What is it about the documentation of patient encounters that is so aggravating? I was ecstatic to be part of the generation that entered the computer age in medicine. The first day I opened a chart on a pull-down menu within a pixelated screen I felt like I was ushering in the future. We would be able to mine charts for clinical gold!
As a clinician, I value the importance of documentation for data collection. It’s essential to record the subjective history and objective findings for future reference as the patient progresses. It is also essential when a diagnosis is not immediately forthcoming and when care must be transferred. Multiple minds need access to patient information to make the best decisions. Data can be pooled and improve care for populations, not just individuals. This wondrous computerized record would make documentation an easy process to promote better patient care.
Fast forward a few decades, and easy is not the word I would use. When will that promise of automation, the type of automation that actually allows me to concentrate on the patient, happen? Every day, I procrastinate and bemoan the task of diagnosis documentation and coding. In the end, I often make it harder on myself than it needs to be as the charts collect in my in-basket, waiting for me. It feels like so much of the documentation is not geared toward painting a picture of the patient at hand but toward creating a charge list for the insurance company. It is easy to forget that the endless strings of number-and-letter codes I am forever looking up correspond to data that helps shape healthcare policy, fuel research, and secure funding that ultimately benefits my patients.
I have tried many ways to make the process more palatable. One way: Do all the narrative documentation and leave the diagnosis documentation and coding to complete as a batch at the end. Another way: play a game by setting a reward for completing all the narrative, diagnosis, and coding documentation all at once, before seeing the next patient. Maybe I’ll just bang my head against a wall instead. I just want to have my tools work for me instead of me working for my tools. I am constantly on the lookout for ways to decrease typing and clicks and make these wondrous technologies the assets they should be.
As part of that quest, I recently had the opportunity to trial an AI-powered bot that read my notes in real-time to help me document applicable diagnoses and codes.
Normally, when I complete my note, the next several minutes are filled with manually searching for diagnoses using a clunky search bar in a slow EHR, gritting my teeth, swearing, eventually finding gold, or settling for “close enough.” In contrast, a list of diagnosis suggestions simply appeared with my AI bot. All I had to do was select the suggestions I found most relevant. While the AI was a bit noisy at times, it was seamless, non-intrusive, and best of all, the more I worked with it, the smarter it would get over time. It helped me with most of the diagnoses I needed right out of the box, saving me precious minutes per patient.
As I think about the experience, working with a clinical AI bot was like having a conversation with someone about the beautiful bouquet you’re designing. While you are talking, they order and deliver all the flowers, vases, and several ribbons to choose from to make it just perfect, all without even being asked. Except, in this case, I got a diagnosis documentation package already translated into ICD-10 elements delivered to my laptop.
After working with this AI bot, my hope is renewed that the automated future I have envisioned will arrive sooner than later.
Tamara E. Jackson is a physician assistant.
Image credit: Shutterstock.com
