Projects
This page highlights selected projects that reflect our approach to transforming physiological signals into clinically meaningful insight.
Each project represents a concrete instantiation of our broader research program, spanning data acquisition, modeling, and real-world deployment.
Problem
Physiological signals in intensive care units are continuously monitored but rarely stored longitudinally, limiting the ability to study developmental trajectories and detect early signs of deterioration.
Approach
We developed a data warehousing system that captures continuous physiological data across NICU beds and integrates with high-performance computing infrastructure for large-scale analysis.
Impact
This platform enables quantitative modeling of neonatal maturation, supports early detection of adverse events, and transforms transient monitoring data into a durable institutional resource.
Problem
Mild cognitive impairment is often underdiagnosed in primary care due to time constraints and limited access to scalable assessment tools.
Approach
We developed a tablet-based application that collects demographic and voice data, which are processed through automated feature extraction and machine learning models via a secure cloud platform.
Impact
This tool enables rapid, scalable cognitive screening in real-world clinical settings and lowers barriers to early detection and intervention.
Problem
Assessment of gait and fall risk often relies on subjective evaluation or expensive motion-capture systems that are not accessible in routine clinical settings.
Approach
We developed a low-cost, contactless motion capture system using depth cameras to track joint movement and extract quantitative gait metrics.
Impact
This system enables objective, scalable assessment of mobility and supports early identification of fall risk in clinical and community environments.
Problem
Traditional neuromodulation approaches rely on fixed stimulation patterns that may be inefficient or non-selective in targeting neural populations.
Approach
We developed self-learning algorithms that discover optimized stimulation waveforms using computational models of neural systems.
Impact
This approach enables more precise and energy-efficient neuromodulation strategies and supports the development of personalized electroceutical therapies.
Problem
Understanding cognitive decline and resilience requires integrating multiple physiological signals across time, which is difficult with fragmented or short-term data collection.
Approach
We are developing systems that combine EEG, cardiovascular signals, voice, and behavioral data in longitudinal, community-based settings.
Impact
This work supports earlier identification of cognitive decline and advances the study of health trajectories in real-world environments.