Harmonycare

Presentation : link

Inspiration

Managing diabetes can feel like a constant battle. Many face daily struggles in balancing glucose levels, calculating insulin doses, and planning meals, all while managing the emotional stress that comes with this condition. HarminiCare was inspired by the need for a holistic solution that makes managing diabetes simpler, more personalized, and less time-consuming. We aimed to create a supportive tool that feels like having a dedicated health assistant at your side, 24/7.

What it does

HarminiCare connects with Continuous Glucose Monitors (CGMs) to monitor glucose levels in real time. The platform offers tailored insulin recommendations, meal planning assistance, and a responsive AI chatbot to answer questions, provide health insights, and give personalized advice. From insulin dosage support to tracking calorie intake, HarminiCare combines all essential tools in one easy-to-use app.

How we built it

We designed HarminiCare using:

Streamlit to build a user-friendly web interface. Langflow for architecture and integration of the various components. AI/ML APIs such as: GPT-4 for text-based assistance and image understanding. ADA-002 for embedding and image processing functionalities. We also developed custom components within Langflow to accommodate these APIs, allowing for seamless integration with CGM data and personalized user support.

Challenges we ran into

One of the main challenges was ensuring smooth integration with CGMs and adapting to the wide variety of available devices. Additionally, providing accurate, real-time insulin recommendations required extensive testing and calibration. Creating a reliable AI chatbot that could offer personalized support based on individual health data also presented technical hurdles.

Accomplishments that we're proud of

We’re proud to have built a platform that combines glucose monitoring, insulin recommendations, and meal planning in a single, cohesive experience. HarminiCare’s AI-driven insights make it a valuable support tool for people with diabetes. Also, achieving partnerships with CGM providers and finalizing the data-driven accuracy of our insulin recommendation feature were significant milestones.

What we learned

Building HarminiCare taught us the importance of user-centered design, especially for health-related applications. We learned a lot about integrating real-time data sources and handling personal health data responsibly. Additionally, developing an AI-powered chatbot taught us to prioritize empathy and clarity in health advice.

What's next for Harmonycare

Our next steps include expanding HarminiCare’s support for other health metrics, like blood pressure and activity levels, to provide more comprehensive health tracking. We’re also working on mobile app development to make the platform more accessible. In the future, we hope to collaborate with more healthcare providers to broaden our reach and potentially add insurance support to make diabetes management more affordable.