search

william-hu-codes / electronic-medical-record

Fully functional electronic medical record with user authentication using Google OAuth. Integrates OpenAI to automatically generate discharge summaries by parsing through the patient's existing progress notes.
MIT License
0 stars 3 forks source link

Team Reflections #51

Open william-hu-codes opened 10 months ago

william-hu-codes commented 10 months ago

What went well?

  1. Problem solving - we faced a vast array of blockers while building our application. Throughout the process, we would always attempt to resolve problems first individually, then as a team, and then externally using online resources or from the very knowledgable and competent developers in our class.
  2. Communication - our team fostered a very collaborative and harmonious work culture even when we divided and conquered different tasks. For the past few days, we all had an unspoken pact that we would remain "on-call" and reachable throughout the day and even towards the late evenings.
  3. Real life applicability - we managed to implement a real patient centric digital solution to one of the biggest problems faced by hospitals -- discharge backlogs. Patient discharge summaries can take hours, if not over several days, to write up (depending on the patient's length of stay in the hospital and complexity of stay). This often leads to a delay in the discharge process, as providers are often too busy to draft these summaries. We demonstrated that EMRs could implement an AI approach to tackle this problem.

What we want to reproduce in the future?

  1. Ownership and pride in the application - we all felt very driven to continue to develop the app after hours out of pure excitements in creating something special.

What did not go so well?

  1. Sometimes need to take a step back and come up with a methodical approach to debugging - while troubleshooting one of our blockers, we spent a good chunk of the day analyzing and trying to identify the issue in our controller action. We eventually simply commented out that entire function and rewriting it from scratch which really took 2 minutes at most. I think sometimes we need to just reconsider our adopted approach in troubleshooting and recognizing when it's time to approach the problem in a different manner.
  2. Tenacity vs stubbornness - There is a fine line between being tenacious and being stubborn, and I think we walked on that line like a tightrope at certain points. We would sometimes get lost in our debugging process, trying to change many many many things before resorting to a simple google search, which often leads to a quick solution offered by StackOverflow

How will we prevent this from occurring in the future

In the future, while debugging, we will remind ourselves to reflect on our current approach to troubleshooting and ask ourselves if this is the most methodical approach or if there is a better approach we can take.

krabecb commented 10 months ago

Wavelength - (Rob, Will, Reggie) Heroku - https://wavelength-emr-ec309aa814db.herokuapp.com/ PROJECT PASSED Instructional Glow: Super professional layout and styling! Meaningful CRUD application, OpenAI - amazing! Deployment link should be at the very top of README/repo. Functionality. Delete functionality is clever! Instructional Grow: Solid/gradient background - current background can clash with text. User notifications and user-centric functionality! Balanced commits. Start looking at health dashboards for future inspiration/dashboard/EMR dashboard enhancement. ID management for users (super long IDs).