57 MedManager

Inspiration

Medication errors are a leading cause of medical harm, accounting for 44% of all cases. With an aging population where over 50% of individuals over 65 take four or more medications, we saw a pressing need for a tool to reduce these risks. MedManager was inspired by the goal of combining technology with healthcare to improve safety and empower patients to manage their medications effectively.

What it does

MedManager allows users to track their medications, automatically checks for conflicts with existing prescriptions, and uses AI to verify that medications are correct by comparing them to prescribed labels. It also features MediBot, an intelligent chatbot that provides users with medication guidance, answers questions, and helps navigate the app.

How we built it

We built MedManager with a Flask backend to handle medication data and API integrations, and a React-native front end for a seamless user experience. The AI verification uses a pre-trained model for image analysis to validate medication labels. The chatbot, MediBot, leverages natural language processing (NLP) to provide accurate, real-time assistance. We used Terraform to help create the Docker image to host the AIs we use data is securely stored in a MongoDB database to ensure scalability and reliability.

Challenges we ran into

A large challenge was training the pill identification model. We wanted a model that could recognize pills in an image, and then identify what medication the pill was. Our team used MobileNetV2 as our foundation. Performed some preprogramming on the image, then trained the model on a data set of pill images. The second issue stemmed from the model's integration with the front end. We needed to post a detailed enough image to allow the model to predict accurately, but a small enough image to maintain a normal API request size.

Accomplishments that we're proud of

We are extremely proud of our entire app as a whole. We worked very efficiently diving up tasks between front end, back end, and model development. Because of this delegation we were able to complete our initial objectives and even begin on some extra features.

What we learned

Our largest learning point was how to work as a team. This improved our efficiency much more than expected. We refined our knowledge of how to train models. We learned some new technologies like Terraform, and got even more experience with technologies like MongoDB, Docker, Tensorflow, and Github.

What's next for MedManager

At MedManager, our vision is to make medication safety accessible to as many people as possible. We plan to introduce multi-language support, ensuring that users from diverse backgrounds can benefit from our app. This inclusivity is at the core of our mission to reduce medication errors worldwide.

Next, we aim to develop a dedicated Healthcare Provider Mode, enabling doctors, nurses, and pharmacists to manage the medications of multiple patients seamlessly. This feature would empower healthcare providers to detect potential conflicts and ensure accuracy at scale, transforming MedManager into a vital tool for both individual users and professionals.

As we continue to grow, we will also explore partnerships with healthcare systems and integrate advanced analytics to provide proactive insights, further reducing the risks of medication-related harm.