EverCare

Inspiration

Having witnessed firsthand the challenges that one of our teammates' grandparents faces living alone, we became acutely aware of the difficulties involved in maintaining their health and well-being. Elders like them struggle with remembering daily health updates, deciding when to seek medical attention, and navigating modern technology. This experience deeply inspired us to create Evercare—an app designed not only to simplify health tracking and make it accessible for elders with speech-to-text capabilities but also to provide peace of mind for both seniors and their families. By offering an AI-powered 100% speech-to-text system, Evercare allows seniors to effortlessly record their health updates, which can be easily shared with doctors. This data exchange helps medical professionals assess conditions more accurately and efficiently, improving care and ensuring that both seniors and their families feel more connected and supported in managing health. Our goal is to make it easier for seniors to manage their health, reduce stress, and foster a sense of security in their daily lives.

What it does

Users can describe their symptoms to a conversational AI agent, which then processes and delivers personalized responses based on a user’s background information and previous entries. These responses help the user better understand patterns in their health and flag issues that may require a visit to a healthcare provider. The responses are converted to speech using text-to-speech software in the backend, allowing users to hear them aloud. Users can access their previous conversations with the AI agent, allowing them to review past discussions and track their symptoms over time.

How we built it

Perplexity Sonar for generating personalized advice and flagging issues that may warrant medical attention.
OpenAI API for speech to text using whisper
ElevenLabs API for natural-sounding text-to-speech
React js frontend, Python backend using Flask

Challenges we ran into

In our original plan we intended to use the FARM approach. However, we ran into issues with integration with our current implementation
We wanted to use Terra API to collect health vitals to use in conjunction with the AI health agent, but realized we needed to focus more energy to perfect fewer tasks, including improving the chatbot and UX.

What we learned

Prompt engineering
Working with generative ai
POST/GET APIs
Webhooks

What's next for EverCare

Live health data from wearable devices (apple watches, oura rings, …)
Keeping track of medications and possible side effects / interactions
Integrate with the sources provided in Perplexity responses to give users external resources that may be relevant to their health concerns.