proceeding

IoMT-Based AI-Driven Respiratory Risk Prediction App with Firebase Integration for Low-Resource Settings

A. S. M. M. R. KhanChittagong University of Engineering and Technology,Dept. of Biomedical Engineering,Chittagong,BangladeshArifa TanzimUniversity of Science and Technology ChittagongTanjim MahmudRangamati Science and Technology University,Dept. of CSE,Rangamati,Bangladesh,4500Abdurahim BoltayevMamun University,Khiva,UzbekistanAbubokor HanipMohammad Shahadat HossainUniversity of Chittagong,Dept. of CSE,Chittagong,Bangladesh,4331

2025

Ochiq kirish: yoʻq

ABI

Annotatsiya

This paper presents a practical, low-cost Internet of Medical Things (IoMT) system for real-time respiratory health monitoring. This framework integrates an ESP32 microcontroller with MAX30102 and LM35 sensors, which capture blood oxygen saturation (SpO2), heart rate (BPM), and body temperature. A Random Forest model was trained on a publicly available COVID-19 dataset and converted to TensorFlow Lite for on-device risk classification. The model achieved 97% accuracy, 100% precision, and an AUC-ROC of 0.996 in offline testing. Real-time data is transmitted via Bluetooth to an Android application for live display. Firebase integration enables remote data storage and automated email alerts for abnormal readings. In testing across 20 individuals, the system reached an average real-time accuracy of 93.3%, including 100% accuracy for stable (non-risk) conditions. Built with affordable, open-source components, this system offers a reliable and scalable solution for monitoring respiratory conditions in home-based or resource-constrained environments.

Hali tarjima qilinmagan

Identifikatorlar

DOI: 10.1109/gitcon65266.2025.11377379

Iqtiboslar va manbalar

0 ta iqtibos14 ta foydalanilgan manba