In EDS ansehen

Building Digital Twins for Elderly Care: An End-to-End Framework From Data Acquisition to Modeling

Gespeichert in:
Bibliographische Detailangaben
Titel: Building Digital Twins for Elderly Care: An End-to-End Framework From Data Acquisition to Modeling
Autoren: Ziaullah Momand, Pornchai Mongkolnam, Jonathan H. Chan, Nipon Charoenkitkarn, Debajyoti Pal
Quelle: IEEE Access, Vol 13, Pp 169415-169445 (2025)
Verlagsinformationen: IEEE, 2025.
Publikationsjahr: 2025
Bestand: LCC:Electrical engineering. Electronics. Nuclear engineering
Schlagwörter: Elderly digital twin, human digital twins, personalized healthcare, physiological data integration, predictive models, Electrical engineering. Electronics. Nuclear engineering, TK1-9971
Beschreibung: The rapid growth of the aging population, which is expected to reach 2.1 billion by 2050, poses profound challenges for healthcare systems and the quality of life of the elderly. Current research on digital healthcare has focused on incorporating Artificial Intelligence (AI) and wearable technology to deliver personalized care to the elderly population. Digital Twins (DTs) are virtual models that replicate real-world entities using real-time data and offer new possibilities for personalized healthcare. While prior studies have explored the adaptation of DTs for personalized healthcare, especially for older adults, the integration of generative AI (Gen AI) for real-time contextualized caregiver support remains underdeveloped. This study introduces an Elderly Digital Twin (EDT) framework that places Gen AI-driven personalized feedback and decision support at its core, supported by a comprehensive data pipeline, predictive modeling, and interactive user interaction. The framework integrates diverse physiological data (heart rate, SpO2, and sleep) and demonstrates feasibility and applicability by developing cardiac DT, sleep quality DT, and SpO2 monitoring DT models. Advanced predictive modeling achieved strong performance, with LSTM reaching 92% validation accuracy in sleep stage prediction, and Bi-LSTM yielding robust performance in real-time heart rate forecasting (MSE = 0.2944 and MAE = 0.3410). The EDT framework was deployed via a Node.js-based web application that offers real-time physiological monitoring, interactive 3D cardiac simulation, and GPT-4o-powered personalized feedback to enhance caregiver decision-making and elderly self-management. The primary contribution of this study is the incorporation of Gen AI into EDT to translate sensor data and predictive outputs into actionable recommendations for caregivers. The pipeline, predictive models, and ethical safeguards serve as enabling components, and this proof-of-concept study lays the groundwork for future multi-organ EDT and broader clinical use.
Publikationsart: article
Dateibeschreibung: electronic resource
Sprache: English
ISSN: 2169-3536
Relation: https://ieeexplore.ieee.org/document/11153809/; https://doaj.org/toc/2169-3536
DOI: 10.1109/ACCESS.2025.3607603
Zugangs-URL: https://doaj.org/article/5ca5851b2b1e48f88a498beb7dd37dc1
Dokumentencode: edsdoj.5ca5851b2b1e48f88a498beb7dd37dc1
Datenbank: Directory of Open Access Journals
Beschreibung
Abstract:The rapid growth of the aging population, which is expected to reach 2.1 billion by 2050, poses profound challenges for healthcare systems and the quality of life of the elderly. Current research on digital healthcare has focused on incorporating Artificial Intelligence (AI) and wearable technology to deliver personalized care to the elderly population. Digital Twins (DTs) are virtual models that replicate real-world entities using real-time data and offer new possibilities for personalized healthcare. While prior studies have explored the adaptation of DTs for personalized healthcare, especially for older adults, the integration of generative AI (Gen AI) for real-time contextualized caregiver support remains underdeveloped. This study introduces an Elderly Digital Twin (EDT) framework that places Gen AI-driven personalized feedback and decision support at its core, supported by a comprehensive data pipeline, predictive modeling, and interactive user interaction. The framework integrates diverse physiological data (heart rate, SpO2, and sleep) and demonstrates feasibility and applicability by developing cardiac DT, sleep quality DT, and SpO2 monitoring DT models. Advanced predictive modeling achieved strong performance, with LSTM reaching 92% validation accuracy in sleep stage prediction, and Bi-LSTM yielding robust performance in real-time heart rate forecasting (MSE = 0.2944 and MAE = 0.3410). The EDT framework was deployed via a Node.js-based web application that offers real-time physiological monitoring, interactive 3D cardiac simulation, and GPT-4o-powered personalized feedback to enhance caregiver decision-making and elderly self-management. The primary contribution of this study is the incorporation of Gen AI into EDT to translate sensor data and predictive outputs into actionable recommendations for caregivers. The pipeline, predictive models, and ethical safeguards serve as enabling components, and this proof-of-concept study lays the groundwork for future multi-organ EDT and broader clinical use.
ISSN:21693536
DOI:10.1109/ACCESS.2025.3607603