A Non-Invasive Smart Anaemia Risk Detection System Using IoT-Enabled Multi-Sensor Fusion and Gradient Boosting Regressor-Based Predictive Analytics
A Non-Invasive Smart Anaemia Risk Detection System Using IoT-Enabled Multi-Sensor Fusion and Gradient Boosting Regressor-Based Predictive Analytics
Authors:
S K Harshita Sree¹, V M Harshini², Mr. D. Syed Ali³, Dr. G. Victo Sudha George⁴, Dr. K. K. Rehkha⁵
¹²Students, ³⁵Assistant Professor, ⁴Professor
Department of Computer Science and Engineering
Dr M.G.R. Educational and Research Institute, Chennai, Tamil Nadu
E-mail: harshitasree27@gmail.com, vmharshini53@gmail.com, syedali.cse@drmgrdu.ac.in, victosudhageorge.cse@drmgrdu.ac.in, rehkhamanuj.cse@drmgrdu.ac.in.
Abstract— Haemoglobin deficiency affects an estimated 2.3 billion individuals globally, yet routine screening remains inaccessible in many settings due to the cost and burden of laboratory-based blood analysis. This paper introduces SmartAnaemia, an IoT-enabled framework for continuous, non-invasive haemoglobin estimation and anaemia risk classification. The system acquires photoplethysmography (PPG) waveforms, peripheral oxygen saturation (SpO₂), heart rate, and skin temperature using a MAX30102 sensor and DS18B20 probe interfaced with an ESP32 microcontroller. Rather than reducing the PPG signal to scalar outputs, 38 morphological and hemodynamic features are extracted per waveform, capturing pulse deformations associated with haemoglobin variation. Sensor data is relayed to a cloud analytics layer via a BLE-to-Python bridge, offloading computational overhead from the edge device. A Gradient Boosting Regressor (GBR) trained on a hybrid clinical dataset achieves a haemoglobin estimation MAE of 1.00 g/dL. A WHO-aligned decision layer maps predicted values to four severity tiers, yielding 91.5% overall classification accuracy with 94% recall for severe cases. The monitoring interface, built as a Progressive Web Application using Vite and React, delivers end-to-end sensor-to-display latency under 300 ms.
Keywords: Internet of Things (IoT); Anaemia Detection; Non-Invasive Diagnostics; Machine Learning; Gradient Boosting Regressor; React.js; ESP32; Telemedicine; Sensor Fusion; Photoplethysmography.