Recent Articles:

Morning Headlines 12/13/18

December 12, 2018 Headlines No Comments

Jonathan Bush on Sale of Athenahealth

Jonathan Bush appears on CNBC six months after stepping down as CEO of Athenahealth to share his side of its sale story.

Nashville Hospital Company Pledges To Share Its Sepsis-Fighting Technology With Others

HCA develops and promises to share its Sepsis Prediction and Optimization of Therapy (SPOT) software, which is capable of diagnosing a patient 20 hours before a physician.

Apple now has dozens of doctors on staff, showing it’s serious about health tech

Sources tell CNBC that Apple has at least 50 physicians on staff, many of whom are keeping their roles across the organization quiet.

Pomerantz Law Firm Announces the Filing of a Class Action against Teladoc Health, Inc. and Certain Officers

A scathing article from the Southern Investigative Research Foundation prompts the filing of a class-action lawsuit on behalf of Teladoc shareholders, citing false statements made by the company and a failure to disclose inappropriate employee relationships and insider trading.

Readers Write: Once Retro, Now Current Again: Why Print is Essential to Your Health Education Program

December 12, 2018 Readers Write No Comments

Once Retro, Now Current Again: Why Print is Essential to Your Health Education Program
By P.J. Bell


P.J. Bell is co-chief content officer at StayWell of Yardley, PA.

Smartphones, tablets, computers. You might think the best way to educate patients about their healthcare is electronically. That’s a fair assumption, considering the average American spends nearly 24 hours a week online. With more than 84 percent of folks accessing the Internet from mobile devices (that’s according to a report from the USC Annenberg Center for the Digital Future), you have good reason to think digital is the way to go. But if your practice is considering going all digital, pause for a moment to rethink that strategy.

While digital experiences are critical for engaging today’s savvy healthcare consumer, printed educational materials remain an important part of the care continuum. People still like to receive printed information at an office visit or during discharge from a hospital stay. Both patients and healthcare providers report that printed brochures or handouts are the most effective means of communication. In fact, more than half of patients say that printed educational material about their diseases or the drugs they’ve been prescribed are more useful than other tools available to them. A similar number of providers also indicate that they rely primarily on print collateral when talking to their patients.

Surprisingly, this majority isn’t just older patients. Even Millennials prefer printed materials. Various surveys have shown that more than three-quarters – and in some cases, 90 percent or more – of Millennials said they preferred reading print materials. In fact, if cost was the same for a print or digital book, they’d pick the paper version.

Paper Rules

Healthcare practices and hospitals are faced with the great challenge of ensuring that their patients fully understand their medical conditions and treatment plans. And just as vital, a patient needs to know what to do after leaving the doctor’s office or being discharged from the hospital. More than $73 billion is spent each year on unnecessary healthcare expenses because patients don’t fully comprehend what their medical providers say to them, according to the Institute for Healthcare Advancement.

The fact is, patients face a number of barriers when trying to follow medical guidance. They can be confused by medical jargon or simply overwhelmed by the amount of information—good or bad—that they can access. Some may be embarrassed to seek additional information or hesitant to ask questions. This alone results in about half of all patients leaving their doctor’s office without a solid understanding of what they were told or what they should do going forward.

However, when printed materials are used as a resource, medical staff can go over the information with patients in the office or at the hospital and get a sense of their understanding. This allows staff to determine any additional resources that could benefit the patient. Just as important, when patients are faced with an unexpected diagnosis or new medicine to take, having something in-hand to take home and re-read later enables them to think more clearly about next steps. That way, they can develop questions they may want to ask of their provider via email, phone, or at a follow-up appointment.

In many cases, given the vast quantity of medical information on the Internet—potentially from questionable sources—patients often believe print materials are more official and trustworthy than electronic documents.

The Prescription for a Well-Rounded Patient Education Plan

If your practice is planning to go all digital, let the evidence showing the importance of printed materials give you pause. Going digital can help you deliver advanced offerings. But ultimately, you need to know your audience and communicate in ways that resonate best with them. This may require a multimedia approach.

To develop a patient education program that will deliver greater value to patients and improve outcomes, consider these tips:

  • Don’t re-create the wheel. Your organization probably has a lot of educational content in its archives, whether it is pre-printed brochures or an electronic library from which you can print on demand. Leverage these existing resources to provide customized education that meets individual needs. Give your patients a takeaway that they, as well as family members and caregivers, can refer to later at home.
  • Use technology where appropriate. Every patient has a different learning style, so offer educational content in a variety of formats to help enable comprehension. Also, keep in mind that a large part of health literacy is ensuring your patients have repeated access to information. For some, printed material that they can read and keep as a reference is ideal. Others may respond better to watching a video made available online or as a DVD they can borrow from the office. Some patients may be tech savvy and prefer to access their information from a patient portal, while a few others may lack internet access or be uncomfortable using a computer. Also, consider the primary language of your patient base. Do you need to provide educational content in languages other than English?
  • Review materials with patient and family members. Sometimes just a few extra minutes can make all the difference. When possible, carve out time to talk through the educational material you’re providing, and use common language that most people will understand. Take cues from your patient. If he or she is impacted by fatigue or the shock of a diagnosis, it can be harder to absorb what you’re saying. It’s also important to consider whether patients have physical, mental, or emotional impairments that may affect their ability to learn. Some may need specialized resources if they are vision- or hearing-impaired. Whenever possible, include family members in the education process, since they often play a critical role in your patient’s healthcare management.

Education is key to ensuring that patients understand what they need to do to address chronic conditions, recover from injury or surgery, or improve their overall health. Digital technologies shouldn’t override your practice’s ability to share healthcare information in a way that enhances patient understanding. As you explore new ways of delivering patient education, don’t miss out on the successful communication available when print materials are part of the process. A winning patient education program is flexible enough to deliver content in the format that works best for each patient.

Machine Learning Primer for Clinicians–Part 8

Alexander Scarlat, MD is a physician and data scientist, board-certified in anesthesiology with a degree in computer sciences. He has a keen interest in machine learning applications in healthcare. He welcomes feedback on this series at


Previous articles:

Predict Hospital Mortality

In this article, we’ll walk through our first end-to-end ML workflow while predicting the hospital mortality of ICU patients.

The Python code I’ve used for this article, is publicly available at Kaggle. The data for this exercise is a subset of MIMIC3 (Multiparameter Intelligent Monitoring in Intensive Care), a de-identified, freely available ICU database. MIMIC3 is a RDBMS (relational database management system) with many tables, relationships, and information on 59,000 admissions.

Since no ML model can digest data in a relational format, the first step is to decide how to flatten the relational structure so that each instance will represent one admission.

Hypothesis on the Number of Patient-Hospital Interactions

One can learn a lot about the patient’s outcome from the number of interactions a patient has with the hospital: number of labs, orders, meds, imaging reports, etc. Not necessarily the quality of these interactions (abnormal or normal, degree of abnormality, etc.), just their numbers. It is of course, an oversimplification of the patient outcome prediction problem, but it is an easier toy-model to explain compared to a full-blown, production-ready model.  

We’ll present the ML model with the daily average of interactions, such as the daily average number of labs, meds, consultations, etc. In real life with this arrangement, the ML model can predict mortality each and every day as the average number of interactions between patient and hospital changes daily. 

Flatten MIMIC Into a Table

Using a method detailed in my second book “Medical Information Extraction & Analysis: From Zero to Hero with a Bit of SQL and a Real-life Database,” I’ve summarized MIMIC3 into one table with 58,976 instances. Each row represents one admission. This table has the following columns:

  • Age, gender, admission type, admission source
  • Daily average number of diagnoses
  • Daily average number of procedures
  • Daily average number of labs
  • Daily average number of microbiology labs
  • Daily average number of input and output events (any modification to an IV drip)
  • Daily average number of prescriptions and orders
  • Daily average number of  chart events
  • Daily average number of procedural events (insertion of an arterial line)
  • Daily average number of callouts for consultation
  • Daily average number of notes (including nursing, MD notes, radiology reports)
  • Daily average number of transfers between care units
  • Total number of daily interactions between the patient and the hospital (a summary of all the above)

The hospital mortality is the label of the set, the outcome we’d like the ML model to predict in this supervised learning, binary classification exercise.

Prepare the Data

As previously detailed, I’ve imputed the missing values with the average, the most frequent value, or just “NA.”


Summary of some basic stats:


Leaking Data from the Future

LOS was eliminated from the dataset as it is never a good idea to provide the model information from the future. When asked to predict mortality, the LOS is not yet known, so it should not be given to the model during training. Leaking data to the ML model is equivalent to cheating yourself — the model will have a stellar performance in the lab and a terrible one in real life.

Skewed Data and Normalization

Histogram of age and the number of patients in each category:


The raw data in the above chart is skewed. There are newborns but no other pediatric patients in MIMIC3. There’s also a sharp cut-off at the age of 90. This is partially corrected during the normalization process. The same parameter of age  after normalization:


The process of normalization was applied to all the features.

Mortality and Imbalanced Datasets

The in-hospital mortality of the patients in MIMIC3 dataset is 5,855 / 58,967 = 9.93 percent  of admissions.


This is considered an imbalanced dataset as the classes of lived vs. expired are imbalanced 9:1. Consider a simple, dumb model, one that always predicts that the patient lives and it assumes 0 percent mortality.

Accuracy is defined as (true positives + true negatives) / all samples, so this model has a fantastic accuracy of 90 percent on MIMIC3 dataset. With 100 patients and a model predicting that all patients lived, TN=90 and FN=10, the accuracy is (0+90)/100=90 percent.

If you need more confusion about TP, FP, TN, FN, and their derivatives, please explore the fascinating confusion matrix.

This exercise with a dumb model provides the necessary perspective on the problem, as it gives us a certain baseline to compare against the machine. We know by now that a 90 percent accuracy is not such a high goal to achieve. It’s considered common sense in ML to try come up with a sanity check, a baseline against which to compare a metric, before we measure the machine performance on a task. 

When a problem involves a moderate to highly imbalanced classes situation == such as mortality in our dataset being 10 percent — accuracy is not the only metric to monitor, as it may be quite misleading. The relevance of the predictions is an important parameter as well:

Precision = TP/(TP+FN)

How many selected items were relevant? Our dumb model’s precision is 0, as no true positives (TP) have been selected (0/0+10).

Recall = TP/TP+FP

How many relevant items are selected?  Recall for this model is also 0, as nothing predicted was relevant (actually recall is indeterminate 0/0).


From Wikipedia.

One Metric Only

A ML model needs one, and only one, metric to use for calculations of the loss function and optimizer. The model will refuse to work if presented with two metrics, such as precision and recall. In the case of imbalanced classes, precision and recall have one prodigy, named F1 score — the harmonic mean of precision and recall. A higher F1 score is better.

In the following examples, I’ve used accuracy, precision, recall and F1 score as the models’ metrics for optimization, but only one metric at a time. In addition, I’ve also optimized the models on the area under the vurve (AUC) of the receiver operating characteristic (ROC), even though ROC AUC is best used with balanced classes. A higher AUC is better.

Task: predict mortality as a supervised learning classification based on a binary decision (yes / no)
Experience: MIMIC3 subset detailed above
Performance: accuracy, F1 score, and ROC AUC 

The original dataset was split into two subsets as previously explained:

Training: 47,180 instances or samples (admissions)
Testing: 11,796 instances


Initially I’ve trained and cross-evaluated seven classifier models: logistic regression, random forest, stochastic gradient descent, K-nearest neighbors, decision tree, Gaussian naive Bayes, support vector machine. These models have used the various metrics detailed above for their optimization algorithms, one metric at a time.

The best model came as random forest classifier with the following learning curves while optimizing on ROC AUC:


The confusion matrix for the random forest classifier (RF) model on the test subset of 11,796 samples, never seen before by the RF model:


  • TN: 10528
  • FP: 101
  • FN: 646
  • TP: 521

The above confusion matrix translates into the following performance metrics for the RF model:

  • Accuracy 93.7 percent
  • Precision 83.9 percent
  • Recall 44.4 percent
  • F1 score 0.581
  • AUC 0.717

Feature Importance

We can ask the RF model to display the most important features in the data that helped the algorithm in the decision making process about hospital mortality:


The daily average number of labs seems to be the most important feature for this ML model, almost twice more important than the age parameter.

Note that six features are more important in predicting the admission mortality than the patient’s age:


I’ve tested several neural network architectures and the best results came from a NN with three layers fully interconnected (dense) of 2,048 units each. 

The output sigmoid unit produces a probability between 0 and 1. If this last unit output is above 0.5, the ML model will predict that the patient died. If the output is below 0.5, the model will predict that the patient lived.

Overall the NN had 8.4 million trainable parameters. In order to prevent overfitting, I’ve employed dropout and regularizers with a relatively slow learning rate, as explained in article #7 in this series. The NN training and validation learning curves showing model overfitting after approximately 55 epochs:


The confusion matrix for the above NN model on the test subset of 11,796 samples, never seen before by the model:


  • TN: 10524
  • FP: 105
  • FN: 627
  • TP: 540

The above confusion matrix translates into the following performance metrics for the RF model:

  • Accuracy 93.8 percent
  • Precision 83.7 percent
  • Recall 46.3 percent
  • F1 score 0.596
  • AUC 0.726

A comparison between RF and NN performances on the prediction of hospital mortality:


Next time you are impressed by a high accuracy of a prediction made by a ML model, remember that high accuracy may be accompanied by very low precision and recall, especially with problems where the data classes are imbalanced. In such cases, politely ask for the additional metrics: confusion matrix, Precision, Recall, F1 score, and AUC. 

As a sanity check, always try to estimate what a No-ML-No-AI kind of model would have predicted in the same situation. Use this estimate as the first baseline to test your ML model against.

Next Article

Predict Hospital Length of Stay

Morning Headlines 12/12/18

December 11, 2018 Headlines No Comments

Dr. Denise W. Hines to Join HIMSS as Chief Americas Officer

Former GaHIN Executive Director and EHealth Services Group CEO Denise Hines, DHA, MS joins HIMSS as Chief Americas Officer.

Health Systems Advise Financially Struggling Peers to Measure EHR Vendor Value Propositions Against Heavy Brand Influences

Black Book finds that 88 percent of mid-sized regional systems regret their EHR implementation due to hidden costs, unexpected consulting fees, lost revenue, patient frustration, and clinician burnout.

‘Hey, Alexa! How’s my blood pressure?’

Amazon works with Omron Healthcare to develop a skill for Alexa that connects the virtual assistant to the vendor’s blood pressure monitor.

Colorado hospital failed to terminate former employee’s access to electronic protected health information

OCR fines Pagosa Springs Medical Center (CO) for failing to cut off a former staffer’s access to a Web-based scheduling calendar that included PHI.

News 12/12/18

December 11, 2018 News 1 Comment

Top News


Allscripts will sell its majority share of Netsmart for $525 million, earmarking the proceeds for paying down debt, investing in specific growth areas, and repurchasing shares.

Allscripts acquired its 51 percent stake in Netsmart in March 2016 by contributing $70 million in cash and its Homecare business, partnering with private equity firm GI Partners to invest a total of $950 million in Netsmart.

The Allscripts ownership share will be purchased by its former co-investor GI Partners and private equity firm TA Associates, with the deal expected to close by the end of the year.

MDRX shares rose 4 percent Monday on the news, tempered by a Leerink analyst’s question of why Allscripts would sell out after touting Netsmart’s growth as its original reason for investing in it. The analyst also noted that Allscripts recently blamed its weak bookings on management’s distraction with the Netsmart business.

Netsmart will operate as an investor-backed independent entity. Netsmart CEO Mike Valentine says the company’s growth will accelerate as an independent company, hinting that Netsmart may pursue acquisitions as  it focuses on homecare growth.

Reader Comments

From Tennessee Tuxedo: “Re: HIStalk. A recent Washington, DC meeting included a technical expert panel discussing whether market forces could influence EHR vendors to change their pricing structure for access to their APIs. One vendor rep said, ‘It does when it shows up on HIStalk.’ I call it being HIStalked. Keep up the great work.” Thanks. I’m happy to be turned into a verb that suggests shining a light on arguably questionable practices. 


From Amatriciana: “Re: DonorsChoose. I’d like to make a donation again this year when the matching funs are the highest, I really appreciate the extra matching you got last year.” I’ve exhausted the matching money from my anonymous vendor executive, but other DonorsChoose matching opportunities abound – the last round of donations were matched 5x and even 10x by foundations. Instructions:

  1. Purchase a gift card in the amount you’d like to donate.
  2. Send the gift card by the email option to (that’s my DonorsChoose account).
  3. I’ll be notified of your donation and you can print your own receipt for tax purposes.
  4. I’ll pool the money, apply the matching funds, and publicly report here (as I always do) which projects I funded.

From Dirk Squarejaw: “Re: Dr. Jayne. She cited a story from CNN. Fake news!” I’m not sure if this is a tongue-in-cheek comment, but I’m not entirely opposed to calling CNN “fake news” (although I prefer “dumbed-down entertainment posing as news that intentionally keeps people agitated and thus coming back.”) CNN and other news outlets have found that factual and nuanced reporting of complex world events doesn’t draw the profitable eyeballs of the intellectually lazy who want sensationalistic stories (including the medical ones), news “celebrities” who blast out opinions that pander to a targeted demographic, and shallow entertainment posing as current events (The Onion satirized it brilliantly back in 2013). However, you can’t blame CNN – they provide the supply of crap that our fellow citizens demand, at least in those rare occasions in which they tire of watching funny YouTube videos, posting nearly indecipherable Facebook rants, and entertaining themselves by using filters on their selfies or selecting lame GIFs as reactions to avoid the intellectual marathon of stringing actual words together. Compare CNN’s choice of top stories to that of the far more responsible American edition of BBC News. Pathetic news reporting reflects rather than causes our increasingly unsustainable culture, which resembles an overweight, angry, and socially outcast teenager who locks themselves in their room surrounded by videogames, drugs, and junk food until something sets them off from self-indulgence to violence.

From Significant Mother: “Re: smartphones. What’s your take on Apple’s high-end models not selling well?” Beyond fanboy status, phones have become a commodity in performing equally well for making calls (a minor use case for most people), texting, running apps, or browsing the web. It’s a mature market in which vendors add questionably useful features and tweak form factors as a differentiator and incumbents are threatened by lower-cost competitors. The only battleground remaining is over the all-important camera, and while Apple has improved in that area at least in terms of pixel wars that matter little for online photo posting, Google Pixel’s Night Sight (AI-powered low-light performance) is the only newsworthy development.

From Fax Me a Simile: “Re: NHS’s fax machine ban. We should do the same here!” You are assuming (incorrectly, I suspect) that providers would be thereby forced to adopt more modern interoperability technologies even though our hospitals aren’t government-run as in the fax-axing England. Most likely they would simply go back the pre-fax standard of mailing photocopies or asking patients to hand-deliver documents. You would also be removing the only form of interoperability that is universal, that costs next to nothing, that never goes down, and that has rarely spilled PHI. Mandate other forms of interoperability (instead of just banning a particular one) if you feel the need to intervene against market forces, but note that providers aren’t paid to share patient data and are rarely punished for refusing to do so, so you’ll just screw patients in trying to force cooler but harder, more expensive technology on providers who aren’t the major beneficiary.

HIStalk Announcements and Requests

Listening: Deadland Ritual, a new bluesy, hard rock band assembled by the underrated Black Sabbath bass player Geezer Butler, also featuring Billy Idol’s highly competent guitarist Steve Stevens, the drummer from Guns N’ Roses, and Scars on Broadway singer Franky Perez. It sounds quite a bit like Black Sabbath, but with a more driving, clean sound and minus Ozzy’s sometimes grating vocal stylings.


None scheduled soon. Previous webinars are on our YouTube channel. Contact Lorre for information.

Here’s the recording of last week’s CitiusTech webinar, “Make the Most of Azure DevOps in Healthcare.”


  • Berkshire Health Systems (MA) will implement Meditech Expanse.
  • Molina Healthcare selects Inovalon for improving member care and documentation.
  • National post-acute care provider Signature HealthCare chooses MatrixCare’s EHR for all of its 115 Signature locations.
  • Normal Regional Health System (OK) will implement Meditech-integrated Access Passport to make electronic forms available on IPads.



Denise Hines, DHA, MS (EHealth Services Group) joins HIMSS as Chief Americas Officer.

Announcements and Implementations

Zen Healthcare IT announces its expanded HIE capability based on its work with Arizona’s Health Current HIE.


Johns Hopkins All Children’s Hospital (FL) parts ways with its president, cardiovascular chief, chief of staff, and surgery department chair following  a newspaper’s report that its Heart Institute mortality tripled between 2015 and 2017 even as employees warned management about the work of specific surgeons. At least 11 children died in the 18 months after the internal warnings. This is yet another reminder that (a) we would be hosed without investigative journalism; and (b) a hospital’s fancy buildings, brand name, and self-stroking advertising aren’t necessarily indicative that they aren’t screwed up internally in a way that may harm patients. US News & World Report must be embarrassed to have named All Children’s to its “Best Children’s Hospitals” for cardiology and heart surgery for 2018-2019 and Hopkins should be equally embarrassed for taking over All Children’s six years ago with a promise to elevate its heart surgery program as one of the country’s best and instead made it the highest-mortality hospital in Florida. 

Black Book finds that non-profit health systems of greater than 1,000 beds are happy with their EHR choice even after suffering through blown budgets and lost revenue, but 88 percent of mid-sized regional systems regret their implementation due to hidden costs, unexpected consulting fees, lost revenue, patient frustration, and clinician burnout. Black Book speculates that those hospitals focused too much on choosing the right functionality and getting the implementation done efficiently while failing to address workflows, usability, and interoperability. Other findings:

  • Three-fourths of C-suite respondents question whether their EHR switch was worth it.
  • Nearly all financially challenged hospitals regret the decision of their executives to replace their EHR.
  • Three-fourths of respondents say interoperability declined after implementing a new system even though the technical capability exists, probably because nobody is paying them to exchange patient information.
  • Hospitals report that their new EHR hasn’t helped them attract doctors.
  • Two-thirds of executive respondents say they worried about their jobs during the replacement.


A small study finds that providing hospital inpatients with tablets that are set up to access a patient portal didn’t improve patient activation, although patients did sometimes use the portal to look up information.

I missed this from a few months back. An expert says that while consumer DNA tests aren’t very useful, any company that can figure out how to make whole-genome sequencing free or cheap can become the Google of that field in providing the “sweet Texas crude” that is needed for clinical treatment and research. He notes that tests such as those offered by and 23andMe lure customers into donating “an intensely personal, incredibly valuable asset” instead of being paid when their data is sold or used to create new drugs or other products. He adds,

The bigger a genomic network becomes, the more likely it is that correlations previously impossible to detect will be uncovered, and the more people and groups will sign on to mine the information for gold … Genomic marketplaces are already attracting partners interested in paying for access to your DNA sequences and related information, with your consent … But the marketplace will really thrive when 2G DNA companies eventually tap into the wellspring of dollars that today supports the Web: advertising. Genomic networks could become the richest source of detailed, opted-in data ever collected for targeted advertising. As more gene-linked products and services appear, these marketplaces should diversify beyond health and medicine, and the revenues flowing through them should explode. And you’ll get your cut.


Brilliant: Texas prisons will begin 3D printing of dentures for inmates, restoring functionality at a cost of just $50 per set and addressing complaints that many US prisons are so financially strapped to provided medical care that dentures are rarely provided and only in cases of medical necessity (the inability to chew doesn’t count). The photo above is of an inmate’s 3D-printed dentures, which are remarkably lifelike. Good job, Texas.


Dr. Gottlieb channels Dr. Suess in a tweet that is as amusing as it is timely.


Shriners Hospitals for Children offers its young patients video visits with Santa Claus this week. Shriners used the Santa visits to test its Dimension Data telemedicine system rollout in 2015.

A woman dies of hypernatremia after attempting a “soy sauce colon cleanse,” an Internet fad that involves drinking a quart of soy sauce over two hours.

Sponsor Updates


  • AssessURHealth raises $6,750 for the American Foundation for Suicide Prevention Tampa Bay Out of the Darkness Tampa Walk.
  • Netsmart profiles Army Reserve / National Guard VP David Aug in its “Meet Our Veterans” series.
  • The Baltimore Sun includes Audacious Inquiry in its list of top workplaces.
  • Atlantic.Net partners with Veeam Software to offer customers data protection and availability solutions.
  • Bluetree launches an Epic-focused service center.
  • Healthcare Growth Partners publishes its November Health IT Monthly Insights report.
  • Datica releases a new book, “Complete Cloud Compliance.”
  • ChiefExecutive profiles Collective Medical CEO Chris Klomp.
  • KLAS rates partial IT outsourcing services from Cumberland Consulting Group with above market average scores in all key performance areas.
  • Gartner includes Dimensional Insight in two hype cycle reports on healthcare.
  • Bernoulli is integrating NIST’s Cybersecurity Framework v. 1.1 into its Bernoulli One medical device integration and continuous surveillance platform.

Blog Posts



Mr. H, Lorre, Jenn, Dr. Jayne.
Get HIStalk updates. Send news or rumors.
Contact us.


Morning Headlines 12/11/18

December 10, 2018 Headlines No Comments

Former Director Of Healthcare Services Company Charged In Alleged $300 Million Investment Fraud Scheme

Federal officials arrest Pavandeep Bakhshi, a former Constellation Healthcare Technologies board member, for his role in a scheme that attempted to trick investors out of $300 million meant to help take the company private.

University of Maryland Medical System investigating malware attack

The University of Maryland Medical System recovers from a ransomware attack that impacted 250 of the system’s desktop computers.

GI Partners and TA Associates to Acquire Allscripts’ stake in Netsmart Technologies

Shares of Allscripts rise on the news that GI Partners and TA Associates will acquire its stake in Netsmart Technologies for $525 million.

U.S. Department of Veterans Affairs Partners with T-Mobile to Help Expand Access to Health Care for Veterans

T-Mobile will provide the VA with 70,000 lines of wireless service to help it expand telehealth services for veterans.

Seattle Children’s spin-out MDmetrix adds operating room capabilities, looks to expand

Health data analytics startup MDmetrix expands into operating room capabilities with OR Advisor, and looks to expand beyond its Seattle Children’s roots.

Curbside Consult with Dr. Jayne 12/10/18

December 10, 2018 Dr. Jayne 5 Comments

The physician lounge was abuzz on Friday due to a piece on CNN claiming that Australian researchers have developed a “10-minute cancer test.” Supposedly it “can detect the presence of cancer cells anywhere in the human body” and stems from research looking at the structure of cancer DNA when placed in water. Physicians were mostly grumbling about having to respond to patient questions about such a sensational announcement when the ink on the publication was barely dry. Patients tend to take hold of these kinds of announcements, especially if they have a particular concern about cancers for which there aren’t good screening tests, such as ovarian cancer.

There’s always more to the story when these announcements are made. Despite author Matt Trau’s statements that the study “led to the creation of inexpensive and portable detection devices that could eventually be used as a diagnostic tool, possibly with a mobile phone,” in this case, the test hasn’t even been used on humans. People tend to hear the part about diagnosing cancer with their phones and miss the part about animal studies. The authors are excited and with good reason, but it’s a long way from where they are with this test to having it available at the primary care office.

The test mentioned in the publication, which was released this week in “Nature Communications,” has only been used to detect lymphoma, along with cancers of the breast, prostate, and bowel. It’s also only been used on around 200 samples, although it did have 90 percent accuracy. Researchers using high-resolution microscopy noted differences between the structure of cancerous DNA fragments and non-cancerous fragments when the DNA was placed in water. The test uses colloidal gold particles to bind to cancerous DNA, creating an electrochemical reaction that can be quantified.

One of the urologists around the table was particularly vocal about suggesting that this test could be used for prostate cancer since there has already been a fair amount of controversy about prostate cancer screening. We’ve seen the Prostate-Specific Antigen (PSA) fall in and out of favor – first approved by the FDA in 1986 to monitor prostate cancer progression, it was approved in 1994 to be used along with a digital rectal exam for screening of asymptomatic patients. Over the next two decades, we saw patients with “abnormal” tests who underwent procedures that may have been overly aggressive given the slow-moving nature of prostate cancer, not to mention the non-cancerous conditions that can cause PSA elevation. Over time, we learned that the test was being relatively overused certain populations without definitive evidence that it drives outcomes in a beneficial way, leading to recommendations that we don’t just order it, but rather have a risk/benefit decision between the patient and the physician before deciding to test.

As we consider new technology and new tests, we need to heed the lessons of the past and proceed with caution, guarding against “shiny object syndrome” and the assumption that just because we can theoretically use a smart phone to do a test that it’s a good idea. CNN ran a similar piece back in January, covering a test developed at Johns Hopkins University that screens blood samples for eight common cancers by detecting cancer proteins and gene mutations. That test, called CancerSEEK, is still being studied to determine its applicability in clinical medicine and whether it can be widely used to screen patients who aren’t experiencing symptoms. CancerSEEK was evaluated in a much larger study that included humans with almost 2,000 patients participating. The test was 70 percent sensitive among the eight cancers, but the range of accuracy for individual cancers ranged from 33 percent in breast cancer to 98 percent in ovarian cancer. The Hopkins team also used an algorithm to evaluate the source of the cancer for positive tests, but the ability to pinpoint a source was only 63 percent.

It will take a tremendous amount of money to bring either of these technologies to the point of care, and unfortunately with medical research, the money doesn’t always follow the hype. Even when tests are promising, they have to be shown to be effective and to be able to make a difference across large patient populations before payers will cover them, which often the main barrier to patients receiving new tests and treatments. EHR and other healthcare vendors follow these discoveries closely since they need to stay ahead of the curve for supplying appropriate clinical decision support information and including new discoveries into order sets and EHR content.

Those changes don’t happen overnight. I work with one EHR vendor that still hasn’t incorporated standard-of-care screenings that were recommended by the United States Preventive Services Task Force (USPSTF) back in 2007. It’s understandable that providers are frustrated when it takes more than a decade to update the EHR.

The conversation about detecting cancer DNA quickly segued into one about the recent “gene-edited baby” announcement coming out of China. A scientist claims to have used the hot new CRISPR gene-editing technology to alter two human embryos to be resistant to HIV. The babies have now been born and the news led to significant outrage from the international scientific community. The processes of announcing the research has broken with the standards of research, with the information being revealed via YouTube rather than through rigorously-reviewed scientific channels. That’s not surprising in the era of social media, but should be viewed with caution. There are many other concerns with the research, including lack of appropriate Institutional Review Board protection for the participants, lack of documentation of the work actually done, and the lead researcher owning patents around the techniques used in the process. It wouldn’t fly in the US or in many other nations.

The conversation came full circle when one of family medicine docs at the table spoke up. She said she felt sad that everyone was excited about these media sound bites around research whose practical use was years away, but she has difficulty getting medical professionals engaged around her work with school-based clinics and mobile outreach to our city’s homeless population. I mentioned working with providers who struggle with EHR adoption and the challenges of trying to get them to use the guideline prompts and alerts that are already in the system for tests that are proven to be clinically effective as well as cost effective. It’s certainly something to think about in this world where we’re used to getting our information 200 characters at a time and the deeper discussions sometimes elude us. Physicians don’t have the time to pull the original articles and read the primary source data, so it’s unlikely that patients asking about these new advances are going to have done so either.

Given our work in healthcare information technology and the seemingly relentless push for innovation, we often become skeptical (if not cynical) about developments. We’ve seen plenty of creative new technologies fizzle and watch the industry continue to search for the next big thing. And we understand how hard it is to take technology from the idea stage to practical use at the patient bedside whether physical or virtual. It will be interesting to look back on these developments in a year, or five or 10, and see where we have landed.


Email Dr. Jayne.

Morning Headlines 12/10/18

December 9, 2018 Headlines No Comments

NHS banned from buying any more fax machines

In England, Health Secretary Matt Hancock bans NHS from buying new fax machines and insists that they be phased out by March 31, 2020.

Hospital Beds Get Digital Upgrade

Hill-Rom’s newest hospital bed will include FDA-approved sensors for monitoring heart and respiratory rates, checking vital signs 100 times per minute, and alerting nurses of abnormalities.

Apple scoops up CEO of Mango Health, a start-up that helps people keep track of their medications

Apple hires former Mango Health CEO and co-founder Jason Oberfest, spurring speculation that the company is looking to tackle medication adherence.

Top cancer center’s business deals created a web of conflicts, say ethics experts

STAT calls into question Memorial Sloan Kettering Cancer Center’s deal with data analytics vendor Cota, whose founder helped to broker a partnership between MSK and his employer, Hackensack Meridian Health.

Monday Morning Update 12/10/18

December 9, 2018 News 3 Comments

Top News


In England, Health Secretary Matt Hancock bans NHS from buying new fax machines and insists that they be phased out by March 31, 2020.

The Royal College of Surgeons agrees, estimating that NHS still has 8,000 fax machines in service.

Here we hospital people thought we were being cutting edge by moving to multifunction devices that at least bundled faxing with printing and scanning. On the other hand, if a business case exists for using something other than fax, they would already be gone.

Reader Comments

From Digital Debonair: “Re: paging systems. A Texas hospital found that Epic-issued consult pages were not being delivered if the message size exceeded character limits – 280 characters for pagers, 160 for mobile phones. The hospital limited Epic’s ‘reason for consult’ field to 100 characters and added an alert to the intended recipient’s mobile device when the limit is exceeded. Once again, technology’s unintended consequences bring us to the least common denominator instead of fixing the problem by breaking the message into segments or getting the communications vendors to increase their character limits. It’s fascinating that each hospital has to discover and solve this problem on their own. Sigh … we have so many miles to go.” Unverified, but the hospital’s email warning to the medical staff was attached. I verified that Sprint and Verizon have 160-character limits, while ATT breaks messages into multiple 160-character segments automatically. SMS stands for “short message service,” so perhaps the real problem is that hospitals try to use that service for something for which it was not intended (not short, in other words) regardless of the convenience of doing so. There’s also the question of whether PHI should be sent over SMS instead of via an encrypted messaging app that could also provide a larger character limit.

From Wan Complexion: “Re: Most Wired. You didn’t list the winners.” I don’t see the point, even as someone who has run IT in organizations that won. We should judge health systems on outcomes, cost, and consumer focus, not on using tools that should drive those results (but usually don’t). I ate at a McDonald’s and it was still awful despite (or perhaps because of) an enviable arsenal of enterprise-wide technology. By “Most Wired” standards, I should have loved it.

HIStalk Announcements and Requests


Poll respondents fear that Amazon will use the medical data they can get to influence their buying habits, although to be honest I’d trust Amazon a ton more than Google or Facebook since Amazon’s business model involves moving merchandise, not serving up ads that clearly were chosen using information those companies really shouldn’t have.

New poll to your right or here: should hospitals be prohibited from using fax machines? Vote and then click the poll’s “comments” link to explain.

I’m questioning those frantically gesticulating TV weather people who this weekend are milking camera time with what they call a “winter storm,” “winter weather,” and of course the inevitable “wintry mix.” It’s not winter until December 21, although I recognize that the less-hysterical “fall storm” won’t keep hunkered-down eyeballs glued to the TV commercials and the result isn’t any different regardless of what the calendar says.

Thanks to the following companies that recently supported HIStalk. Click a logo for more information.




None scheduled soon. Previous webinars are on our YouTube channel. Contact Lorre for information.

Acquisitions, Funding, Business, and Stock

Allscripts shares hit a 52-week low last week, having shed 34 percent in the past three months. Anonymous posters on claim that around 80 percent of the 1,700 McKesson EIS people who joined Allscripts with the acquisition 14 months ago are no longer there.

IBM sells off several software lines to an India-based company, among them Lotus Notes/Domino, which should elicit hope from IBM’ers who have been stuck on that unpopular platform while the rest of the world moved on. Maybe they’ll replace it with GroupWise.


Medication reminder technology vendor MyMeds issues a press release whose headline appears to be intentionally misleading, dutifully picked up by some crappy health IT sites as a “partnership” between the company and Mayo Clinic. Plowing through the fluff reveals the actual development – the app will offer users Mayo Clinic’s drug information (for which I assume the company is paying). Any resemblance to “teaming up” appears to be coincidental.

InterSystems releases a cloud-hosted version of its TrackCare EHR for hospitals in the UAE and Middle East, licensed in a pay-per-usage model.

Hill-Rom’s newest hospital bed will include FDA-approved sensors for monitoring heart and respiratory rates, checking vital signs 100 times per minute and alerting nurses of abnormalities. The price was not announced, but the company’s traditional bed is among the most expensive with a list price of $20,000.


  • Northside Hospital System (GA) replaced Allscripts with Cerner in October 2018.
  • Gifford Medical Center (VT) went live on EClinicalWorks in April 2018, replacing Evident.

These provider-reported updates are supplied by Definitive Healthcare, which offers a free trial of its powerful intelligence on hospitals, physicians, and healthcare providers. )

Announcements and Implementations

Citizens Memorial Hospital (MO) upgrades to Meditech Expanse.

Hospital Sisters Health System integrates Epic with SeamlessMD’s patient engagement solution using SMART on FHIR. 

Government and Politics

Six pain management doctors in Michigan are charged with insurance fraud and unjustified opiate prescribing in submitting $464 million in phony insurance claims.



Here’s an interesting tweet from Apple CEO Tim Cook. I’m not sure the silver bullet for people managing their health lives inside of an IPhone, but I’m sure a citation-desperate academic will compare life expectancy of IOS and Android users vs. a control group of non-cell users.

An article by Penn’s Wharton School weighs in on Amazon’s announcement that it will mine unstructured patient data using AI and machine learning in its Comprehend Medical program, saying the service could:

  • Empower consumers.
  • Deliver new insights, particularly with regard to radiology, and connect people with clinical trials.
  • Allow insurers to deny enrollment of patients with potentially expensive conditions.
  • Lighten the workload of doctors.
  • Erode physician loyalty as patients could manage their own medical information or choose to share information with competitors such as retail clinics.
  • Replace consultants who perform custom predictive analytics for individual clinical conditions.
  • Raise questions about data accuracy, especially if consumers are allowed to add or change their information.
  • Cause major problems if Amazon were to be breached.
  • Raise questions of who’s paying the bill for the Amazon service.
  • Lure clinicians into becoming overly reliant on technologies instead of learning, improving, and questioning how the models work.

A ProPublica report finds that journal articles written by physician researchers often don’t disclose the money they’re paid by drug and medical device companies as required, with the medical journals doing little checking of their own. Among them is the dean of Yale’s medical school, the president-elect of the American Society of Clinical Oncology, and the president of clinical operations at Sarah Cannon Research Institute. The reports didn’t have to dig all that deeply – they simply looked up compensation as reported to CMS’s Open Payments Database and compared that to the disclosures section of published articles.

Weird News Andy says this patient hacked up a lung, kinda. A patient coughs up what looks like a bright red, leafless tree, which turned out to be a six-inch-wide blood clot formed in his right bronchial tree (and now you can see how apt that name is). I’ll spare you the photo just in case you’re eating  breakfast since it’s both fascinating and disturbing.

Sponsor Updates

  • Liaison Technologies awards its Data-Inspired Future Scholarship to BYU dual-major student Andrew Pulsipher.
  • Loyale Healthcare introduces the Patient Financial Bill of Rights.
  • Mobile Heartbeat will exhibit at the ONL Winter Meeting December 14 in Burlington, MA.
  • National Decision Support Co. and Redox will exhibit at the IHI National Forum December 9-12 in Orlando.
  • NextGate launches a fundraising campaign to help customer HealtheConnect Alaska recover from the earthquake.
  • Netsmart will exhibit at the TAMHO Annual Conference December 11 in Franklin, TN.
  • The Business Gist features Sansoro Health CEO Jeremy Pierotti in a new video, “The challenge of sharing medical records.”
  • New data from Surescripts shows that its benefit optimization tools have saved patients as much as $8,032 in out-of-pocket costs on a single prescription.
  • Vocera launches three leadership councils to accelerate healthcare transformation.
  • ZappRx will exhibit at Advances in IBD December 13-15 in Orlando.
  • Healthwise discusses why its partnership with ZeOmega benefits clients.

Blog Posts



Mr. H, Lorre, Jenn, Dr. Jayne.
Get HIStalk updates. Send news or rumors.
Contact us.


Weekender 12/7/18

December 7, 2018 Weekender No Comments


Weekly News Recap

  • Apple’s Watch 4 OS update includes the ECG app and arrhythmia notification capability.
  • Meditech acquires its London-based partner Centennial Computer Corporation as part of its creation of Meditech UK.
  • A KLAS report finds that most EHR vendors are progressing well toward supporting a national patient record network now that CommonWell is connected to Carequality.
  • In Australia, Queensland Health’s hospital EHR project will run $188 million over budget if implemented in the remaining hospitals.
  • A ProPublica report concludes that three supporters of President Trump had influence over the VA’s $10 billion Cerner contract and got former VA Secretary David Shulkin fired.
  • Allscripts confirms an unstated number of employee layoffs.
  • Athenahealth files shareholder notice of a vote on its proposed acquisition by Veritas Capital and Elliott Management.
  • Connected health technology vendor ResMed will acquire Madison, WI-based Propeller Health for $225 million.
  • Leading UK EHR vendor Emis Group will shift 40 million patient records from its servers onto AWS as part of a continued  push in the UK for more flexible health data exchange.

Best Reader Comments

Interoperability will never be fully solved by creating more regulations and layering on all sorts of requirements on data then making portions of it voluntary. It’s truly a confusing system mired in all sorts of administrative burden and muck with too much conflicting self-interest. There are many models from other countries that work more effectively, have lower mortality rates, less physician burnout. Perhaps instead of spending billions on more regs and administrative burden, maybe step back spend some of that on evaluating effective healthcare delivery models and select one that works. (Renee Broadbent)

Cerner is THE founder of CommonWell and they make it hardest for their customers to implement. Further mucks up DoD and VA plans for interoperability, though they seem to be all talk little action on interoperability anyway. Thank you Athena, EClinical, and Epic for leading the way! (Charlie Harris)

Is the above for real? Who dreams this stuff up? Mixing two disparate protocols for a transaction activity? Lets make this a complex as possible! It is as if they really don’t want organizations implement this functionality so they make the cost of entry as high as possible. (David Coffey)

Dentists are taught in dental school that they are going to be small business owners, and taught how to run a profitable business. Medical schools seem to focus on a world where all doctors stay in academia, instead of the reality that millions of doctors are small business owners. The expectations that dentists have for the successful operations of their dental healthcare businesses drives the advances in their industry. (Julie McGovern)

I am sure the bigwigs and muckity-mucks that come into consulting after losing their comfy jobs make the rest of us look pretty bad and desperate to outsiders, but from my experience (seven years of consulting, running my own little shop, loving it each and every day) there are plenty of opportunities to work, great clients to help, unbelievable experiences to have, and we have a bit more freedom to live a life that supports having a family, raising children, and balancing a life that isn’t just an identity of “I work for [blank company name].” (Consulting Union Needed?)

An ONC Safety Center (which Congress didn’t fund) with peer review and anti-trust protection for IT vendors is the right answer here. Maybe ONC could focus on that instead of dithering around with tefca and “information blocking.” (Charlie Harris)

Watercooler Talk Tidbits

image image

Readers funded the DonorsChoose teacher grant request of Ms. G in Utah, who asked for an an Osmo Wonder Kit for her third grade class. She reports, “We have been using the kit during our small group time. The games that came with the kit help the students practice phonics, number sense, math facts, logic, and other important skills. The students beg to get it out and use it, and even want to stay in during recess to play! I love watching them manipulate the tools to get the right answers. The looks on their faces when they get the answers right are priceless! My absolute favorite part, however, is watching them work together as a team to find the answers. They help and encourage one another, and even when someone gets an answer wrong they encourage their classmates with phrases like, ‘Everyone makes mistakes! Let’s try again!’ I never expected the Osmo Genius Kit to have that sort of impact in my classroom.”

Ben and Michelle of ST Advisors always include my DonorsChoose project in their annual charity support. Their generous donation, matched with funds from my anonymous vendor executives and other sources (some with 10-times matching!), fully funded these teacher projects:

  • Robotics tools for Mr. D’s junior high class in Cedar Creek, TX (classroom was affected by Hurricane Harvey)
  • Math and reading centers for Ms. T’s kindergarten class in Oroville, CA (classroom was affected by the Camp Fire)
  • Programmable robots for Mr. A’s grade school class in Bronx, NY
  • 30 sets of headphones for Ms. B’s sixth grade class in Spring, TX (classroom was affected by Hurricane Harvey)
  • Four Chromebooks for Mr. V’s high school class in Bridgeport, CT
  • Math manipulatives for Ms. L’s first grade class in Washington, DC
  • 14 sets of headphones for Ms. H’s high school class in Mesa, AZ
  • 25 sets of headphones and solar system learning tools for Mr. F’s elementary school class in Porter, TX (classroom was affected by Hurricane Harvey)
  • Diversity and multicultural learning activities for Ms. H’s elementary school class in Wellington, KS

I heard back quickly from several of these teachers, including Ms. T, who said, “I was so surprised when I peeked at my email at lunch and read the great news. I wish I had recorded the squeals of joy from my students when I shared the fun that is to come in the mail for them. Your generosity is appreciated. Merry Christmas!”

This research might have been more appropriately released on April 1. A study finds that a parasite found in cat poop is associated with a higher likelihood of entrepreneurial behavior (I would have expected bull manure given the success of some executives). Actually, my theory is this – Toxoplasma gondii is more commonly acquired by consuming contaminated food or water, which would be far more commonly found in countries such as India whose society values entrepreneurial behavior, hard work, and academic achievement more than ours. I love that many US business are created and run by hardworking, well-educated, family-focused people from other countries who in many ways exemplify the American dream better than many native-born citizens whose goals seem to be consuming mindless entertainment, taking advantage of entitlement programs, and ridiculing those who work harder and smarter and are rewarded accordingly.


I wanted to replace my old, cheap wireless router to make sure I’m using the most current protocols and ran across this fantastic $75 mesh router. I plugged it into the modem, connected to it via its app, entered my desired network name and password, and it was running flawlessly literally within two minutes of opening the box. Setting up a guest network took another 30 seconds (again, just entering a network name and password). The range is excellent, but I had ordered a second one just in case and the only setup required was to plug in the power cord – it instantly connected to the first router and started beaming the signal even further away.

Walgreens partners with FedEx to offer next-day prescription delivery, with same-day service in some cities. 


Hurricane-damaged Bay Medical Sacred Heart (FL) will lay off 800 employees – half its workforce – when it reopens in January at one-fourth its original size.


Former MD Anderson CIO Lynn Vogel, PhD publishes “Who Knew? Inside the Complexity of American Health Care.”

In Case You Missed It

Get Involved



Morning Headlines 12/7/18

December 6, 2018 Headlines No Comments

Partners goes down (and then back up)

Politico reports that Partners HealthCare (MA) briefly took its Epic EHR offline Wednesday to handle unspecified technical issues.

U.S. Department of Veteran Affairs and Walmart Announce Telehealth Collaboration to Reach Underserved Veterans

The VA announces at its telehealth event in Washington, DC that it will offer telemedicine services to vets at select Walmart stores.

Apple now says its smartwatch tech to detect atrial fibrillation is not for those with atrial fibrillation

Apple emphasizes that the ECG app and irregular heart rhythm feature launched today on the Apple Watch 4 are not intended for people with atrial fibrillation, but should rather serve as conversation starters with physicians.

Telehealth Virtual Care Platforms 2018

Epic and InTouch lead in telemedicine value and impact, according to a new KLAS report, while only Epic, American Well, and MDLive have more than half their customers moving along an EHR integration path.

News 12/7/18

December 6, 2018 News No Comments

Top News


It’s been a busy week for Apple when it comes to healthcare:

  • The FCC clears an Apple-branded sleep monitor built using technology the company gained from its Beddit acquisition last year. 
  • Apple Watch 4 users who update to watchOS 5.1.2 can now use the ECG app and notification feature for irregular heart rhythm.
  • The US PTO awards the company a patent for interchangeable AirPod earbuds that can incorporate biometric sensors for heart rate and temperature monitoring.

Reader Comments

From Bjorn Again: “Re: out-of-work executives temporarily consulting. Many just need a title while playing out their parachute and await their next position. I’m a career consultant and these folks distract our prospects from the skills and work we propose, sometimes even making us look bad as we don’t expect to be paid $300/hr. Sometimes they bid or leverage their previous relationships to win over a better, but slightly lesser known option. The big one for me is the old-time vendor execs who have been culled out and are now consulting, suddenly claiming to understand BI, blockchain, machine learning, cloud, etc. after working 27 years for a mainframe-based company, passing off a hobby or reading LinkedIn articles as a professional skill.”

From Former Startup CEO: “Re: startups. Graduating from an incubator or developing a minimally viable product is just the beginning. Companies don’t know how to grow to profitability and the time and expensive of onboarding one new client doesn’t match growth expectations of 10 per week for several months. They don’t know how to gain business or traction. Investor portfolios are filled with dogs (bad investments) and puppies (soon to be dogs) because it’s too hard to deploy their solution.”


None scheduled soon. Previous webinars are on our YouTube channel. Contact Lorre for information.



Pam Matthews, RN, MBA (Collie Group Consulting) joins Georgia Health Information Exchange Network as executive operations officer.


  • Nicklaus Children’s Health System (FL) selects Health Catalyst’s Data Operating System to optimize its RCM.
  • CaroMont Health (NC) will deploy physician time-tracking and payment software from Ludi.

Announcements and Implementations


A new KLAS report on telehealth platforms finds that few vendors have customers using their product for all three forms of telehealth (on-demand care, virtual visits, and specialty consultations). Epic — whose product works only within its own system — and InTouch lead in value and impact, while only Epic, American Well, and MD-Live have more than half their customers moving along an EHR integration path.

Privacy and Security


Politico reports that Partners HealthCare (MA) briefly took its Epic EHR offline Wednesday to handle unspecified technical issues. A hospital spokesperson was quick to rule out the possibility of a data breach. This Twitter thread, prompted by the Partners event, provides some amusing insight into provider attitudes towards downtimes.

Government and Politics


The VA announces at its telehealth event in Washington, DC that it will offer telemedicine services to vets at select Walmart stores.



In Canada, physicians argue for more input into the already-contentious bidding process for Nova Scotia’s One Person One Record System. Cerner and Allscripts are vying for the contract. The Doctors Nova Scotia association says the process needs more providers involved to avoid the EHR problems faced by Cerner customer Island Health in Vancouver. According to DNS President Tim Holland, MD, “If you look at how the electronic health record was set up on Vancouver Island, it crippled their healthcare system, it completely paralyzed their ability to deliver care in hospital, and it had a huge negative impact on patient health and patient safety … if done poorly, this could cripple our healthcare system. It’s very important that frontline healthcare workers — doctors, nurses, and the organizations that represent them — are involved in the development and implementation of this system.”


I missed this in Health Affairs last month: Pascal Metrics develops software that uses machine learning and EHR data to detect and alert providers to medical errors in real time. Developers found that the program could detect errors as they happened at higher rates than current methods, but experts have pointed out that the false positives triggered by the software would a pain for hospitals to deal with.

Medical City Dallas mistakenly bills a patient for $13,000 after a “patient portal mix-up,” according to MCD. The situation was remedied only after the patient took her predicament to the local news. Coincidentally, University of Michigan researchers find that out of 2,300 patients, only one-third used a patient portal in 2017. Respondents cited lack of need, a desire to speak with their provider face to face, and not knowing about portal availability as top reasons for their lack of use.


Teladoc is quick to refute claims of inappropriate employee relations and insider trading that were made in an article from the Southern Investigative Reporting Foundation. The report says the CFO was having an affair with the lower-level employee and shared company stock advice with her. The employee bragged to co-workers who complained to their boss, who pushed through an investigation. The CFO got off with a warning and a one-year loss of share vesting, his girlfriend was not disciplined and later left the company with an unstated severance, but the boss who pushed the investigation was fired. Nobody was investigated by the SEC for insider trading. The company said it acted swiftly and fairly in taking appropriate disciplinary action.


I find it ironic that Googlers argue for fairness in machine learning when their co-workers are preparing to strike over the company’s plan to launch a censored search engine in China.


A Weird News Andy wannabe reader is happy he beat WNA to the punch with this story. In England, a pharmacist faces life in prison for strangling his wife in a staged burglary that he hoped would allow him to collect $2.6 million in life insurance. He planned to use the money to join his same-sex lover in Australia, where they would use the wife’s frozen embryos to start a family. Police examined the IPhones of the man and his wife, discovering that Apple Health showed her resting while he was frantically staging the phony crime. It also showed that her phone was moved 14 steps as he took it outside and dropped it for police to find, with the time stamp disproving his claim that she was alive when he left.

Sponsor Updates


  • The CoverMyMeds team stuffs backpacks for chronically ill campers and their families at Flying Horse Farms.
  • Imprivata partners with DigiCert to enable remote identity proofing for electronic prescribing of controlled substances.
  • EClinicalWorks will exhibit at the 2018 National Ryan White Conference on HIV Care & Treatment December 11-14 in National Harbor, MD.
  • The EHealthcare Leadership Awards honors Formativ Health as the Platinum winner in the Best Patient Access & Convenience category.
  • FormFast and Healthgrades will exhibit at the IHI National Forum December 9-12 in Orlando.
  • HCTec features former University of Virginia Health System CIO Rick Skinner in a new Executive Insights video on “Characteristics of a Trusted Partner.”
  • The Health Information Resource Center honors Healthwise with three Digital Health Awards for its patient education videos.
  • Imat Solutions releases a new podcast, “Phil Beckett of HASA Discusses Why Data Quality Matters.”
  • Wolters Kluwer joins the Healthcare Services Platform Consortium to help advance interoperability efforts and improve patient care.
  • Forrester ranks’s Analytics as top in the current offering category in its Healthcare Analytics evaluation.
  • Spok partners with Standard Communications to implement Spok Care Connect across VA hospitals.
  • Healthfinch releases a new e-book, “Implementing Standardized Refill Protocols.”
  • T-Systems offers its T-Sheets flu templates to all EDs and urgent care staff free of charge during National Influence Vaccination Week.
  • Solutionreach partners with Jive by LogMeIn to offer customers easier, faster communication options.
  • Nuance will integrate clinical data exchange capabilities from Halfpenny Technologies with its AI-powered clinical documentation solutions.

Blog Posts



Mr. H, Lorre, Jenn, Dr. Jayne.
Get HIStalk updates. Send news or rumors.
Contact us.


EPtalk by Dr. Jayne 12/6/18

December 6, 2018 Dr. Jayne 3 Comments


Following the ONC annual meeting at the end of November, I received an email that the slides and webcast would be “made available in the near future.” This always aggravates me after conferences, because by the time they make the content available, people have moved onto other things and momentum is lost. Especially with a relatively small (two-day) meeting, it shouldn’t be that hard to get the materials together since presumably people had to submit their slides in advance for review and approval. Webcasts also aren’t that hard to get online, especially if they’re not edited. Making the materials available quickly would help engage those who couldn’t be there and allow them to be part of the discussion.


I finally had some time to dig into the draft “Strategy on Reducing Regulatory and Administrative Burden Relating to the Use of Health IT and EHRs” document that ONC issued last week. It offers three goals for reducing clinician burden, including reducing time and effort to record information, reducing time and effort for reporting requirements, and improving EHR functionality and ease of use. I’m not sure whether or not I should read something into how those goals were constructed, since fixing the third goal would likely solve a big portion of the first one. When you dig deeper into the document, it becomes apparent that the first item refers not only to documentation effort, but the fact that the documentation required for billing is a burden above the documentation required for clinical care.

The usability discussion specifically addresses poor design of clinical decision support tools, including pop-ups that require “excessive interaction.” It also mentions poor implementation of electronic summary of care documents, lack of standardization around the presentation of clinical content, and the need for improvements to configuration and implementation processes that should “proactively engage the end user.”

One of the problems here is the fact that EHR vendors simply don’t want to spend as much money as would be needed to make EHR systems substantially better. I worked with one vendor that had a limited development budget, which essentially meant that the only work they could afford to do was that which was mandatory – either required for them to maintain certification or to address severe patient safety defects. Even minor patient safety defects were put into the deferred maintenance bucket to sit until more development hours became available, which often meant that they didn’t get fixed. When there’s not enough money to fix patient safety issues, that means that the “nice to have” and usability enhancements logged by customers over the years rarely made it to the requirement stage.

They also go in-depth about reporting issues and the fact that “regulatory requirements and timelines are often misaligned across programs and subject to frequent updates, which require significant investments from clinicians to ensure annual compliance. Government requirements are also poorly aligned with the reporting requirements across many of the federal payer programs in which clinicians may participate …” How about this — let’s put a freeze on federal reporting requirements until the federal payers can get their own houses in order. Present us with a unified set of reporting requirements that make sense clinically and actually allow us to drive the needle for clinical quality rather than just make us report for reporting’s sake.

While we’re at it, here are my other suggestions to solve the issues (although I’m sure they’d never be accepted): First, allow physicians to bill office visits based on time. Not the current “greater than 50 percent of this visit was spent in counseling and coordination of care” nonsense, but actually billing on time like a lot of businesses do, including attorneys, accountants, auto mechanics, and the guy who does my hair. If you’re more complex and take more time, allow us to be compensated for what we do. If you’re a quick visit, let us see you and get you on your way. One might say this may lead to abuses, so let’s put reasonable caps on it, such as a maximum of 16 hours a day. It can’t be any worse than our current system that doesn’t even detect fraudulent physicians that are billing many more procedures than they could possibly do in a day.

Second, let’s also address the usability issue by requiring vendors to issue standardized reports to their clients on how much development time is spent on regulatory requirements, remediation of software defects, patient safety issues, usability, new content, and the like. I know vendors hate this idea because they’re afraid the information will wind up in the public eye, but it’s important for customers to understand whether their vendor is really putting their money where their mouth is. This is hard for publicly traded companies, since actually spending money on development eats into the profit margin. Still, there has to be some kind of accountability for where the millions of R&D dollars are being spent.

While we’re at it, let’s also think about adding some requirements that will just make everyone’s lives easier. Let’s standardize to LOINC for laboratory orders and results. It’s there, it works, and it would save time for hospitals and healthcare organizations. Not just in the EHR, but with the laboratories – I’m tired of federal mandates that put the onus on the physicians, but don’t do anything to make lab vendors comply. I can’t even count the number of practices I’ve worked with whose vendors aren’t sending LOINC codes with results, but the practices have to have the codes mapped in the EHR, so much manual mapping occurs. Why not just fix the problem at the source? The strategy does allude to this a bit with standardization of medication information, order entry content, and results display conventions, but it’s shameful that we’re still talking about this a decade after the start of Meaningful Use.

What about patient matching and interoperability issues? There’s no federal funding for a universal identifier, but what if the vendors came together and created a voluntary one? Let patients opt in or opt out, but if they want to opt in, let’s give them a unique ID they can carry around to their providers that can be used to assist with matching. It’s clear that it’s never going to be a federal priority even if they blockages in front of it are cleared.

I ended up having to stop reading the document, because what I thought was going to be a quick blurb about it has rapidly turned into a semi-angry rant about the state of things. I’ll have to refine my thoughts before I enter my formal comments, which I will certainly do before the January 28, 2019 deadline. ONC plans to post all the public comments that are received, which should make for some entertaining reading in front of a nice fire on a snowy evening.

If you were in charge of all things healthcare IT, how would you fix these problems? Leave a comment or email me.


Email Dr. Jayne.

Morning Headlines 12/6/18

December 5, 2018 Headlines No Comments

Enzyme Health Adds $1.7M for Clinician Telemedicine Job Marketplace

Austin, TX-based Enzyme Health raises $1.7 million in a seed funding round led by Silverton Partners.

Teladoc Health Refutes SIRF Report Claims

Teladoc refutes claims of inappropriate relations and insider trading made in an article from the Southern Investigative Reporting Foundation.

Researchers find way to catch medical errors as they happen

Pascal Metrics develops software that uses machine learning and EHR data to detect and alert providers to medical errors in real time.

Readers Write: What’s Good for the Dentist is Good for the Medical Doctor as Well

December 5, 2018 Readers Write 4 Comments

What’s Good for the Dentist is Good for the Medical Doctor as Well
By Robert Patrick

Robert Patrick is president of dental at Vyne of Dunwoody, GA.


Medical professionals might be tired of the endless requirements of mailing x-rays or other documentation to the insurance company every time they file a claim. Some of them might simply want the ability to add their supporting documentation to claims electronically for easier adjudication.

While medical professionals continue to wait for developments and guidance related to the use of electronic attachment solutions and technologies, their dental colleague counterparts have no such obstacles. Even though there’s no formalized standardization from an organization like the Center for Medicare and Medicaid Services (CMS) for dental, there is a range of solutions that have permeated the sector and enjoy robust use by many thousands of dental practices. Why the disparity?

The simplest reason is that the solutions are readily available in the dental sector, Their use has been embraced despite there being little formal regulation or guidance related to submitting electronic attachments. For example, as long as the solutions are compliant with HIPAA, their use is fair game. Per recent reporting, some on the medical side of healthcare are waiting for a push toward standardization in the way electronic attachments are sent before moving forward with similar solutions.

According to reporting by MedPage Today, Robert Tennant, director of health information technology policy at the Medical Group Management Association — a trade group that represents medical practices — said that HIPAA includes a directive for the federal government to develop standards for electronic attachments. But the HIPAA provision still is not seeing traction or light of day. Even when the Affordable Care Act (ACA) was passed in 2010, it included a provision requiring the federal government to issue a final rule on standardizing electronic attachments, and a deadline of January 1, 2014, for doing so, but nothing yet.

The delay, Tennant speculates, might relate to how CMS can address “solicited” versus “unsolicited” attachments. Maybe the use of a secure attachment protocol or portal for data submission could eliminate this concern. For example, with dental electronic attachment solutions, providers can simply upload their supporting documentation via HIPAA-compliant software services. The respective payer is then notified that attachments are available for claim processing. No muss, no fuss. 

While there’s no requirement or mandate for dental providers to submit attachments, just like there is not one for medical doctors, dental providers are leading the way having embraced the move to electronic attachments years ago, unlike their medical colleagues. Any care professional can (?) make use of the technology, and there is a market on the medical side of the fence, so why the delay in adoption?

One potential issue is that some believe submitting attachments to be “a fairly complex transaction” for health plans to implement. “Since CMS also controls Medicare and Medicaid, they would be required by law to implement this standard, and maybe there is some pushback in terms of the cost to implement this transaction,” said Tennant in the report.

Is regulation on electronic attachments forthcoming for medical providers? The federal electronic attachment conversation continues and was included in the federal government’s unified agenda — a plan of action issued by the Office of Management and Budget — that might not be considered until later this year.

According to regulatory guidance, the electronic attachments rule must contain data formats to be used for the attachments. In 2016, the National Committee for Vital and Health Statistics, a public group that advises the Health and Human Services (HHS) secretary on health data issues, laid out its recommendations for electronic attachments, including suggested formats, in a letter to then-HHS Secretary Sylvia Burwell:

  • For the request for attachments, the group recommended using the ASC X12 format
  • For the response with a submission of attachment, the HL7 format is recommended
  • For the acknowledgement of the response, the ASC X12 format is recommended

For reference, the Accredited Standards Committee X12 (ASC X12) provides standards that can be used for nearly all facets of business-to-business operations conducted electronically. The committee aims to:

  • Develop high-quality e-commerce standards that are responsive to the needs of the standards user
  • Collaborate with other existing standards to make the standards developed more interoperable
  • Avoid any conflict, confusion, and duplication of effort
  • Publish and promote the standards along with their education
  • Drive the implementation and adoption of the standards developed by the committee

Health Level 7, or HL7, refers to a set of international standards for transfer of clinical and administrative data between software applications used by various healthcare providers. These standards focus on the application layer, which is “layer 7” in the OSI model.

The group also recommended that HHS define attachments as the “supplemental documentation needed about a patient(s) to support a specific healthcare-related event (such as a claim, prior authorization, referrals, and others) using a standardized format.”

One thought is that with such guidance and with the backing of CMS, there might be a reduced “provider burden.”

What about the payers? Why not a push by payers for standardized operations? Why don’t payers and providers just decide on standards and implement them without any government help? This hasn’t happened because payers argue that it will cost too much money to implement; no one is going to bother if vendors don’t create products for the providers. Some vendors, of course, are not willing to produce a solution for such without payer’s backing.

In medical care, it seems that everybody’s waiting for somebody else, and no one will do it until the government issues the standard. Perhaps these arguments are valid for physicians, but for dentists, this foundation already is laid. Perhaps infrastructure is the real problem for medical providers. Nevertheless, the technological capabilities exist and have for many years.

If electronic attachments were implemented in medical care, the result could be savings for both health plans and providers, according to the Council for Affordable Quality Healthcare (CAQH), a non-profit alliance of health plans and other organizations whose goal is to streamline healthcare administration. The 2017 CAQH Index report found that only six percent of medical attachments were submitted electronically that year, but the report also found that providers could it save 51 cents per claim – 30 percent of their current cost — if electronic submission were employed, while health plans could save $1.64 per claim, a 94 percent savings.

CAQH launched a project under its Committee on Operating Rules for Information Exchange (CORE) division — a group of about 130 organizations developing operating rules for healthcare administration — to scan and discover where the healthcare industry stands in relation to electronic attachments, including use of a standard format. The organization is examining the varying types of use cases for documentation and the products available in the marketplace to support an automated approach to move the industry forward.

While the number of electronic attachments exchanged is quite small in volume, at least for medical providers, there is a clear path in place that can be executed with or without the support of an organization like CMS or others, as we have seen on the dental side of the house. While doctors may have been waiting for some guidance since HIPAA’s creation in 1996, dentists have been successfully using electronic attachment solutions since at least 1997, and with great results.

Thus, if more than 60 percent of the dentists in America who need to send supporting documentation to payers to get paid for their service are doing so electronically, why can’t the medical professionals of America do the same? America’s dental payers have agreed to participate in electronic attachments while America’s medical payers seem to be waiting for a mandate.

Machine Learning Primer for Clinicians–Part 7

Alexander Scarlat, MD is a physician and data scientist, board-certified in anesthesiology with a degree in computer sciences. He has a keen interest in machine learning applications in healthcare. He welcomes feedback on this series at


Previous articles:

  1. An Introduction to Machine Learning
  2. Supervised Learning
  3. Unsupervised Learning
  4. How to Properly Feed Data to a ML Model
  5. How Does a Machine Actually Learn?
  6. Artificial Neural Networks Exposed

Controlling the Machine Learning Process

We’d like a ML model to learn from past experiences, so post-training, it should be able to generalize when predicting an output based on unseen data. The ML model capacity should not be too small nor too large for the task at hand, as both situations are not helping to achieve the goal of generalization.

Under and Overfitting

In the funny yet accurate description below: 

  • Knowledge sits in some form, but a ML model with not enough capacity will fail to see any relationships in the data.
  • Experience is the capability to connect the proverbial dots. Once a ML model achieves this level, training should stop. Otherwise,
  • Overfitting is when the model tries to impress us with its creativity. The ML model just had too much training and is now overdoing it.


Regression and Classification Examples of Under / Overfitting

We are searching for the sweet spot — a good, robust fit so the model would be able to generalize with unseen data.

The model should have sufficient capacity to be able to learn and improve and yet at the same time, not necessarily become the absolute best AI student on the training set.



Consider the left side of the above figure. The upper diagram displays data which is obviously not linear. Still, the ML model we’ve applied is linearly restricted – the model capacity is limited for the task.

The lower diagram displays a classification task, but the model is restricted to a circle. Its capacity is limited, so it cannot classify the dots better than with a circle separation line

When a ML model is underfitting, it basically doesn’t have enough or the right type of brain power for the task at hand or the model is exposed to a poor choice of features during training. We can help the ML model by:

  • Using non-linear, more complex models.
  • Increase the number of layers and / or units in a NN.
  • Adding more features.
  • Engineering more complex features from existing ones (using BMI instead of weight and height).

Underfitting is also called high bias and low variance and is one of the causes for a model to underperform. The model has a high bias towards a linear solution (in the regression example above) and a low variance in terms of limited variability of the features learned


You’ve trained your ML model for some time now and it achieves an amazing performance on the training set. Unfortunately, once in production, the ML model is only slightly better than just random predictions. What happened?

As the right side of the above figure shows, the model has used its large capacity to memorize the whole training set. The ML model became a memory bank for the training samples’ features, similar to a database. This overfitting caused the model to over train, to become “creative,” and also to become the best-ever on the training data. 

However, the overfitted model fails on real-life test data because it has lost the ability to generalize. We need the ML model to learn with each experience to generalize, not to become a memory bank

Overfitting is also called low bias and high variance, as the model has a low bias to any specific solution (linear, polynomial, etc.). The model will consider anything, any function, and it has a huge variance. Both factors contribute to an increased overall model prediction error.


How do we achieve a balance between the above two opposing forces of bias and variance? We need a tool to monitor the learning process — the learning curves — and a method to continuously test our model at each and every epoch, the cross-validation technique.

Training, Validation, and Testing Sets

Once you’ve got the data for a ML project, it is customary to cut a random 20 percent of samples, the test set and put it aside, never to be looked at again until the time of testing. Any transformation you plan on doing (imputing missing values, cleaning, normalizing, etc.) should be done separately on the training and test sets. 

This strict separation will easily prevent the scenario where normalizing over the whole data set and learning the average and standard deviation of the test set in the process may influence the model decision making in a way similar to cheating or letting the model know information about the test set, which the model should not know. The rest of the data after removal of this test set is the original training set.

As the model is going to be exposed to the training data multiple times — with different hyper-parameters (see below), architectures, etc. — if we allow the model to “see” the test data repeatedly, the model will eventually learn the test set as well. We want to prevent the model from memorizing all the data and especially to prevent the model exposure to the test set .

The original training set is used in a cross-validation scheme, so the same training set can be used also for validating each learning epoch. In a fivefold cross-validation scheme, we create each epoch, a 80 percent subset from the original training set and a validation set from the remaining 20 percent. Basically, we create a mini-test set for each learning epoch — a validation set — and we move this validation set within the original train set with each learning epoch (experiment in the figure below):


Learning Curves

With a cross-validation arrangement as detailed above, we can monitor the learning process and identify any pathological behavior on behalf of our student ML model during training.


Underfitting learning curves above show both the training and the validation curves remaining above the acceptable error threshold during the epochs of the learning process. Basically the model does not learn: either not enough model capacity or not good, representative enough features it can generalize upon. We need to either increase model capacity, increase the number or complexity of the features, or both. Adding more training samples will not help.

Overfitting learning curves show that pretty early during the learning process, the model started overfitting, when the two learning curves separate. The training curve continued to improve and reduce the training error, while the validation curve stopped showing improvement and actually started to deteriorate. Decreasing the model capacity, decreasing the number of features, or increasing the number of samples may help.

Perfect fit happens when the validation error is below the acceptable threshold and it starts to plateau and separate from the training curve. At that number of training epochs, we should stop, call an end for the learning session, and give our ML model a short class break.

Learning Rate

A ML model has parameters (weights) and hyper-parameters such as the learning rate.


With a too-low learning rate, the model will take its time to find the global minimum of the cost function (left in the above figure). Too high a learning rate will cause the model will miss the global minimum because it jumps around in too large steps. Modern optimizers can automatically modify their learning rate as they approach the minimum in order not to miss it with a too large jump above it.

Data Augmentation

Usually collecting more samples to feed an overfitting model is a time, money, and resource-consuming activity. Consider an image analysis ML model that identifies between dogs and cats in an image. Until recently, this exercise was used by CAPTCHA to distinguish between humans and malicious bots trying to impersonate humans. Machines recently achieved the same level as humans, so CAPTCHA is not using this challenge any more. Nevertheless, dogs vs. cats became one of the basic, introductory exercises in computer vision / image analysis.

While developing such an image classification model, one usually increases the model capacity gradually until the model starts overfitting. Then its customary to add data augmentation, a technique used only on the training set, in which images are being reformatted randomly around the following image parameters:

  • Zoom
  • Scale
  • Brightness
  • Skew
  • Mirror around vertical / horizontal axes
  • Colors

By exposing the algorithm during training to a more diverse range of images, the ML model will start overfitting at a much later epoch, as the training set is more complex than the validation set. This in turn will allow the model to bring the validation error to an acceptable level.

Data augmentation allows a ML model to realize that a cat looking to the right side is still a cat if it looks to the left side. With data augmentation, the ML model will learn to generalize that a dog is still a dog if it is scaled to 80 percent, flipped horizontally, and skewed by 20 degrees. No animals were harmed during this data augmentation exercise.



Eventually, a big enough model will start overfitting the data, even if the training set has been augmented. Another technique to deal with overfitting is to use a regularizer, a model hyper-parameter. Basically it penalizes the model loss function on any large modifications to the model weights. Keeping the changes to the weights within small limits during each epoch is important, as we don’t want the model to literally “jump to any conclusions.”


An interesting, different, and surprisingly very efficient approach to overfitting that can prevent a ML model from learning the whole training set by heart is called dropout. Like data augmentation above, dropout technique is used only on the training set. It takes out randomly up to 50 percent of a NN layer units from one learning epoch, like randomly sending home half of the students for one class. How can this strategy prevent overfitting? 

The analogy with these students being dismissed from that class / epoch caused all the other units (students) in the layer to work harder and learn features they were not supposed to learn otherwise. This in turn zeroed the weights for up to half of the units while forcing other units to modify their weights in a way that is not conducive towards a “memory bank.” Shortly, dropout destroys any nascent memory bank a ML model may try to create during training.


Once training is completed, hyper-parameters have been optimized, data has been re-engineered, the model has been iteratively corrected, etc. then and only then one brings out the hidden testing set. We test the ML model and its performance on the test set will hopefully be close to its real-life performance.

Next Article

Predict Hospital Mortality

Morning Headlines 12/5/18

December 4, 2018 Headlines No Comments

Curavi Acquires TripleCare, Prepares to Rapidly Expand Telemedicine Services

UPMC-backed Curavi Health acquires competitor TripleCare, a telemedicine vendor that caters to the post-acute, long-term care market.

MEDITECH UK Announces Acquisition of Centennial and the Formation of Medical Information Technology UK LTD

Meditech acquires its London-based partner Centennial Computer Corporation as part of its creation of Meditech UK.

Interoperability 2018: Real Progress with Patient-Record Sharing via CommonWell and Carequality

A KLAS report finds that most EHR vendors are progressing well toward supporting a national patient record network now that CommonWell is connected to Carequality.

Queensland digital hospitals program facing $250m cost blowout

In Australia, Queensland Health’s hospital EHR project will run $188 million over budget if implemented in the remaining hospitals, with an auditor-general’s report noting that Cerner can name its price for contract extensions knowing that alternative systems haven’t been considered.

Subscribe to Updates



Text Ads

Report News and Rumors

No title

Anonymous online form
Rumor line: 801.HIT.NEWS



Founding Sponsors


Platinum Sponsors
























































Gold Sponsors

















Reader Comments

  • FRANK POGGIO: Shows to go ya...answering black and white Jeopardy questions is a far cry from the massive grey area of medicine/pharma...
  • Number Cruncher: You are right AC. The cost is seriously underestimated here. Just looking at the numbers - $1 B for 5 years = $200 M ...
  • Abraham Van Helsing: Re Theranos. Will be interesting to follow the saga. As I and others had noted going back 2+ years, something was obvi...
  • Prof. Moriarty: Re: Watson pull out. I've not been directly involved with this product, but from its beginning I have always seen Watso...
  • mih: Of course they can, and for much much cheaper. But why would they do it? Existing arrangement works for everyone in the ...
  • Andrew M. Harrison: Thanks for (actually) reading our paper. I enjoyed the story of your friend, as well as the translation of numbers to em...
  • Mike: I would love to see this type of discussion around Blockchain. It is being hyped heavily currently. Yet, I wonder how we...
  • Brian Too: Just slightly off-topic, but I recently heard an interesting downtime rule-of-thumb: Every hour of downtime requires 2 ...
  • James E Thompson: AI in particular isn't disruptive until it can offer an effective alternative against which a go/no-go decision makes se...
  • Former Newspaper Guy: I applaud your attention to grammar and style. In high school, I worked for the local newspaper in the sports department...

RSS Industry Events

  • An error has occurred, which probably means the feed is down. Try again later.

Sponsor Quick Links