HackStack

Project Info

MedEasy

Devpost

Inspiration

One of the key problems with today's healthcare systems is the lack of patient agency. Patients rarely access their medical health records: medical health records can be cumbersome to obtain and difficult to understand. As a result, MedEasy was developed with the objective of assisting individuals who have trouble viewing and understanding their personal medical records.

What it does

MedEasy offers a simple platform for patients to access information about their medical health records and their current health condition. Patients can log in to the platform, where they are presented with a summary of important details like relevant care plans, current conditions, and vitals. Patients can then understand more about their condition through the chat interface on the platform; they can ask follow up questions regarding their health and a chatbot will explain the answers to them in an easily understable way.

How we built it

We used Synthea, a Synthetic Patient Generation tool to generate realistic electronic health records in the form of CSV files. We then processed these CSV files to glean relevant information, and fed a textual representation of the electronic health record to a Code Llama LLM using together.ai's API, prompting it to simplify the input so it can be better understood. We also used together.ai's API and the Code Llama LLM to create a chatbot-like interface on the website, where patients can ask follow-up questions and the LLM will answer them to the best of its abilities. We created a front-end website using React.JS and connected our code to the website using the Python Flask module.

Challenges we ran into

We had difficulties finding a reliable free database solution for our product, and ended up using a local postgresql database for EMR tabular data. Additionally, we struggled with implementing the website since most of our team had very little experience with front-end coding. Lastly, we had many concerns related to resolving dependencies and using virtual environments, as our front-end platform was very sensitive to them.

Accomplishments we're proud of

We are proud to announce that we brainstormed an idea given real issues in the world, and worked hard at it, which resulted in us creating a possible solution to these issues.

What we learned

Through the long process of brainstorming, developing and testing MedEasy, we learned invaluable time-management, collaboration and technical skills. In addition, we learned that given an idea, it is possible for us to create that idea given enough effort.

What's next

Given how rapidly artificial intelligence is developing, MedEasy is only the beginning of future developments. With enough time, AI technology will be able to process tremendous amounts of information that no human could in a lifetime, which will result in scientific breakthroughs that benefit society as a whole. MedEasy has the potential to empower patients by giving them the ability to understand more about their health and have better access to electronic health records. During TreeHacks, one of the ideas we experimented with was using voice-to-text software to allow people to "talk" to LLMs. This idea has the potential to significantly increase the population that MedEasy can serve, such as the elderly and people with debilitating health conditions.

Analysis

Compare with all teams

View

Metric

Figures cover GitHub contributors during the hackathon window. A co-authored commit counts in full for each author, so per-member totals add up to more than the whole-team figures.

Technology

Found in codeClaimed only
  • CSSIn code
  • HTMLIn code
  • JavaScriptIn code
  • PythonIn code
  • ReactIn code
  • SQLIn code
  • FlaskClaimed
  • OpenAIClaimed
  • PostgreSQLClaimed

6 of 9 appear in the indexed code. 3 claimed on Devpost could not be matched to code, which may simply mean the tool leaves no trace in the repository.

AI coding agents

No AI coding agent signals were found in this repository.

Detected from committed agent config files and commit authorship. Absence of a signal is not proof an agent was unused.

Codebase size

Source size

45 KB

Source files

19

Counts recognized source files only; vendored directories, binaries and lockfiles are excluded, so this is smaller than the repository on disk.

0 stars