Inspiration
According to the AMA, 62.8% of the physician population experienced some type of burnout in 2021. This was consistent no matter the specialty. A significant portion of a physician's role is the necessary paperwork and documentation that has to be done. If we can provide an automated solution to this issue, it would be a stepping stone to a larger product. On top of documentation, physicians also have to go through personal research to understand each patient's condition and history, and make a plan of action.
What it does
The Scribe Health web application allows physicians to record patient conversations in real-time and summarize them into significant medical key points. This text can be integrated into any patient record within the EHR system. AI scrapes the conversation to provide a recommended plan of action for each patient for physicians to consult and maybe proceed with.
How we built it
I built this web application using NextJS, OpenAI for calls, Speechly for voice-to-text, and Google Cloud Healthcare Natural Language Processing for medical terminology recognition. No data is being persisted currently.
Challenges we ran into
My main challenge was researching and collecting the necessary data to train the program. I had to obtain datasets from credible healthcare websites to feed into technology like Speechly, OpenAI, and GCloud.
Accomplishments that we're proud of
I'm proud of the MVP I have created because it has real-time functionality that can be used today. Brainstorming, researching, implementing and testing by myself was the most challenging portion of this Hackathon but also the most rewarding experience.
What we learned
There are lots of regulations when it comes to AI and ML in the healthcare sphere. For example, the FDA has created a proposed regulatory framework for any AI and ML used within healthcare electronic products. And these have to go through "good practices" that comply with the framework.
What's next for Scribe Health
Scribe Health will be continuous innovation through this product by completing integration with EHR systems and having AI create the recommended plan of action for physicians to use.
Log in or sign up for Devpost to join the conversation.