Machine learning for digital health: Transforming clinical data into knowledge

Kaur, Rachneet

Permalink

https://hdl.handle.net/2142/117741 Copy

Description

Title

Machine learning for digital health: Transforming clinical data into knowledge

Author(s)

Kaur, Rachneet

Issue Date

2022-12-02

Director of Research (if dissertation) or Advisor (if thesis)

Sowers, Richard B.

Doctoral Committee Chair(s)

Sowers, Richard B.

Committee Member(s)

• Beck, Carolyn L.
• Kesavadas, Thenkurussi
• Hernandez, Manuel E.

Department of Study

Industrial&Enterprise Sys Eng

Discipline

Industrial Engineering

Degree Granting Institution

University of Illinois at Urbana-Champaign

Degree Name

Ph.D.

Degree Level

Dissertation

Keyword(s)

• Artificial intelligence
• Deep learning
• Health analytics
• Gait
• Vision
• Multiple sclerosis
• Alzheimer's disease
• Parkinson's disease

Abstract

Promoting well being and healthy aging in older adults is becoming increasingly crucial to advance towards a better quality of life. In an effort towards the same, we propose some current and novel approaches to the study of neurological disorders in this thesis. Broadly, we attempt to address recurring problems in neurological disorders, namely early-stage disease diagnosis and progression prediction. Diagnosing neurological conditions is difficult, especially in the early stages, many individuals go undiagnosed partly due to the complex heterogeneity in disease progression. Thus, we study the integration of artificial intelligence (AI) and health data that we believe may provide a viable patient-centric approach to aid clinicians in designing novel AI-based disease prediction strategies and monitoring disease progression. Our ultimate objective is to facilitate the future developments of AI in digital healthcare. This work proposes new data-driven machine learning-based solutions utilizing health data from multiple modalities, such as gait (e.g., spatiotemporal gait metrics, acceleration, ground reaction forces), cognitive, functional, and longitudinal clinical assessments, and neural responses measured via electrophysiological (e.g., electromyography (EMG), electrocardiography (ECG)) signals, to improve early disease prediction and progression in neurological movement disorders. We measure our ability to use these signals to classify disability and predict progression of cognitive and motor changes in persons with neuromuscular disorders. This thesis is a multidisciplinary effort that involves novel combinations of sensors, vision, machine learning, bio-mechanics, and dynamical analyses to better characterize neurological disorders. These studies on the integration of AI and health data may provide a viable patient-centric approach to aid clinicians in designing novel AI-based disease prediction strategies and monitoring disease progression. This may help providers to individualize treatment plans and design improved clinical trials; thus, help reduce the skyrocketing healthcare costs in the future. The focus of this dissertation is on the following three areas under the broad umbrella of AI for digital healthcare: 1) Gait analysis for differentiation of neurological disorders, where we focus on disease diagnosis and study gait data-driven methodologies for an automated quantification of neurological gait disorders, such as multiple sclerosis and Parkinson's disease, 2) Clinical data analysis for prediction of disease progression, where we focus on early-stage disease progression prediction and propose machine learning models to identify etiological disease subtypes and study trajectory progression in Alzheimer’s disease, and 3) Virtual reality for analyzing neural responses to anxiety, where we examine the potential of virtual reality neurorehabilitation for ameliorating fall-related anxiety in adults.

Graduation Semester

2022-12

Type of Resource

Thesis

Copyright and License Information

Copyright 2022 Rachneet Kaur

Owning Collections