SEHAT 360
AI-Powered Health Predictions, Personalized Care, and Real-Time Guidance – All in One App.
Created on 13th April 2025
•
SEHAT 360
AI-Powered Health Predictions, Personalized Care, and Real-Time Guidance – All in One App.
The problem SEHAT 360 solves
Healthcare is often not immediate, personalized, or accessible, leaving people to face several challenges:
Delayed Diagnosis: Without the right tools, individuals may miss early signs of serious conditions like joint issues, diabetes, or heart attack risks.
Lack of Personalized Guidance: Even after a diagnosis, many people are unsure of the next steps or how to manage their health.
Access to Healthcare: In emergencies, locating nearby hospitals or doctors can be confusing and time-consuming.
Health Management Overload: Keeping track of medications and doctor visits, especially for chronic conditions, can be overwhelming.
How It Makes Existing Tasks Easier & Safer
Instant Health Predictions & Diagnosis
Joint Analysis: Upload a knee X-ray and get an AI-driven severity analysis, along with personalized recommendations for care.
Diabetes & Heart Attack Risk Prediction: Quickly assess your risk level based on health data, empowering you to make informed decisions about your health.
Real-Time Personalized Guidance
After predictions, the app provides tailored lifestyle suggestions and exercises to help you manage your health proactively and make informed choices.
Nearby Hospitals & Doctors
Use your current location to find nearby hospitals and doctor departments, providing quick access to doctor details, OPD schedules, and directions—reducing wait times and confusion.
How It Makes Tasks Safer
Early Detection & Prevention: The app helps you detect risks early (diabetes, heart attack), allowing for prevention before health issues become serious.
Access to Healthcare: The hospital and doctor features reduce the delay in receiving care, ensuring quick action in emergencies.
Personalized Health Monitoring: AI-powered recommendations guide you to make safer.
This app revolutionizes health management by providing immediate predictions, personalized guidance, and easy access to healthcare, making it safer and easier for users to take control of their health.
Challenges we ran into
Challenges We Ran Into
Machine Learning Model Development and Integration
Model Creation: We used open-source datasets from Kaggle to train our models. For knee severity prediction, we utilized ResNet. For diabetes prediction, we applied a Support Vector Machine (SVM), and for heart attack prediction, we used Linear Regression.
TensorFlow Lite Conversion: After training the models, we converted them into TensorFlow Lite to integrate them into the Android app.
Integration Challenges: The biggest challenge was integrating these models into the Android app. Writing functions to make the models work seamlessly with TensorFlow Lite was complex, especially optimizing them for mobile use. We faced several technical issues during this phase, requiring extensive debugging and adjustments to ensure smooth performance on Android.
Hospital Data Collection & Integration
Targeting a Hospital: For the nearby hospital feature, we targeted hospitals in Nigeria, which posed challenges in gathering accurate hospital data, including doctor details and schedules. The diversity of medical systems across regions made it difficult to get real-time, accurate information.
API Development & Integration
Setting Up API Routes: Setting up the API routes to fetch hospital and doctor data dynamically from JSONBin was another hurdle. We needed to ensure accurate and up-to-date information was available to users.
Hospital Data Integration: Integrating hospital details and doctor schedules into the app based on user location added complexity. Ensuring a smooth and responsive user experience with real-time data was time-consuming, especially under a tight development timeline.