TwinSight: A Data-Driven Digital Twin Framework for Human-Centric Health Monitoring
PI: Dhananjay Singh (Information Sciences and Technology)
This project proposes the development of an intelligent and privacy-preserving digital twin system to support agingin-place by enabling continuous, unobtrusive monitoring of older adults’ daily activities and living environments. Leveraging a network of low-cost, privacy-conscious IoT sensors, including motion sensors, thermal arrays, and environmental monitors. We aim to generate real-time data that will feed into AI-driven models for activity recognition, posture detection, and health status inference. The digital twin will serve as a dynamic virtual representation of the resident’s home and behavior, offering interactive 3D visualization and real-time alerts to caregivers or family members via a mobile interface. Key project objectives include improving the interpretability of neural network decisions, adapting the system for multi-resident households, and scaling deployment through automated calibration techniques. Over the next year, we will expand our current pilot from two to at least five homes, refine our models with diverse datasets, and collaborate with ICDS researchers to explore robust privacy-preserving AI methods. This interdisciplinary work aligns with the ICDS Digital Twins hub by applying computational modeling and human-centered sensing to support healthcare innovation, aging populations, and data-driven public health strategies.
Desired Skills/Expertise:
Activity recognition and time-series analysis, Neural network training and evaluation such as using PyTorch, TensorFlow, Signal processing and data preprocessing, Privacy-preserving machine learning techniques, IoT sensor integration and embedded systems data collection, Full-stack mobile development for data visualization/dashboard tools.
Other Requirements or Expectations:
Availability for weekly meetings; post-comp’s graduate student status; experience or interest in Digital Twin; Willingness to work with sensitive human data under appropriate IRB and data governance protocols; Interest in interdisciplinary research combining AI and human-centered computing
Specific Objectives:
• Develop and refine posture and activity recognition models from thermal and motion sensor data
• Deploy and maintain sensor systems in real-world home environments
• Generate and analyze pilot data to support future NSF and NIH proposals
• Submit conference and journal publications such as IEEE Pervasive Computing, ACM UbiComp
• Prepare documentation, open-source tools, and a data/code repository
• Support dissemination efforts such as demos, ICDS symposium presentations
Medium/Long-Term Goal:
Submit a large-scale external research proposal such as NIH’s Smart Health program, and NSF’s Smart and Connected Health initiative, based on pilot work. Establish a publicly available dataset and open-source platform to support broader community research on aging-in-place digital twins.
Connection to ICDS Mission:
This project aligns with ICDS’s mission by integrating data science, AI, and IoT technologies to address the critical societal challenge of aging-in-place through a privacy-conscious digital twin framework. It supports interdisciplinary research, real-world impact, and the development of AI-enabled solutions to improve health and quality of life.
Engagement with ICDS:
This project will actively participate in ICDS activities, including talks, symposiums, and collaborative forums.