Medify

Inspiration

We joined HackMIT interested in trying the health and accessibility track. One of our members lives with her grandparents at home and recognized the challenges that older people face in recognizing and communicating their health concerns. The elderly are more susceptible to getting illnesses and at their age, it is important for them to quickly identify possible conditions that they have based on their symptoms. This way, it will be easier for doctors to accurately and quickly understand the situation and formulate a treatment plan.

Goal

We wanted to create a hack that would make it easier for the elderly to understand and communicate any health conditions they may have. By simply entering fragments of ideas or a list of symptoms, the hack should provide seniors with a coherent, professional-sounding description of possible medical conditions that they could bring to a physician.

Tools

We built our project using OpenAI's GPT-3 API. We were fortunate enough to meet a mentor that had been working on a similar idea. He helped us in editing the API to customize the chatbot towards our medical focus. Afterwards, we prompt-engineered the bot to be an empathetic doctor named Sue who listens to suggested possible ailments, specifically ones common among the elderly, based on provided symptoms. Then, we modified the chatbot’s user interface by editing the HTML and CSS code in the API.

Challenges

We faced several challenges as beginners to hackathons. Despite having basic and general coding experience, we did not have any prior experience with big coding projects and using APIs. We were also not familiar with typescript and javascript so we worked a lot with mentors to modify the chatbot code.