PhD Dissertation Proposal by TEH Seng Khoon | Machine Learning for In-Home Detection of Cognitive Decline

Dementia is a neurodegenerative disease with a prevalence rate expected to triple by 2050, posing a significant challenge for health services. In-home digital biomarker technology is an emerging pragmatic approach to permit objective, ecologically valid, and long-term continuous measurement of cognitive health status, potentially increasing opportunities to detect cognitive decline at a reversible stage which is known as Mild Cognitive Impairment (MCI). Though promising, the large volume of noisy data measurements as a consequence of long-term continuous monitoring in a naturalistic setting, coupled with limited human sample sizes mainly due to ethical and financial reasons, poses significant challenges to extracting reliable predictive information from digital biomarker technology. In this dissertation proposal, we aim to systematically address these challenges using Machine Learning (ML). First, we addressed the knowledge gaps in the understanding of the predictive accuracy performance of ML employed for in-home digital biomaker technologies to detect cognitive decline by proposing a systematic review and meta-analysis of relevant studies. Application Programming Interface (API)-assisted literature search was employed on major academic research databases including IEEE-Xplore to extract more than 60,000 articles for analysis. Next, we addressed the technical challenge of tackling the prevalent small and noisy real-world data problem in digital biomarker studies by investigating a unique ML technique, known as the predictive self-organizing fuzzy neural network architecture. Data from a Singapore cross-sectional study with 49 subjects measured using multiple sensors over two months were employed for the evaluation of MCI detection efficacy. Lastly, we address the problem of effective extraction of meaningful predictive features for MCI detection from long-term noisy continuous data. Specifically, we propose the novel encoding of potential episodes of aberrant spatio-temporal behavioral data that is indicative of MCI into a self-organizing episodic memory neural network to learn the differential changes in the episodic sequence over time. Currently, we are actively designing and evaluating the new self-organizing neural network with real-world data from an ongoing digital biomarker longitudinal study that collects more than a year of high-temporal resolution intra-day spatio-temporal movement data and clinically validated cognitive status data from 40 older adults.

Seng Khoon TEH is a part-time Ph.D. candidate at the School of Computing and Information Systems at Singapore Management University (SMU) under the guidance of Professor Tan Ah-Hwee. His Ph.D. research focuses on transdisciplinary domains encompassing the development of self-organizing neural networks for application on digital biomarker technologies to early detect geriatric diseases. Prior to the pursuit of the doctorate degree, he received B.Eng and M.Eng in Biomedical Engineering from National University of Singapore (NUS). He further received his Master of Information Technology for Business (Analytics) from SMU. To date, he is the author of nine peer-reviewed articles and holds five granted patents in technology at the interaction of Biomedical Engineering and Informatics.