Project Overview:

The Smart Health Monitoring System is an IoT + AI-powered web application that continuously monitors vital signs such as heart rate, temperature, and blood oxygen level (SpO2) of patients. It sends real-time data to a web server, where ML algorithms detect anomalies and notify medical personnel in case of emergencies.

This system is ideal for remote patient monitoring, especially for elderly or chronic patients, and reduces the need for constant manual checks.

Technologies Used:

Hardware:

• Microcontroller: Arduino Uno / ESP32 / Raspberry Pi

• Sensors:

  • MAX30100 / MAX30102 – Heart rate and SpO2 sensor

  • DHT11 / LM35 – Body temperature sensor

  • Wi-Fi Module: NodeMCU / ESP8266 (for online data transfer)

Frontend:

• HTML, CSS, Bootstrap

• JavaScript (Chart.js for live graphs)

Backend:

• Option 1: PHP with MySQL

• Option 2: Node.js with MongoDB

• Option 3: Java Spring Boot with PostgreSQL

AI/ML:

• Python (scikit-learn or TensorFlow)

• Anomaly detection using decision tree, logistic regression, or threshold-based logic

System Architecture:

1. Sensors collect real-time data from the patient.

2. Microcontroller transmits data to a web server via Wi-Fi.

3. Data is stored in a database through REST APIs.

4. Machine Learning model processes historical + real-time data to detect:

   • Abnormal heart rate (e.g., < 60 bpm or > 100 bpm)

   • Fever (e.g., > 100.4°F / 38°C)

   • Low oxygen levels (SpO2 < 95%)

5. If abnormalities are detected:

   • Alert is triggered via SMS or Email.

   • Dashboard shows “Critical” tag with patient’s vitals.

6. Web dashboard displays:

   • Live readings

   • Graphical trends over time

   • Alert log

ML Model Description (Simple Example):

• Train a logistic regression model using sample health data:

  • Inputs: Heart rate, SpO2, temperature

  • Output: 0 = normal, 1 = abnormal

• Model is trained offline using Python and deployed in the backend.

• For real-time predictions, latest vitals are sent to the model API and decision is returned.