CardioCare AI

Inspiration

Cardiac rehabilitation is crucial but often inaccessible, especially for patients in remote areas or those recovering at home. Many lack access to physiotherapists or personalized care. I was inspired to build CardioCare AI to bridge this gap using mobile technology to deliver personalized exercise plans and real time pose correction, ensuring safety and effectiveness during rehab.

What it does

CardioCare AI is a mobile application that:

1. Assesses patient risk based on age, medical history and physical condition.
2. Generates a customized cardiac rehab exercise plan by using the rule based scoring system.
3. Uses Google ML Kit to track user movements and corrects posture in real time.
4. Provides visual guides, performance feedback and progress tracking.

How we built it

1. Frontend: Developed in Flutter for a smooth and accessible user experience across devices.
2. Pose Detection: Integrated Google ML Kit's Pose Detection package to track body joints and form.
3. Backend Logic: Built a Python-based module for generating risk-adjusted exercise routines.
4. Customized Exercise plan generation: Used rule based scoring algorithms.
5. Animations: Used adobe after effects and adobe illustration for animating 2d exercises.

Challenges we ran into

1. Ensuring pose detection worked accurately across different body types, lighting conditions and mobile devices.
2. Developing a real-time feedback loop without creating lag on low-end devices.
3. Balancing medical safety with usability, especially when designing exercise rules.
4. Personalizing rehab plans dynamically while keeping the system simple for non-technical users.

Accomplishments that we're proud of

1. Built an end-to-end solution for cardiac rehabilitation entirely on mobile - no external hardware required.
2. Integrated real-time pose detection and correction that works smoothly even on budget Android devices.
3. Created a dynamic exercise engine that adjusts workouts based on user risk and progress.
4. Designed an inclusive and accessible UI suitable for elderly users with minimal tech experience.
5. Achieved high pose detection accuracy through testing in various real-world conditions.
6. Developed animated exercise demonstrations that guide users without overwhelming them.
7. Successfully conducted initial user testing and validation with non-technical participants.

What we learned

1. How to combine healthcare logic with ML and app development for a real world problem.
2. The practical limits and capabilities of Google ML Kit in pose tracking.
3. Importance of UI/UX in medical applications, simplicity can be life saving.
4. How small innovations can make a big difference in patient safety and autonomy.

What's next for CardioCare AI

1. Voice-Assisted Guidance: Integrate voice prompts to guide users through exercises hands-free, especially beneficial for elderly patients.
2. Wearable Integration: Connect with smartwatches or fitness bands to monitor heart rate, oxygen levels, and fatigue in real-time.
3. Doctor Dashboard: Build a web-based portal for physiotherapists and cardiologists to remotely monitor patient's rehab progress.
4. Multi-Language Support: Add regional language options to ensure accessibility across diverse linguistic communities.
5. Gamification and Motivation: Introduce progress badges, streaks and motivational messages to encourage consistent rehab.
6. Posture Heatmaps: Visualize pose accuracy over time using heatmaps to show which parts of the body need improvement.
7. Emergency Alert System: Add real time alerts if the app detects risky behaviour like signs of overexertion or falls.
8. Offline mode: Optimize the app for offline use so patients can access their plans without active internet.
9. Integration with EMR/EHR Systems: Allow hospitals or clinics to sync rehab data with existing medical records.
10. Clinical Trials and Validation: Partner with healthcare institutions to conduct formal clinical testing and get validation for medical use.
11. AI-Powered Feedback Loop: Use anonymized training data to continuously improve the accuracy of pose correction models.
12. Insurance/Health Providers Partnerships: Explore integration with health insurance for reimbursement or incentives for app usage.
13. Expanding use cases: Adapt the platform for other chronic conditions such as stroke rehab, orthopedic recovery or diabetic exercise management.

Built With

• dart
• flutter
• google-ml-kit
• lottie