HeartRate AI
As part of the ApplicationsAI module, my team and I developed a project combining AI and IoT for patient medical monitoring. My main role was to design artificial intelligence models capable of classifying heart diseases, predicting heart rate (BPM), and detecting potential pathologies.
Project Demo
Watch HeartRate AI in action
Project Overview
Problem
Traditional heart rate monitoring systems struggle with motion artifacts and require expensive medical equipment, limiting accessibility for continuous patient monitoring and early disease detection.
Solution
Developed an AI-powered heart rate monitoring system combining IoT sensors (ECG, PPG, 3-axis accelerometer) with XGBoost machine learning algorithm to filter motion artifacts and provide accurate heart rate predictions with confidence scores.
Results
- • Successfully implemented XGBoost algorithm for heart rate prediction
- • Integrated multiple sensor data streams (ECG, PPG, accelerometer)
- • Developed motion artifact filtering system for accurate readings
- • Created confidence scoring mechanism for prediction reliability
Technologies
XGBoost
Machine Learning
Python
Language
Troika Dataset
Data
ECG/PPG Signals
Medical IoT
3D Accelerometer
Sensors
Photoplethysmography
Physiology
Development Team
This project was developed in collaboration with Zayd El Ouaragli, Abdoul Latif KINDA, Ali Guinga, OUADIA AMEKRAN, Tefinjanahary Anicet MAHARAVO and Michel Sagesse Kolié, under the supervision of Mr. Otman Aghzout.
Technical Description
We used the Troika dataset, integrating signals from ECG, PPG, and 3-axis accelerometer sensors. The algorithm used, XGBoost, allows us to estimate heart rate by filtering out arm movement interferences through comparison of PPG and accelerometer signals.
A confidence value accompanies each estimation to indicate its reliability. The operation is based on the physiological principles of photoplethysmography, distinguishing signals related to heartbeats from those caused by movement.
Future Improvements
Future improvements are planned, including patient testing and public presentation of the developed tool.
Explore More Projects
Sentiment Analysis
Advanced sentiment analysis system using NLP and machine learning for customer feedback analysis.
Movie Recommendation System
Machine learning movie recommendation system using KNN and text similarity algorithms.