Re: "People preferred the [patient] portal over the telephone for getting test results, updating personal information, getting medical records copies,…
Curbside Consult with Dr. Jayne 10/21/19
The past week has been entirely too crazy, working on projects for a couple of vendor clients whose offerings are starting to overlap with each other. Everyone is chasing after certain industry buzzwords, and if you can add a feature or two that gets you into a different space where people might be more willing to spend money than your current prospective clients, apparently you go for it.
It’s been challenging for me as it’s hard to keep track of what’s a real feature that is able to be delivered vs. something that is in the works, but only parts of it are actually in the code, especially in an agile development organization. It’s particularly challenging when company leaders talk about features under development as if they are actually part of a general release.
In a startup organization, words are thrown around much more freely than they typically are in publicly traded companies, who sometimes mind their language a bit to stay on the right side of the shareholders. With that in mind, I was excited when a reader clued me in to a recent analysis looking at corporations’ earnings call transcripts in an attempt to determine whether the words used in the calls are harbingers of credit risk. Researchers at Washington University in St. Louis looked at over 132,000 earnings call transcripts and used machine learning methods to create a measure of credit risk. The resulting information informs an algorithm that assesses elements from changes in credit ratings to risk of bankruptcy.
What if you could extrapolate those findings to examine the earnings call transcripts of EHR vendors to create a model that would let you know how much of what they were saying was possibly accurate vs. what might be considered creative accounting? What if you could use proxy words to identify impending layoffs or bad decisions that were about to make your stock drop? Mr. H follows some of the earnings calls much more closely than I do, but I have definitely heard some interesting comments on calls in the past that could be interpreted in a variety of ways:
“Our cash generation capabilities continue to expand our storehouse of dry powder.” I’m not sure using gunpowder as a comparative term is a great idea in this day and age. Does the company want to seem like it’s war-like and on the offense? Or perhaps it’s a commentary on the executive team’s ability to blow up the company by continued poor decision-making?
“Scary.” I recently read an earnings call transcript that used the word three times in a variety of contexts. Could there be something predictive about the state of mind of the people on the call?
“Our team is always impatient to go faster.” That’s the kind of reassuring language end users love hearing from vendors with a track record of under-delivering or overpromising on features. Not to mention that when you’re caring for people’s loved ones, you typically don’t want your primary tools to be fueled by a spirit of impatience.
Those are immediately attention-getting, but I’d be more interested in the subtle comments that show that something is slightly off or that there might be some level of obfuscation going on. One recent call characterized what clearly looked like a cost-cutting layoff as an effort to speed compliance with new regulatory requirements. With a training database of thousands of vendor earnings call transcripts, I bet you could come up with some very interesting themes and potentially useful indicators.
In the next year or so, I have to start thinking seriously about a performance improvement project so that I can complete it and check the box for renewal of my clinical informatics board certification. Maybe I could build a tool that would better enable the HIStalk team to detect language that would be most predictive of a company that might have juicy things for us to write about, or what their odds might be of winding up on the HISsies ballot. I doubt those applications would be approved by the folks at the American Medical Informatics Association, but they would be entertaining.
If I don’t come up with a project soon, I’m going to have to seriously think about letting my certification lapse. The required projects are actually referred to as “Improvement in Medical Practice” projects, and since I don’t actually practice clinical informatics where I practice clinically, that gets a little dicey. My primary clinical employer doesn’t want the physicians to have anything to do with the EHR – there isn’t even an informatics committee. The COO (who is a practicing physician) calls all the shots on whether we’ll implement new features and how they will be shared with the masses. The likelihood of my being granted any ability to query the data or perform any kind of project is exactly zero.
They do allow a diplomate to substitute a 360-degree evaluation project instead, where they survey a half dozen of their colleagues to find an area that needs improvement, then work on it and survey again. That doesn’t exactly work in the consulting model, where I think my clients might be generally appalled if I asked them to spend resources essentially providing job coaching to someone they’re paying as their expert advisor.
Doing these projects as a way to maintain certification is frustrating regardless of your specialty. During a time when I wasn’t practicing clinically, I had to do a mock “hand hygiene” project where I had to manually enter a downloaded data set and then analyze it. The goal was to simulate the paper surveys that my peers were getting from live patients, but I learned exactly zero sitting there and keying in the data. It’s just another hoop that physicians have to jump through to try to stay certified.
That takes me back to the earnings call transcript project. Maybe if I write it as an abstract with enough sexy buzzwords I can sneak it past the evaluators. Sprinkle in some artificial intelligence, machine learning, and blockchain to get the job done.
If any of you other clinical informaticists out there have creative ideas for what a consulting clinical informaticist can do as a project, I owe you a drink at HIMSS.
For the rest of you: what’s the wackiest thing you’ve ever heard in an earnings call transcript? Leave a comment or email me.
Email Dr. Jayne.
Discerning the financial health of an IT vendor in healthcare doesn’t require AI nor Machine Learning algorithms. Basic common sense is all that’s needed. The more criptic the language, the more creative language used by executives to describe basic economics and forecasts the more you know you’re listening to BS.
It’s all in the numbers.
I like your idea of the BS barometer. Perhaps one of the investment firms would sponsor it.
Jayne, Dr, 2020. Journal of ToBeDecided. Predicting EHR vendor system enhancements, performance, and corporate survival; an innovation application of NLP & deep learning technologies.
I am 99.9% confident that many research efforts with less useful premises have been published.