Level of Education of Students Involved
Graduate
Faculty Sponsor
Hugh Gong
College
College of Arts & Sciences (CAS)
Discipline(s)
Mathematics & Statistics
Presentation Type
Poster Presentation
Symposium Date
Spring 4-24-2025
Abstract
Wearable health monitoring systems, integrated with data-driven predictive models, are transforming personalized healthcare by enabling early disease detection and proactive intervention. Traditional health monitoring relies on rule-based thresholds and prior medical knowledge, limiting adaptability to individual variations. This project leverages machine learning models—Random Forest, Support Vector Machines (SVM), and Neural Networks—to analyze physiological data from wearable devices, enhancing predictive accuracy in detecting conditions like heart disease and sleep apnea.
The study addresses the growing need for real-time, personalized health monitoring, particularly as populations age and chronic diseases become more prevalent. Arizona serves as a testbed for validating model performance under real-world conditions.
Data is collected from pulse ellipsoid levels, sleep wave patterns, ECG readings, and blood oxygen saturation (SpO₂) measurements. A structured preprocessing pipeline ensures data quality, employing missing value imputation, normalization, and feature engineering to enhance model performance.
Evaluation metrics, including accuracy, precision, F1-score, and cross-validation, validate the models. A key challenge—performance deterioration with new data—is mitigated through retraining strategies. Live health monitoring with alert mechanisms further enables real-time detection of abnormalities and early medical intervention.
Findings demonstrate that machine learning-based predictive models outperform traditional methods in detecting health risks and enabling preventive care. This study concludes that integrating predictive analytics into wearable devices enhances personalized healthcare, improving patient outcomes and reducing emergency health events.
Recommended Citation
Akkinapelli, Manideep, "Predictive Models for Health Monitoring in Wearable Devices" (2025). Symposium on Undergraduate Research and Creative Expression (SOURCE). 1449.
https://scholar.valpo.edu/cus/1449
Biographical Information about Author(s)
Myself Manideep Akkinapelli from India. I am a experienced Business Analyst dedicated to continuous professional development and leveraging my skills to drive business success. My core competencies include SQL, Excel, Agile methodologies, UAT, requirements elicitation, and stakeholder communication. My experience involves developing pricing solutions, conducting gap analysis, creating storyboards, and resolving system defects. Currently, I am expanding my expertise through a Master's in Analytics and Modeling at Valparaiso University (expected May 2025), building upon my previous education at ITM Business School and Satavahana University. My background demonstrates a strong commitment to analytical thinking, collaboration, and effective problem-solving.