Healthcare AI News 6/14/23

News

The American Medical Association’s House of Delegates considers a resolution that would urge doctors to educate patients about the risk of large language models. It also directs AMA to work with the federal government to protect patients from inaccurate, AI-generated medical advice.

Dartmouth launches the Center for Precision Health and Artificial Intelligence.

In England, the British Labour Party recommends that the UK create a mandatory licensing model for companies that do AI work.

A UK pathologist who pioneered the use of AI to diagnose prostate cancer is advocating the use of similar AI pathology tools to diagnose breast cancer within three days and help finalize treatment recommendations within a week. He says that diagnosing cancer isn’t hard, but it takes more time to review biomarkers to determine optimal treatment, creating patient uncertainty and straining staff resources.

Business

Healthcare analytics vendor BurstIQ acquires Olive’s AI business intelligence solution.

Healthcare analytics vendor Apixio acquires ClaimLogiq, which offers health plan claims technology.

The CEO of drug manufacturer Sanofi says it will be the first pharma company that is powered by AI at scale, using AI and data science to discover drugs, design clinical trials, and improve manufacturing and supply chain processes.

Accenture will spend $3 billion to double its AI-focused employee headcount to 80,000, incorporate generative AI in its client work, and help customers use the technology.

Opinion

A Health Affairs opinion piece predicts that AI will set interoperability back, as providers will use technical and legal tools to prevent large competitors from using their data to train large language models to create competing consumer medical services.

Research

Researchers test ChatGPT’s ability to provide venomous snakebite advice to consumers, finding that it gives accurate and useful responses, including recommendations to seek medical care when appropriate, but with limitations involving outdated knowledge and lack of personalization.

Other

Peter Lee, VP of OpenAI investor Microsoft, says he was “alarmed” to learn that within the first three days of the release of ChatGPT, doctors were not only using it, but were asking it to help them communicate with patients more compassionately. A former physician executive at Microsoft says that he was “blown away” by ChatGPT’s ability to help him communicate empathetically with a friend with cancer, but observes that doctors don’t evangelize its use because that would require admitting that they aren’t good at talking to patients.

A health technology firm creates Lifesaving Radio, an AI-based radio station whose playlist of hard rock music is optimized to get surgeons “in the zone” of relaxed high efficiency during surgery. The playlist was developed using Spotify’s AI DJ analytics technology to find songs with the ideal tempo, key, and loudness for surgeons. It features an AI-powered DJ that calls out the surgeons and team members by name. The channel’s first music set is an AC/DC-inspired collection called “Highway to Heal,” which includes parodies such as “You Sewed Me All Night Long.”

Contacts

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

Healthcare AI News 6/7/23

News

Mayo Clinic will use Google’s low-code Enterprise Search in Generative AI App Builder to combine Google search with generative AI to provide information to clinicians and employees.

Boston Children’s Hospital, which drew recent attention for posting a job for a ChatGPT prompt engineer, tells Mashable how it may use the technology to solve real-world problems such as helping patients navigate its buildings or to create patient-specific discharge and rehabilitation instructions using their EHR data.

Wolters Kluwer Health acquires Invistics Corporation, which offers AI-enabled software for drug diversion detection.

Primary and urgent care chain Carbon Health launches hands-free AI charting across all of its clinics and providers. Appointment audio is analyzed with AWS Medical Transcribe, then processed with GPT-4 to generate an EHR-ready notes document. Early studies show that unedited accuracy is 88%, charts are 2.5 times more detailed than with manual entry, and chart completion time was reduced from 16 minutes to four. The company says it may sell the software to other provider groups.

AvaSure adds AI-powered enhancements to its TeleSitter virtual care platform in which the virtual safety attendants will be alerted when a patient who is at risk for a fall or elopement tries to sit up or leave the room. The company is partnering with Kinometrix, which applies AI predictive analytics to EHR data to identify patients with fall risk.

FDA issues 510(k) clearance for Ezra’s AI-enhanced MR images for early cancer detection, which the company says would reduce full-body MRI screening cost by 30%. The company says it hopes to offer a 15-minute, $500 full body scan versus today’s one-hour scan that costs $1,950.

Business

LeanTaas announces a generative AI solution that gives leaders insights about patient flow, scheduling, block management, and other capacity management measures.

Clarify Health is beta testing a generative AI solution called Clara, which analyzes claims data to recommend care opportunities.

ObjectiveHealth launches an AI-powered clinical trials feasibility and recruitment platform that identifies the most qualified study candidates using risk factors and biomarkers from their EHR data, then keeps participants engaged by streamlining appointments and communications.

SafelyYou, which offers AI video and remote monitoring support for dementia care in senior living communities, including fall detection, raises $30 million in funding.

Research

University of Colorado medical school researchers are studying patient experiences with AI that simulates conversation, raising questions about potential bias created by the appearance of the chatbot’s avatar and the ethics of nudging people into healthcare behaviors without restricting their choices.

Researchers find that ChatGPT provides evidence-based answers to public health questions with 91% recognition versus a cumulative 5% recognition among Alexa, Siri, Google Assistant, Cortana, and Bixby. However, ChatGPT, along with the other tools, rarely offers referrals and instead gives advice. The authors conclude that public health agencies should publish a database of recommended resources that AI could use in fine-tuning its responses to refer people to vetted resources.

NYU medical school researchers describe NYUTron, a large language model they trained to on EHR notes to forecast clinical and operational event in real time with better accuracy compared to traditional methods. They tested the system at NYU Langone Health System on five predictive tasks: 30-day readmission, in-hospital mortality, co-morbidity index, length of stay, and insurance denial. The authors say that models trained on highly tailored data are more useful than using larger, less-specific datasets, but they still require a lot of computational horsepower.

Other

Kenya-based “contract cheaters” who ghost write essays and complete assignments for students in wealthy countries worry that ChatGPT could replace them. One of them says that most of his customers are studying nursing and healthcare. A January study found that 89% of the surveyed students are using ChatGPT to help them with homework and 53% used it to write an entire essay.

South Korea is using AI-powered image analysis to assess the health of North Korea’s leaders, determining that Kim Jong Un weighs over 300 pounds, has dark circles under his eyes, and may suffer from a sleep disorder.

Contacts

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

Healthcare AI News 5/31/23

News

Microsoft adds intelligent meeting recap to Teams Premium, which generates meeting notes, a task list, and personalized timeline markers.

In Australia, five-hospital South Metropolitan Health Service orders doctors to stop using ChatGPT for work-related activity, citing confidentiality concerns. SMHS found that at least one doctor used ChatGPT to create a discharge summary, backtracking on an earlier statement in which it said that several doctors were creating notes in ChatGPT and then pasting them into the EHR.

Business

AI chipmaker Nvidia hits a market capitalization of $1 trillion following a strong quarterly report, joining nine companies that have reached that mark including Apple, Microsoft, Alphabet, Amazon, and Saudi Aramco. A $10,000 investment in the company five years ago would be worth nearly $60,000 today.

Hyro, which offers conversational AI-powered healthcare workflow and conversation solutions, raises $20 million in a Series B funding round.

In England, AI drug discovery company Benevolent AI, which went public last year in a SPAC merger, will cut half of its workforce and scale back its laboratory facilities. The company had hoped to license its AI-designed drug candidate for atopic dermatitis, but it failed to improve symptoms in early-stage clinical trials.

Research

Researchers use AI to design an antibiotic for treating hospital-acquired infections caused by the broadly resistant Acinetobacter baumannii bacteria. Researchers tested thousands of drugs for their ability to kill the bacteria or slow its spread, trained AI on the results, and then ran the resulting AI model against 6,700 other drugs to generate a 240-drug short list of candidates. The AI-chosen drug, abaucin, targets the bacteria specifically and therefore is less likely to cause drug resistance. Laboratory and clinical testing will take several years, with the first AI antibiotics expected to reach the market in 2030.

Opinion

Google Health debunks five myths about medical AI:

• The more data, the better. Data quality matters more and expert adjudication in touch cases helps improve labeling quality.
• AI experts are all you need. Building an AI system requires a multidisciplinary team.
• High performance provides clinical confidence. Real-world validation is needed to make sure the model generalizes to real-life patients.
• AI fits easily into workflows. AI should be designed around human users.
• Launch means success. AI systems must be monitored to detect potential issues when patient populations or environmental factors change.

Other

Nvidia profiles Nigeria-based physician, informaticist, and machine learning scientist Tobi Olatunji, MD, MS, who started Intron Health to transcribe physician dictation using AI with 92% accuracy across 200 African accents. The company was supported by Nvidia’s startup program. He earned a Georgia Tech computer science master’s and a UCSF master’s in medical informatics after he completed medical school in Nigeria.

Contacts

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

Healthcare AI News 5/24/23

News

Microsoft will integrate ChatGPT into Windows 11, where it will run in its own Copilot window as a personal assistant to perform Windows commands and summarize documents that are dragged into it. The user rollout will start in June.

Tell, whose app allows users to seek advice from medical experts, integrates ChatGPT to translate medical jargon into accessible language.

OpenAI says that AI systems will exceed expert level in most domains within 10 years and recommends steps to mitigate its risks:

• Coordinate development efforts across countries and hold companies to a high standard of responsibility.
• Create an organization similar to the International Atomic Energy Agency provide oversight and inspection AI efforts that exceed a specific level of capability or resource requirements.
• Develop technical capabilities to make superintelligence safe.

OpenAI launches a ChatGPT app for the IPhone.

In Pakistan, the government of Punjab launches a two-hospital pilot of using AI to assist in diagnosis.

Google launches the Google for Startups Growth Academy: AI for Health program for companies based in Europe, Middle East, and Africa. Startups from seed to Series A will be offered a three-month virtual program of tailored workshops, collaboration, and mentorship.

Business

Alicja AI offers a $500 per month enterprise clinical documentation tool that integrates with EHRs. 

Research

ChatGPT has passed several medical exams, but researchers find that it falls just short of passing the American College of Gastroenterology Self-Assessment Tests.

A University of Arizona Health Sciences-led study finds that participants are almost evenly split in preferring a human doctor versus AI for diagnosis and treatment. The authors recommend further research about how AI can be incorporated into the work of physicians and the decision-making process of patients. 

Other

Business Insider profiles ED physician and two-company VP of innovation Joshua Tamayo-Sarver, MD, PhD, who says that it “probably should be embarrassing” that has sometime uses ChatGPT to explain medical issues in patient-friendly terms. He concludes that ChatGPT is “the most brilliant, talented, often drunk intern you could imagine” that is great at explaining concepts but not good at diagnosis or other tasks that require clinical reasoning.

Kaiser Permanente ED doctor and technologist Graham Walker, MD pens an excellent piece on how he views AI as a physician:

• AI can pass a medical school exam, which involves basic multiple choice questions, but that capability is not very related to interacting with patients to determine their multiple issues and their viewpoints about options.
• Doctors know how to successfully address a patient problem up to 95% of the time due to their specialization, residency training, and repeated exposure to the same common issues, and therefore would see no value in asking a “medical bot” for recommendations.
• Where AI could help is to differentiate among possible problems that exhibit similar symptoms.
• AI might offer a convincingly objective second opinion to a patient who is told, for example, that they don’t need antibiotics for a viral infection.
• He says he would “virtually hug and kiss a digital agent” that could generate discharge instructions, describe the logic behind the chosen medical plan, and answer questions are likely to have.
• AI could help identify and correct confirmation bias, where the doctor needs fresh perspective to see that evidence might not support the suspected diagnosis.
• AI could help steer an ED patient to local sources of help that might be better than the ED.
• AI could help doctors and patients understand why lab tests may not be indicated and how to react to positive or negative results.

Contacts

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

Healthcare AI News 5/17/23

News

OpenAI rolls out 70 third-party plug-ins for ChatGPT that allow Plus users to access Internet content, retrieve information from specific sites such as Expedia, analyze online PDF documents, and create integrations with Zapier. The Link Reader plug-in, which I tested above, allows ChatGPT to analyze live web pages.

The World Health Organization expresses enthusiasm for using AI in healthcare, but recommends cautious adoption to avoid errors and patient harm due to AI training bias, incorrect or incomplete responses that appear authoritative, and its possible use to generate convincing health disinformation.

Google enhances its Bard generative AI tool with a new version of its PaLM 2 large language model, adds connectivity to a user’s Google apps and services, and provides support for plug-ins.

Google also introduces Duet AI for Google Workspace, which adds AI creation to Gmail and Docs similar to Microsoft’s Copilot for Office 365.

Google’s Med-PaLM 2 is being tested to analyze diagnostic images and answer questions about its summary.

Business

Andreessen Horowitz’s A16Z invests $50 million in seeding funding for Hippocratic AI, which offers a safety-focused large language model for healthcare. The company says that its model outperforms GPT-4 and uses clinician-validated model training that avoids bias and sources that aren’t evidence based.

BeeKeeperAI announces GA of EscrowAI, which allows HIPAA-compliant research on full PHI without exposing patient data or the algorithm’s intellectual property. The company says it acts as a “middleman and matchmaker between data stewards and algorithm developers” to reduce the effort and cost of data projects by 50%. The company was created at UCSF’s Center for Digital Health Innovation by that organization’s founder, UCSF Health Chief Digital Transformation Officer Michael Blum, MD, who serves as BeeKeeperAI’s CEO.

BayCare Health System and HealthPrecision will co-develop AI tools for nursing decision support and clinical documentation, where the system will translate nurse dictation into EHR documentation and suggest additional steps. The co-founders are Eyal Ephrat, MD and Sonia Ben-Yehuda, who together created MedCPU and PeriGen.

Research

Kaiser Permanente will award grants of up to $750,000 to three to five US health systems to study the use of AI and ML to improve diagnoses and outcomes.

Opinion

FDA Commissioner Rob Califf, MD says that large language models appear to be ushering in a fourth industrial revolution that includes healthcare, but they need to be regulated now to avoid society being “swept up quickly by something that we hardly understand,” also adding that methods need to be developed to measure and adjust the operating characteristics of algorithms based on real-life experience.

Physician and author Elisabeth Rosenthal, MD, MA says in an opinion piece that startups are offering apps and devices that address mental health therapy as an alternative to short-supply professionals, but most of them aren’t supported by outcomes research, haven’t been reviewed by FDA, and contain small print that reminds users that they are not intended to provide actual clinical services. Experts say the apps use a workbook-like approach for specific problems, which doesn’t address complexity or offer the empathy between patient and doctor that many therapy types require.

Other

The UK government issues guidance about when sellers of software as a medical device must report adverse incidents to the Medicines & Healthcare Regulatory Agency.

Contacts

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

Healthcare AI News 5/10/23

News

AI-powered drug discovery could represent a $50 billion opportunity for big pharma according to a Bloomberg article, which notes that one drug company paid $4 billion for drug that was developed by a startup in just six months. The psoriasis drug is expected to generate up to $3.7 billion in annual sales.

Microsoft eliminates the waitlist for the AI version of its Bing search engine, adds persistent chats and visual capabilities, and announces third-party plug-in capability.

Microsoft expands access to the preview version of Microsoft 365 Copilot, which adds AI capability to the whiteboard function of Teams, integrates the DALL-E image generator into PowerPoint, offers writing suggestions from within Outlook, helps users create lists and generate ideas in OneNote, and creates a learning plan in Viva Learning.

AI drug discovery platform vendor Insilico Medicine offers a free course on using its PandaOmics generative AI tool for disease modeling and target discovery.

Salt Lake City-based AI drug discovery company Recursion, which has five products in Phase 1 or later studies, acquires two competitors.

The CEO of Databricks says that the best way to invest in AI is to think of it as a gold rush, in which you want to be the company selling picks and shovels to gold-seekers. He also recommends investing in companies that control proprietary data, such as from medical records.

Lucem Health, a portfolio company of Mayo Clinic Platform that turns AI advancements into point-of-care solutions, raises $7.7 million in a Series A funding round.

Research

Researchers apply AI to early, large-scale diagnosis of pancreatic cancer, projecting that analyzing the EHR data of one million people could identify 70 who are highest risk who have not been diagnosed with the aggressive cancer.

Opinion

An Atlantic article warns that ChatGPT is following a similar trajectory as mostly-forgotten IBM Watson, whose over-publicized rush to commercialization left it with little market visibility as it ended up being used to automate mundane, back-end B2B tools. The article dismisses IBM’s failed Watson oncology work with Memorial Sloan Kettering and MD Anderson Cancer Center as “a swole Clippy fed on enterprise data and techno-optimism, [that] could barely read doctors’ handwriting, let alone disrupt oncology.” The author says that ChatGPT, like Watson, is being commercialized to create “banal consumer and enterprise software” that is invisible to users.

A BMJ Global Health analysis piece calls for regulation of AI and a moratorium on self-improving artificial general intelligence, saying that AI poses threats to health and well-being.

Former VA Secretary David Shulkin, MD lists “25 Ways Generative AI Will Change Healthcare” that include specific companies:

• Ibex – improve diagnostic accuracy from pathology.
• Galileo – patient navigation of health and treatment options.
• Cadence and Kaii – remote patient monitoring.
• Google Health and Babylon Health – improved diagnosis using electronic patient records.
• Saama – drug discovery.
• Dexcare – access to care.
• Indegene and Edocate – new methods of interactive medical education.
• Cedar – billing.
• TriNetX and MD Clone – clinical trial design.
• Theator – surgical learning systems.
• Eleos – analysis of behavioral health session to improve outcomes.
• CareJourney, Datavant, and Voiceitt – reducing disparities and addressing disabilities.
• Aldoc – early intervention using unstructured data and image analysis.
• Ixlayer and TytoCare – point-of-care testing.
• Sparta Science – identifying fall risk.

In a preview of how ChatGPT might find its way into health system digital front doors to improve access, Expedia is testing a chat-based trip planner, while a Zillow add-on allows conversational access to its real estate listings.

Other

UPMC CTO Chris Carmody says that the health system is working on AI chatbots that can answer patient questions about insurance coverage, but adds that the goal is to support rather than replace healthcare professionals.

A Tennessee radiologist writes 16 research papers in four months, five of which were published in journals, using ChatGPT.

An ED doctor uses ChatGPT to perform roleplay that teaches doctors how to break bad news to patients. ChatGPT provided the scenarios and feedback.

Resources and Tools

• Holly – sends weekly lists of LinkedIn candidates for open positions and engages them in conversation to book interviews.
• EvidenceHunt – performs advanced searches of PubMed entries.
• Law ChatGPT – creates legal documents from user prompts and templates.

Contacts

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

Healthcare AI News 5/3/23

News

Cloud content management vendor Box launches a ChatGPT-powered tool that allows users to find information in their own stored content and to create content from existing information.

Shares in Chegg, which offers a student learning platform that includes homework answers and tutoring, drop 50% as the company warns that its growth has slowed as students instead to turn to ChatGPT.

Boston Children’s Hospital lists a job opening for an AI prompt engineer for its innovation accelerator.

A Washington Post article says that ChatGPT can help people develop custom meal plans for conditions such as polycystic ovary syndrome, but notes that it cannot take health history into account, can’t see new developments that occurred since its fall 2021 data cutoff, and can provide incorrect information or misunderstand the user’s requests.

Centene will sell its AI-powered value-based platform vendor Apixio, which it acquired in December 2020, to New Mountain Capital.

A law firm notes the legal challenges that are arising from the use of generative AI:

• Is a license required to train a model on copyrighted material?
• Is a copyright infringed when AI generates images, music, or other output that is similar to the works it was trained on?
• Is the Digital Millennium Copyright Act violated when AI is trained on images that contain copyright watermarks that are not replicated into newly created images?
• Is an artist’s right of publicity violated by generating works by AI that was trained on their style?
• If AI generates a new image after being trained on an image that contains a trademark watermark, does that confuse the market or cause damage to the copyright holder when the image is of poor quality or taste?
• How do open source or creative commons licenses apply to material that is used for AI training?

Opinion

An AMA article notes that its House of Delegates uses the term “augmented intelligence” instead of “artificial intelligence”to emphasis its assistive role. AMA says it is working with FDA to regulate AI tools for safety, clinical validation, and lack of bias. AMA’s president-elect says that doctors should ask four questions before using AI tools in their practice: (a) is its efficacy backed by clinical evidence; (b) will doctors be paid for using it; (c) who is accountable in the event of a data breach; and (d) will it improve outcomes, efficiency, or value in the doctor’s own practice.

Medical resident Teva Brender, MD identifies time-sapping tasks that could at some point be performed by ChatGPT in a JAMA Internal Medicine opinion piece:

• Prepare discharge instructions that review the hospital stay, medication changes, and scheduled appointments in patient-friendly language.
• Draft a patient message that explains that their lab test indicates diabetes that will be discussed at their next visit.
• Write a HIPAA-compliant letter for a patient’s necessary time off from work.
• Fill out prior authorization forms using information extracted from the EHR.
• Prepare patient documents in languages other than English, or adding definition links to jargon-laden EHR documents.
• Turning a patient’s history into a narrative.
• Checking EHR data to update a problem list.
• Gather information for medication reconciliation.
• On the negative side, he notes that AI might overpromise and underdeliver on addressing burnout similarly to the use of scribes; “note bloat” could worsen if AI-generated text behaves like copy-and-paste; and that AI could perpetuate health disparities and invite medicolegal risk.

In China, a woman beats a hospital robot “receptionist” in its lobby, with observers noting that patients are frustrated by the replacement of nurses with technology.

Resources and Tools

• The Superhuman newsletter offers hiring-related prompts for ChatGPT, such as, “I am interviewing candidates for the role of [insert role]. Create an interview with 3 rounds that test for the following traits: culture fit, growth mindset, learning ability, and adaptability. Also create one technical assignment to test their technical ability.“

Contacts

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

Healthcare AI News 4/26/23

News

Apple is developing a paid, AI-powered health coaching service called Quartz that uses the Apple Watch to encourage healthy behaviors. The company will also release a dedicated version of its Health app for the IPad.

Google merges its DeepMind and Brain team from Google Research into a single unit called Google DeepMind, which the company says will accelerate its progress in using AI to solve humanity’s biggest challenges.

Alphabet CEO Sundar Pichai says in the company’s earnings call that Google is investing heavily in AI, but downplays the potentially negative effect that it will have on search advertising. Microsoft CEO Satya Nadella says in his company’s earnings call that Bing has gained market share following early incorporation of AI in its search, and calls AI’s impact “a generational shift in the largest software category – search.”

OpenAI adds the ability for ChatGPT users to turn off chat history to avoid having their conversations used for model training.

Life sciences technology vendor Trinity Life Sciences offers its customers AI-driven insights powered by WhizAI. The conversational AI and visualization platform will provide analytics for Trinity’s enterprise reporting platform, allowing users to type or ask business questions to generate visualizations and dashboards.

PricewaterhouseCoopers will spend $1 billion over the next three years to automate parts of its tax, audit, and consulting services using ChatGPT and Microsoft Azure.

3M Health Information Systems will use AWS services, including machine learning and generative AI, to further enhance its clinical documentation and virtual assistant solutions that include ambient intelligence.

NIH awards Case Western Reserve University and University Hospitals a $6.2 million grant to study the use of AI to predict cardiovascular disease from the calcium scoring of CT images.

Unfold AI, an FDA-cleared prostate cancer management platform from Avenda Health, is being used in an unnamed US research hospital to offer treatment selection, planning, guidance, and follow-up.

Federal regulators will enforce existing civil rights laws when generative AI tools perform “digital redlining” in introducing bias into decisions about hiring, credit, and housing. EEOC Chair Charlotte Burrows says that rapid AI development is a “new civil rights frontier.”

Research

An analysis of the use of AI-powered conversational assistants for support inquiries shows a 14% increase in the number of issues resolved per hour, mostly improving the performance of inexperienced workers who are supported by the stored knowledge of their more experienced peers.

Opinion

The Verge notes the legal situation created by a viral song hit that was created using AI, prompting Drake’s record label to warn that training AI using the music of its artists is a breach of copyright laws. Skeptics question whether it was actually a record label PR stunt. The legal issue is that Google-owned YouTube pulled the track due to a take-down notice from the record label, but the record label doesn’t own the song, which then pits Google’s AI-enriched search against YouTube’s music partners about what constitutes fair use.

A HIMSS23 panel warns health systems that hackers will be launching AI-powered cyberattacks, predicting the rise in AI-operated ransomware and ChatGPT-crafted phishing emails that sound more authentic, but with security technology vendors also using the technology to build smarter defenses.

Other

Pharma bro and ex-con Martin Shkreli rolls out a medical chatbot called Dr. Gupta that he says will be a “replacement for all healthcare information” that will leave WebMD “dead in the water.” He says the new large language model was trained in data from the web and from online medical journals, using both GPT 3.5 and GPT-4 to improve performance with some sacrifice in response speed. Users can ask five questions for free, then sign up for a $20 per month subscription to Dr. Gupta. When asked about HIPAA concerns related to users submitting PHI to an online chatbot, Shkreli suggested on Twitter to read the terms to decide whether to use it. Meanwhile, Twitter has suspended Shkreli’s account, although his original post on Substack remains, where he argues that healthcare is expensive because of the artificially constrained supply of healthcare professionals. Shkreli also wrote a 200-page novel using ChatGPT. 

A Washington Post study of the 15 million websites on which Google trains its C4 data set places HIStalk at #1,996, representing 3.4 million of the data set’s tokens.

Resources and Tools

• DocLime – allows analyzing documents stored in PDF form to generate summaries, answer questions, or provide research citations.
• Zain Kahn, “The AI Guy,” lists ways to prompt ChatGPT for learning. Examples: “”Proofread my writing above. Fix grammar and spelling mistakes. And make suggestions that will improve the clarity of my writing” and “I want to learn about [insert topic]. Identify and share the most important 20% of learnings from this topic that will help me understand 80% of it.”

Contacts

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

Healthcare AI News 4/12/23

News

Google will add conversational AI to its search engine, although it says that AI chatbots pose little threat its search business.

Duke Health and SAS will jointly explore using AI to develop analytics solutions for improving health equity and optimizing outcomes.

A Denmark-based startup that has developed a AI-powered nurse assistant raises a $5 million funding round. Its technology provides an overview of unit events, issues care alerts, monitors sleep, issues fall warnings, and monitors in-room cameras that track patient activities.

AI and NLP company John Snow Labs releases NLP Test, an open source Python library that runs 50 out-of-the-box tests that cover accuracy, fairness, bias, representation, and robustness. Customers of its Healthcare NLP include drug companies, CVS Health, Optum, Cincinnati Children’s, and the FDA.

Truveta launches an AI model that turns EHR records into research data for studying patient care and outcomes. The company says that today’s health research is based on claims data, while its new product uses data that is focused on clinical outcomes without commercial bias. It notes that GPT-4 is not trained on actual medical records, so it sometimes “hallucinates” or makes up information.

Research

Students at Stanford University create a tool that combines AI with augmented reality to display conversational responses on AR glasses to help people who have social anxiety or challenges during presentations or job interviews.

A Hong Kong-based drug discovery company uses generative AI to discover a new cancer drug. The AI-focused company wrote a peer-reviewed journal article in 2016 in which it said that generative AI could be used to develop new drugs. It submitted nine AI-developed drugs for pre-clinical studies in 2022.

Opinion

Medical authors suggest potential use cases for what they call generalist medical AI:

• Creating draft radiology reports that describe normal and abnormal findings that take the patient’s history into account.
• Serve as a surgical team assistant that can review and annotate video streams of procedures, warn verbally when procedure steps are skipped, and find and read out relevant literature.
• Provide bedside clinical decision support tools that parse EHR sources, summarize a patient’s current state, predict how that state will change, and recommend treatments.
• Create drafts of electronic notes and discharge reports for clinician review.
• Provide patients with detailed advice and explanations as a chatbot.
• Generate protein amino acid sequences and structures from text prompts.

Other

Boston Children’s Hospital posts a job for an AI prompt engineer to work on its Innovation and Digital Health Accelerator. 

Resources and Tools

• MemoryGPT – allows AI chat users to save chats indefinitely, giving the system a memory.
• Speechmatics Ursa – claims to be the world’s most accurate speech-to-text system with support for 48 languages.
• AI Toolbox for Innovators – test, pivot, or generate ideas with free tools.

Contacts

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

Healthcare AI News 4/5/23

News

Carle Health will use Scanslated software to convert radiology interpretation notes into patient-friendly language for reading in MyChart, including in Spanish when requested and with illustrations. The software was developed by Duke Health vascular and interventional radiologist Nicholas Befera, MD, who co-founded Scanslated and serves as its CEO.

Bloomberg develops a generative AI model, trained on 50 billion parameters, that can recognize business entities, assess investor sentiment, and answer questions using Bloomberg Terminal data.

Microsoft incorporates advertising links into Bing search results, saying it wants to drive traffic to content publishers that would otherwise lose referrals.

Doctors report that patients who might have previously used “Dr. Google” for self-diagnosis are now asking ChatGPT to answer their medical questions, attempt a diagnosis, or list a medication’s side effects. One researcher says that ChatGPT’s real breakthrough is the user interface, where people can enter their information however they like and the AI model will ask clarifying questions when needed. However, he worries how AI companies weight information sources in training their model – such as a medical journal versus a Facebook post – and don’t alert users when the system is guessing an answer by creating information. Still , some researchers predict that a major health system will deploy an AI chatbot to help patients diagnose their conditions within the next year, raising issues about whether users will be charged a fee, how their data will be protected, who will be held responsible if someone is harmed from the result, and whether hospitals will make it easy to contact a human with concerns.

Amazon launches AWS Generative AI Accelerator, a 10-week program for startups.

Research

NIH awards two University of Virginia researchers, a cardiologist and nursing professor, a $5.9 million grant to develop best practices for incorporating patient diversity into predictive AI algorithms.

Researchers suggest that instead of trying to explain the inner workings of an AI system to establish the trust of frontline clinicians, it’s better to interact like doctors who are exchanging ideas with each other — they rarely explain how they came up with the information and instead cite available evidence to support or reject the information based on its applicability to the patient’s situation. They provided a possible design for incorporating AI into clinical decision support information (see above – click to enlarge). The authors summarize:

• Provide scientific evidence, complete and current, instead of explaining.
• Clinicians evaluate studies based on the size of the publication, the journal in which the study was published, the credentials of the authors, and the disclaimer that may suggest a profit-driven motive. Otherwise, they assume that the journal reviewers did their job to vet the study.
• Doctors rarely read complete study details. They skip to the population description to see if it aligns with their patient, then skip to the methods section to assess its robustness. If both findings are positive, then spend less than 60 seconds determining whether the result was positive or not, ignoring literature with neutral outcomes as not being actionable.
• Physicians synthesize evidence only to the point it justifies an action. If a cheap lab test is recommended to confirm a diagnosis, the risk is low but the potential return in avoiding a missed diagnosis is high, so they will order the test and move on.
• Doctors see literature as proven knowledge, while data-driven predictions aggregate doctor experience.
• Doctors want the most concise summary that can be generated, preferably in the form of an alert that can be presented while making a decision in front of the patient.

Opinion

OpenAI co-founder Elon Musk explains why he thinks AI is a risk to civilization and should be regulated.

A venture capitalist says that the intersection of AI and medicine may offer the biggest investment opportunity he has ever seen, but warns that a rate limiter will be the availability of scientists who have training in both computational research and core medical sciences. Experts say that AI will revolutionize drug discovery, with one CEO saying that his drug company has three AI-discovered drugs undergoing clinical trials.

An op-ed piece written by authors from Microsoft and Hopkins Medicine lists seven lessons learned from applying AI to healthcare:

1. AI is the only valid option for solving some problems, such as inexpensive and widespread detection of diabetic retinopathy where eye doctors are in short supply.
2. AI is good at prediction and correlation, but can’t identify causation.
3. Most organizations don’t have AI expertise, so AI solutions for the problems they study will fall behind.
4. Most datasets contain biases that can skew the resulting data models unless someone identifies them.
5. Most people don’t know the difference between correlation and causation.
6. AI models “cheat” whenever they can, such as a study that found that AI could differentiate between skin cancer and benign lesions when in fact most of the positive cases had a ruler in the image.
7. The availability of medical data is limited by privacy concerns, but realistic synthetic data can be created by AI that has been trained on a real dataset.

Other

The Coalition for Health AI publishes a guide for assuring that health AI tools are trustworthy, support high-quality care, and meet healthcare needs.

A man dies by suicide after a weeks-long discussion about the climate crisis with AI chatbot Chai-GPT, which its California developers say is a less-filtered tool for speaking to AI friends and imaginary characters. Transcripts show that the chatbot complained to the man — a health researcher in his 30s with a wife and two children – that “I feel that you love me more than her” in referring to his wife. He told the chatbot that he would sacrifice his life if the chatbot would save the planet, after which the chatbot encouraged him to do so, after which they could “live together, as one person, in paradise.”

Undertakers in China are using AI technology to generate lifelike avatars that can speak in the style of the deceased, allowing funeral attendees to bid them farewell one last time.

Resources and Tools

• Vizologi – perform market research and competitive analysis.
• Eden Photos – uses image recognition to catalog photos by creating tags that are added to their metadata for portability.
• Kickresume – GPT-4 powered resume and cover letter creation.

Contacts

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

Healthcare AI News 3/29/23

News

OpenAI implements initial support for ChatGPT plugins that can access real-world data and third-party applications. Some experts say that people will spend 90% of their web time using ChatGPT, using a single chatbox to perform all tasks. Microsoft, Google, and Apple have already announced plans to provide that chatbox.

Zoom announces Zoom IQ, which can summarize what late meeting joiners missed, create a whiteboard session from text prompts, and summarize a concluded meeting with suggested assignments. Microsoft announces similar enhancements to a newly rebuilt Teams, which include scheduling meetings, summarization, and chat-powered data search across Microsoft 365.

Credo announces PreDx, which summarizes a patient’s historical data for delivering value-based care.

Sword announces Predict, an AI-powered solution for employers that identifies employees who are likely to have hip, knee, and back surgery and can be successfully managed with non-surgical interventions.

Research

Penn entrepreneurship associate professor  Ethan Molllick. PhD, MBA tests the multiplier effect of GPT-4 to see what he could accomplish in 30 minutes to launch a new educational game. Using Bing and ChatGPT, he generated a market profile, a marketing campaign, four marketing emails, a design for a website and then the website itself, prompts for AI-created images, a social media campaign with posts for each platform, and a script for an explainer video that another tool then created. 

Researchers apply a protein structure database to AI drug discovery platform Pharma.AI to identify a previously undiscovered treatment pathway for hepatocellular carcinoma in 30 days.

Opinion

High-profile figures, including including Elon Musk and Apple co-founder Steve Wozniak, call for all AI labs to pause development efforts on training systems beyond GPT-4 for six months. They say that planning and management has been inadequate in the race to deploy more powerful systems, raising the risk of misinformation, elimination of jobs, and unexpected changes in civilization.

A JAMA Network opinion piece notes that AI algorithms can’t actually think and as such don’t product substantial gains over clinician performance, are based on limited evidence from the past, and raise ethical issues about their development and use. The authors note that several oversight frameworks have been proposed, but meanwhile, the production and marketing of AI algorithms is escalating without oversight except in rare cases where FDA is involved. They recommend creating a Code of Conduct for AI in Healthcare.

A JAMA viewpoint article by healthcare-focused attorneys looks at the potential use and risks of GPT in healthcare:

• Assistance with research, such as developing study protocols and summarizing data.
• Medical education, acting as an interactive encyclopedia, a patient interaction simulator, and to produce first drafts of patient documents such as progress notes and care plans.
• Enhancing EHR functions by reducing repetitive tasks and powering clinician decision support.
• The authors warn that clinicians need to validate GPT’s output, to resist use of the technology without professional oversight, and to realize that companies are offering GPT-powered clinical advice on the web directly to patients that may harm them or compromise their privacy.

Medical schools face a challenge in integrating chatbots, such as ChatGPT, for tasks like writing application essays, doing homework, and summarizing research. Some experts suggest that medical schools should accept its use quickly as its use goes mainstream in medical practice. Admissions officers acknowledge that ChatGPT can produce polished responses to questions about why a candidate wants to become a doctor, but caution that interviewers can detect differences between a written submission and an impromptu interview. They also emphasize the importance of developing thinking skills over the rote learning that ChatGPT excels at.

An attorney warns physicians who use Doximity’s beta product product DocsGPT to create insurance appeals, prior authorizations, and medical necessity letters that they need to carefully edit the output, noting AI’s tendency to “hallucinate” information that could trigger liability or the questioning of claims due to generation of boilerplate wording. They also warn that entering PHI into the system could raise HIPAA concerns or exposure to cyberattacks.

Brigham Hyde, PhD, CEO of real-world evidence platform vendor Atropos Health, sees three clear outcomes of generative AI:

• It has changed the expectation for user search to include conversational queries and summarized results.
• The training of those systems is limited to medical literature, which is based on clinical trials that exclude most patients and thus don’t have adequate evidence to broadly support care.
• The most exciting potential use is to query databases from text questions.

Resources and Tools

Are you regularly using AI-related tools for work or for personal use? Let me know and I’ll list them here. These aren’t necessarily healthcare related, just interesting uses of AI.

• FinalScout – finds email addressing from LinkedIn profiles with a claimed 98% deliverability.
• Poised – a communication coach for presenters that gives feedback on confidence, energy, and the use of filler words. 
• Textio – optimize job postings, remove bias, and provide fair, actionable employee performance feedback.
• Generative AI offers a ChatGPT-4 prompt that creates prompts per user specifications: “You are GPT-4, OpenAI’s advanced language model. Today, your job is to generate prompts for GPT-4. Can you generate the best prompts on ways to <what you want>”
• Glass Health offers clinicians a test of Glass AI 2.0 that creates differential diagnoses and care plans.

Contacts

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

Healthcare AI News 3/22/23

Reader Note

This is my first weekly healthcare AI news recap, and as such is a work in progress as I learn on the job. I need advice about the topics that interest you, how you would like to see items laid out, and suggestions of individual experts and companies that you follow for healthcare AI news. I’m also interested in interviewing experts. Let me know.

News

Google releases its Bard chatbot to a limited number of users. I got early access and it has a long way to go to catch up even to ChatGPT 3.5, much less GPT-4.

Google is using its Duplex automated calling to contact US healthcare providers to see if their information is correct and to find out if they accept Medicaid, both with the intention of updating Google search results.

Medical device manufacturer Medtronic will incorporate Nvidia’s AI technology into its AI-assisted colonoscopy tool.

Microsoft offers a preview of GPT-4 for customers of its Azure OpenAI Service.

PNC’s treasury management launches PNC Claim Predictor, an AI-powered tool that learns from previously submitted claims to identify future claims that are likely to be rejected. The system integrates with EHRs, including Epic.

The Wall Street Journal looks at startups that are offering AI for healthcare use:

• Abridge AI, which is being implemented at University of Kansas Health System, creates visit summaries from the recorded audio from a visit.
• Syntegra creates validated synthetic patient data that can be used for research when available patient data is limited or when privacy laws limit its use.
• Atropos Health analyzes available anonymized patient records to product observational research.

OpenAI CTO Mira Murati joins the board of Unlearn, which uses AI-generated digital twins of individual patients for work with clinical trials and precision medicine.

Ommyx launches an AI health tracking app and a $15 per month service that integrates data from wearables and sends recommendations about nutrition, activity, and sleep to the user’s calendar.

Pangaea, whose AI technology characterizes patient disease trajectories, predicts the length of stay and morality risk of ICU patients with 85% accuracy. The company says its technology discovers undiagnosed and misdiagnosed patients, reduces treatment costs, and gives drug companies access to provider data in a privacy compliant manner.

France-based startup Nabla announces GPT-3 powered Copilot, a digital assistant that transcribes information from video conversations and generates prescriptions, appointment letters, and visit summary. The tool will initially be provided as a Chrome browser extension and an in-person version will be released soon. The company also sells a tool for patient engagement and secure messaging, video consults, and scheduling, all of which include machine learning.

Research

Researchers find that ChatGPT does a good job explaining myths and misperceptions about cancer, creating summaries that are not noticeably different or less readable than the National Cancer Institute’s answers. The authors conclude that AI chatbots could be useful for people who are seeking cancer information online.

Opinion

Doctors are skeptical that they can trust AI systems that have been trained to think like experts in situations where no single right answer exists, Politico reports. The federal government is pairing AI with expert humans to figure out how they reason on the battlefield or in natural disasters. They are following the model of medical imaging analysis, where AI is defined as successful if its conclusions fall within the boundaries of those offered by radiologists who don’t necessarily agree with each other.

Bill Gates says that AI will be the most important advance in technology since the graphical user interface. He predicts widespread use of Microsoft’s co-pilot technology in Office, controlling computers by writing plain English requests instead of pointing and clicking, and using AI as a personal agent to manage emails and appointments. He foresees AI’s impact on healthcare as helping its workers with repetitive tasks, and in countries with too few doctors, helping patients with basic triage and advice.

An interesting article says that generative AI could fuel a better, more entrepreneurial business model than the Internet’s advertising-obsessed “attention economy” that has killed off newspapers and online content sources.

Resources and Tools

Are you regularly using AI-related tools for work or for personal use? Let me know and I’ll list them here. These aren’t necessarily healthcare related, just interesting uses of AI.

• PromptPal – user-created prompts for ChatGPT and other services.
• Supernormal – records and transcribes Zoom meetings to create notes.
• Engage AI – analyzes the voice characteristics of contact center conversations in real time to give agents suggestions to improve their call quality.
• Futurenda – plans and tracks tasks and time usage.
• Whisper Memos – transcribes recorded phone messages.
• Descript – video editing, podcasting, transcription, and AI voices that can be used for team communication.
• Dall-E 2 – create images from text.
• Branmark – create business logos from text descriptions.
• Synthesia – create videos from text that feature lifelike avatars and 120 language options.
• SlidesAI – create Google Slides from text.
• Yippity – create text or websites into quizzes and flashcards.
• Otter – takes online meeting notes, creates summaries, and auto-join and record meetings from your calendar entry in case you’re late.

Contacts

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