HealthMentor AI

AI-powered health advisor analyzes medical records instantly, delivering precise clinical insights and actionable recommendations to improve patient care and decision-making.

Comment

Inspiration Healthcare in Nigeria and across Africa faces significant challenges—limited access to specialized medical expertise, overburdened clinics, fragmented patient records, and diagnostic delays that can cost lives. Many regions suffer from a shortage of trained clinicians and infrastructure, making accurate diagnosis and timely treatment difficult. This inspired us to create a solution that leverages cutting-edge AI to bridge these gaps, democratizing access to expert medical insights regardless of location or resources. We envision a future where technology empowers frontline healthcare workers and patients alike, transforming the quality of care across the continent.

What it does HealthMentor AI is an intelligent medical record advisor that helps healthcare providers and patients by analyzing queries and delivering expert-level advice based on similar real-world cases. Users input their medical concerns, which are converted into AI-generated embeddings. The system performs a vector search to find closely related medical records, then synthesizes a clear, actionable summary report with clinical insights and recommendations. This enables faster, data-driven decision-making even in environments with limited specialist access.

How we built it We sourced a diverse collection of medical records from Hugging Face and processed them for storage in MongoDB Atlas. Using Google Cloud’s Vertex AI, we generated high-dimensional vector embeddings of each record and created an efficient similarity index to support rapid querying. The frontend was developed using Next.js, delivering a responsive and intuitive user interface, while the backend was built with FastAPI to handle API requests and manage search logic. Both frontend and backend are containerized with Docker and deployed on Google Cloud virtual machines, ensuring scalability and reliability.

Challenges we ran into One major challenge was working within the constraints of cloud services and managed databases, especially in enabling advanced vector search capabilities on MongoDB Atlas’s available tiers. We also faced hurdles optimizing embeddings generation and search latency to deliver near real-time responses. Ensuring secure and seamless authorization flow between frontend and backend required careful handling of authentication tokens. Lastly, tailoring the AI-generated medical summaries to be both accurate and accessible demanded iterative prompt engineering and domain-specific adjustments.

Accomplishments that we're proud of We successfully built an end-to-end AI-powered platform that can interpret complex medical queries and provide evidence-based advice using real-world medical data. The system operates reliably in a cloud environment and supports secure, authenticated user interactions. We achieved performant vector search indexing with Vertex AI embeddings, enabling precise retrieval of relevant clinical cases. Most importantly, we created a user-friendly application tailored for healthcare contexts in Nigeria and Africa, addressing a critical unmet need.

What we learned This project deepened our understanding of AI embeddings for medical language and the technical nuances of scalable vector search. We gained practical experience deploying full-stack applications with secure authentication flows in cloud environments. We learned the importance of prompt design in guiding AI to generate domain-appropriate, trustworthy advice. Additionally, we saw firsthand the power of AI to amplify healthcare expertise and the ethical responsibility that comes with building such systems.

What's next for HealthMentor AI Our next steps include expanding the medical record dataset with more diverse and localized African healthcare data to improve contextual relevance. We plan to enhance the AI’s interpretability by integrating more transparent reasoning and confidence metrics. Improving multilingual support to serve Nigeria’s many languages is a priority to increase accessibility. Finally, we aim to pilot HealthMentor AI in healthcare facilities across Nigeria, gathering user feedback and measuring impact to refine and scale the solution, bringing high-quality medical insights to those who need them most.

What's next for HealthMentor AI

Built With

Share this project:

Updates