QuickDeets

QuickDeets Home Page

Inspiration

Three of the people on the team have family members in healthcare services. After talking with their family members, we found many ways in which we could improve patient care and automate some of their tasks.

What it does

It is a website that is capable of adding new patients or importing patient data from an EHR system. It then takes this data and populates a table of patient harm indicators. A healthcare professional can then record their own notes and optionally add an LLM's analysis of their notes to the table. All appended notes are timestamped.

How we built it

In Python, there is a Audio Recording class that handles the function of starting and stopping audio recordings. In JavaScript, requests are made to Python using the API created in Django. These requests will start and stop the recordings. Once a recording is stopped, it is passed into a ChatGPT API prompt to return a JSON file through the API and back to the requester to update the website. A user can decide to update the website with the suggested updates or they can edit the suggestions and save those changes.

Using an FHIR compatible server and Synthea, we created mock patient data in the format that you would see at a hospital. This data was then used in the website and each patient was assigned an ID. With this ID, you can pull up the data related to that patient and you will see the patient harm indicator data for them as well.

Challenges we ran into

There are a lot of moving parts and there were conflicts at certain points during development. We also initially struggled to find open source data that is in a format that hospital systems use.

Accomplishments that we're proud of

We were able to resolve the conflicts between the separate components to create a working project.

What we learned

We learned that communication is paramount when working across different languages and frameworks. We had some communication related issues that set back our progress that could have easily been avoided.

What's next for QuickDeets

We plan on scaling the project and creating more security features for deployment in real-life settings. There is also room for a lot more AI and machine learning components to personalize the project's use for any patient's condition.