🤰 AI-Powered Fetal Monitoring Belt

This project is part of a year-long Engineering Design Project (EDP) at IIIT Jabalpur. The AI-Powered Fetal Monitoring Belt is an intelligent, wearable medical device that non-invasively tracks fetal and maternal parameters using ECG and EMG sensors. It applies advanced signal separation (CAA-CycleGAN) and AI classification (LightGBM) to provide real-time fetal health insights via a cloud-based dashboard. Designed for affordability and accessibility, the system bridges the gap between hospital-grade prenatal care and home-based monitoring.

---

🚀 Project Summary

The AI-Powered Fetal Monitoring Belt is a non-invasive, smart wearable that monitors critical fetal and maternal parameters using embedded sensors and intelligent AI models. Unlike traditional Cardiotocography (CTG) systems, this belt offers a low-cost, real-time alternative with cloud connectivity, making fetal healthcare accessible even in remote or resource-limited settings.

---

🎯 Objectives

• Enable real-time monitoring of fetal health during second and third trimesters.
• Use multimodal physiological data (ECG, EMG, accelerometer, temperature).
• Apply AI techniques for fetal health classification.
• Deliver real-time alerts and historical trends to doctors and expecting mothers via Web UI.
• Offer a cost-effective and scalable solution for maternal healthcare.

---

🧠 Novel Contributions

• Dual-Sensor Integration: Combines ECG (fetal heart rate) and EMG (uterine contractions) in one belt.
• CAA-CycleGAN for Signal Separation: A deep learning model that isolates fetal ECG from noisy abdominal recordings.
• AI-Powered Health Prediction: Uses a LightGBM classifier trained on medical data to detect fetal health conditions with 98.38% accuracy and 0.9985 ROC AUC.
• MERN Stack Web UI: Real-time dashboard for clinicians and mothers.
• Edge + Cloud Integration: Supports both local processing and cloud-based visualization.

---

🛠️ System Architecture

1. Data Collection

• AD8232 ECG Sensor: Monitors fetal heart rate.
• EMG Sensor: Detects uterine contractions.
• Temperature & Accelerometer Sensors: Monitor maternal vitals and fetal movement.

2. Signal Processing

• CAA-CycleGAN: Deep learning model for separating fetal ECG (FECG) from maternal ECG.
• Attention Mechanism: Enhances relevant signal components for cleaner extraction.

3. AI Prediction

• LightGBM Classifier: Predicts fetal health status – Normal, Suspect, or Pathological.

4. Web UI (MERN Stack)

• Real-time visualizations and alerts for patients and clinicians.
• Historical data trends, downloadable reports, and anomaly notifications.

---