Inspiration
The Inspiration of my project was actually my Dad. He is a Physician's Assistant with a small clinic startup in Miami. He mentioned to me that he gets WAY too many emails so I figured a tool that could supported automated patient correspondence would help him out. Plus patient correspondence seemed like a rich source of potential unstructured text for AWS Comprehend Medical to parse.
What it does
The ecosystem opens up all sorts of potential but the primary use case (as shown on the video) is an patient self-service process to inquire about new medications via a custom form. The new medication information is then passed to AWS comprehend medical to generate an rxnorm code. The rxnorm code is then combined with the AWS HealthLake data to get a list of currently active medications. The full list of new and current medications is then feed into a drug negative interactions API. The results of that API are passed into AWS SES to generate an email to the healthcare provider highlighting the potential negative interactions of the current and new medications.
How we built it
I literally had to stitch together several github projects and tweak them to get the end-to-end deliverable up and running.
Challenges we ran into
Lambda functions were a pain to figure out for the first time. Also the chat bot with AWS Connect and AWS Lex took some time to adjust / debug.
Accomplishments that we're proud of
The ecosystem. This tool has a AWS Connect chat bot that can facilitate interactions with the AWS HealthLake environment I stood up and populated with Synthea data. The project as a whole is an easy tie into existing FHIR server instructures so it has tangible real world potential. I could easily see my Dad utilizing this tool for his practice.
What we learned
A ton about AWS services and intergrating those services for a project deliverable.
What's next for PatientCom
The next steps are to align with actual end-to-end HIPPA compilance then to explore healthcare organizations willing to incorporate this ecosystem for their patient correspondence but also form data processing processes. PatientCom can also be refined to incorporate document generation automation as well.
Log in or sign up for Devpost to join the conversation.