International Research Journal of Engineering and Technology (IRJET)
e-ISSN: 23950056
Volume: 12 Issue: 11 | Nov 2025
p-ISSN: 2395-0072
www.irjet.net
MedAi: A Multi-Sensor Smart watch and Machine Learning Framework for Real-Time Multi-Disease Prediction ARJUN M MANOJ Arjun M Manoj, MSc Computer Science St. Thomas (Autonomous) College, Thrissur 680001, Kerala, India ---------------------------------------------------------------------***--------------------------------------------------------------------statistics or providing feedback on unusual results for Abstract - This review examines the "MedAi" framework, a smart watch-based application designed for the prediction of multiple common diseases using machine learning algorithms. The core objectives include providing early disease vulnerability detection and offering real-time health assistance. The framework comprises three main modules: a prototype smart watch called 'Sense O’Clock' equipped with eleven sensors for bodily data collection, a machine learning model for data analysis and prediction, and a mobile application for displaying results. The study employed various machine learning algorithms, with the Random Forest (RF) algorithm demonstrating the highest accuracy at 99.4% in predicting diseases such as ischemic heart disease, hypertension, and diabetes. This system represents a notable advancement in health information technology, aiming to mitigate the time consuming and expensive nature of traditional doctor visits and pathological tests.
multiple diseases. There is a demand for affordable smart wearables equipped with numerous sensors to monitor bodily statistics and prevent sudden tragedies. Machine-Learning for Disease Prediction Machine learning (ML) has emerged as a vast and continuously growing field of research for disease prediction. While many existing ML-based prediction systems focus on single diseases, such as heart disease (due to its high death rate), there is a critical need for systems capable of predicting multiple disease vulnerabilities simultaneously. The "MedAi" system addresses this by presenting a smartwatch based prediction framework for twelve common diseases, leveraging various machine learning algorithms to achieve high accuracy.
I.
II.1 Smartwatch Monitoring: hardware, sensors, data challenges Existing smart wearables often fall short in comprehensive disease detection. Many branded smartwatches are expensive and mainly function as fitness trackers, not providing in-depth analysis or feedback for health issues. For instance, Medibot, a medical chatbot for smartwatches, claimed to be the first disease prediction app with a smartwatch, but it only included four sensors, which could bias predictions. In contrast, the 'Sense O'Clock' prototype, central to the MedAi system, is designed with eleven sensors to collect a wider range of bodily statistics, making it more feasible for multi-disease prediction. It aims to be more cost- and power-efficient than smart garments or high-functionality branded watches. Data collection for disease prediction systems is challenging; readily available open-source datasets for all targeted diseases are scarce, and collecting observational data from patients wearing smartwatches is not always feasible. Privacy concerns arise when utilizing third party APIs for data storage, as health related information is sensitive. The accuracy and reliability of smartwatch data for clinical measurements remain an ongoing discussion. Some studies show smartwatches under/overestimate measurements (e.g., blood pressure) and can be affected by rapid acoustic fluctuations in noise monitoring, indicating they are not always ready for clinical usage without systematic bias. While some commercial smartwatches have demonstrated moderate
II. LITERATURE SURVEY
INTRODUCTION
Context & Motivation Healthcare information technology is rapidly advancing, driven by the need for quick illness prediction and medication access, as traditional medical visits and tests are often time-consuming and costly. Globally, prominent diseases like cardiovascular diseases, cancers, respiratory diseases, diabetes, and kidney diseases cause alarming numbers of deaths and contribute to massive healthcare expenditures. Many countries, especially in Asia, struggle to provide adequate health coverage, leading to concerns about diseases like malaria, dengue fever, and typhus. A significant issue is the occurrence of sudden deaths due to undiagnosed illnesses, such as advanced cancer, heart attack, or brain hemorrhage. Routine checkups are recommended, but often hindered by work pressure, family responsibilities, or lack of time. This highlights a critical need for convenient, real-time health monitoring systems that can provide full-time assistance and suggest remedies. Wearable Health Monitoring The modern era offers technology-driven solutions for healthcare, with health and fitness mobile application downloads increasing significantly. The rapid growth in intelligent wearables, predicted to exceed one billion by 2022, has made realtime health checkups possible. However, many branded smartwatches are expensive and primarily function as fitness trackers, lacking features for analyzing bodily
© 2025, IRJET
|
Impact Factor value: 8.315
|
ISO 9001:2008 Certified Journal
|
Page 329