Unleashing Efficiency, Unlocking Empathy
How and why AI scribes can shape the landscape of veterinary care for the better.
I accept as a lifelong obligation the continual improvement of my professional knowledge and competence.
Veterinarian’s Oath, final line
“Writing is eternal.”
-F. Scott Fitzgerald
I don’t know any veterinarians who became veterinarians to pursue a lifelong dream of writing medical records. None of us, whether we dreamed of being a vet as children or pursued it first as adults, did so to fulfill our unending passion for filling out charts. I’m as avid a reader as I am dedicated a veterinarian, but even I was skeptical about the validity of a paper that explored medical records as a genre.
Nonetheless, it’s a vital part of our work. Good record-keeping enhances better patient care and continuity, it provides legal documentation of our work, it facilitates research and quality improvement, and even works as a communication tool to clients. But following the standard SOAP format doesn’t mean the note will be any good. And evaluation of records is deeply subjective.1 Doctors’ language can be stigmatizing, painful for clients or patients to read, and sometimes indecipherable even by other doctors. Anecdotally, I had a previous employer who intentionally wrote with abominable penmanship in order to protect herself against future scrutiny.
I use a combination of AI scribe and human scribe. Most of us are familiar with what a human scribe does, so I’ll briefly describe the AI version. An AI medical scribe is a software tool designed to streamline clinical documentation tasks for healthcare providers. Using technologies like speech recognition and natural language processing, AI scribes transcribe spoken words into text in real-time during patient and client encounters. These tools are trained to understand medical terminology and contextual nuances, ensuring accurate documentation. Integrated with electronic health record (EHR) systems, AI scribes automatically populate transcribed information into patient records. Some systems also offer real-time feedback to enhance accuracy. Ultimately, AI medical scribes are intended to improve the efficiency and quality of clinical documentation, allowing healthcare providers to focus more on patient care.
Benefits
How much do medical professionals enjoy medical records? Oh, I’m so glad I asked…
Completing medical records and billing records is so necessary and time-consuming that it’s often completed after-hours during what some physicians call “pajama time” and contributes significantly to physician burnout.
The federal government subsidizes electronic medical record adoption, but physicians are concerned about poor EHR usability, time-consuming data entry, reduced patient care time, inability to exchange health information, and templated notes.
80% of medical progress notes are copied and pasted, propagating errors.
Completing EHR accounted for almost 17 minutes per patient encounter.
When was the last time your physician spent 17 minutes talking to you?
Completing EHR accounted for at least 17% of nurses workload.
Relieving the burden of medical records dramatically reduces burnout in physicians.
Using an AI scribe to enhance record-keeping resulted in:
81% of patients reported the doctor spending less time looking at a screen during the visit.
Reduced time spent on EMR
Physician-reported “transformational impact” on their care.
There are a number of medical scribes available on the market. Personally, I’ve been impressed by a few. I use Digitail for my hosital’s PIMS, and it counts an AI scribe among its features. And I use a human scribe to help organize and double check the notes against the recordings before I review them. I strongly recommend using both, I use Mascotte for the human support. Several months I managed to cobble my own AI-scribe together by duct-taping23 various softwares together. It’s truly that accessible.
The scribes are good, and and save a lot of time. About a third of physicians offered it in a study above tried one, and the results were overwhelmingly positive on doctors and patients alike.
But I think it’s more than time they save. I believe that the cost of paying attention to medical records instead of to the patient and to the client is more significant than we acknowledge. The doing of good work requires time and focus, and the more time and focus we spend on the actual patient than on their records, the better a job we can do for them. I don’t dispute that medical records are a vital part of patient care, but I don’t believe they should interrupt or interfere with our physical examination and evaluation as often as they do.
Put simply, if I’m thinking about how to write the patient’s medical record, I’m not really thinking about the patient. Those are different parts of my brain at work and none of us possess an unlimited amount of bandwidth. I’d much prefer my bandwidth to be dedicated to patient care than to what is literally my least favorite part of my job. I think AI scribes will let us do both things better. A study in human healthcare providers found that every extra minute spent with a patient reduces the likelihood of hospital readmission by 8%. Imagine what that means for our patients when we use an AI scribe to allow our focus to be on the patient.
Risks
There are concerns that the use of artificial intelligence will dehumanize us, and that is a risk. Those concerned range from union bosses worried about their workers to statisticians writing editorials about how AI girlfriends are ruining a generation of men. And maybe they’re right, maybe the end of the world is nigh. Although scientific research refutes such opinions, it’s hard not to recall mythology and feel something at least near worry when one thinks of how Amazon named their now nearly ubiquitous AI speaker. Every Narcissus needs an Echo.
But to do almost anything, and certainly almost anything worthwhile, is not without inherent risks. I’ll share a story of a patient I saw some time ago. I was working as a locum surgeon at another hospital in the area and I was asked to evaluate a patient with a pyometra. A pyometra is an infected uterus, and this patient had been on several rounds of antibiotics to resolve it to no avail. The preferred treatment is an ovarihysterectomy surgery to remove the uterus entirely, but the owners’ financial constraints led them to try medical management. Now that medical management failed, surgery was the only option left for recovery.
A colleague noted the patient’s severely elevated kidney values, a common problem in these cases,4 and openly expressed her worry about the risk of anesthesia in a patient whose kidneys were significantly injured.5 Her concern was entirely valid and her reasoning was medically sound, but the patient’s pathology persisted. If we did surgery, she might die. If we didn’t, she would die. I did the surgery,6 and we were fortunate that the patient survived and lives on today. It was only by acceptance of the unavoidable risk were we able to get better.
Risk of error, fault, or outright failure cannot be entirely avoided in the practice and performance of medicine especially as we continue to improve.
Conclusion
Performing surgery on a sick patient - or any patient - was not without risk, but without the acceptance and management of that risk, there was only a vanishing hope of improvement. So too it is with advancing technology. Just as the patient would not get better without the risk of anesthesia and surgery, we doctors will not get better without the risk that comes with the use of using new technology. The use of artificial intelligence has the potential to harm us in the way that a scalpel has the potential to harm a patient. With thoughtful, well-informed, well-intentioned use, we can minimize the negative impact of these necessary advances.
And I do mean necessary. I opened this piece with the final line of the Veterinarian’s Oath, which is a mandate to our continual improvement. That pursuit of improvement cannot be accomplished by only acting on what is established, conventional, or certain. We must be bold in our pursuits of efficiency and efficacy as well as in our imaginations and aspirations. We cannot improve what is without dreaming of what can be. Our Oath’s mandate is at once utilitarian in its goal and existentialist in its pursuit. “What is happiness?” asked a noted philosopher. “The feeling that power is growing, that resistance is overcome,” was his answer.
The relatively simple technology of transcription by artificial intelligence has the potential to dramatically improve not only a veterinarian’s quality of life, but also the quality of their care. That’s reason enough to try it for now and develop it for future.
In an effort to prove a point as a self-righteous veterinary student, I submitted a surgery report identical to a classmate whom I believed was comparatively favored by the professor. She received an A. I received a C+. Providing this as evidence of unfairness was not well-received. In hindsight, the outcome was entirely predictable.
A phrase I use to mean “cobbling together in an unaesthetic fashion.” Fully focused on function rather than form. A bit like how Mary Shelley described the reaction of her modern Prometheus to his own creation (see below). I’m no coder, but the software is so accessible that by being a little clever you can get away with not being very smart. The ones produced by the professionals work dramatically better than mine.
“For this I had deprived myself of rest and health. I had desired it with an ardour that far exceeded moderation; but now that I had finished, the beauty of the dream vanished, and breathless horror and disgust filled my heart. Unable to endure the aspect of the being I had created, I rushed out of the room, and continued a long time traversing my bed-chamber, unable to compose my mind to sleep.”
Shelley, Mary. Frankenstein (AmazonClassics Edition) (pp. 51-52). Amazon Classics.
Pyometrae are frequently the result of E. coli infections. E. coli produces an endotoxin, a lipopolysaccharide, that interferes with, among other things, anti-diarrhetic hormone and kidney function. The subsequent dehydration can cause a rise in kidney values and sometimes severe damage to the kidneys. This information isn’t pertinent to the essay, but it was a question I got right in veterinary school to the extreme (almost impolite) surprise of my professor and classmates and I enjoy the memory.
The conversation went something along the lines of “You’re not going to put this patient under anesthesia, are you?” To which I replied approximately, “Unless you think we can do it with her awake? Do you want to hold her? I’ll be fast!” I suppose John Knowles may have been onto something when he called sarcasm the protest of the weak, but I thought I was being funny. My colleague was not remotely amused. Again hindsight reveals that to have been a predictable outcome.
With the patient under anesthesia, of course.
Great article, Bill. I think a big challenge here is switching costs when it comes to adapting new software. Software must be 1. Cheap 2. Simple to use 3. Simple to integrate. The job of an entrepreneur is to change the mind of the customer. It's a tall order!