Title: Novel Machine Learning Approaches with Applications in Healthcare and Social Welfare
Date: Friday, August 1st, 2025
Time: 12pm EST
Location: https://gatech.zoom.us/j/98892301027 (or in person - Groseclose 404)
Anjolaoluwa Popoola
PhD Candidate in Machine Learning
H. Milton Stewart School of Industrial and Systems Engineering(ISyE)
Georgia Institute of Technology
Committee
1. Dr. Kamran Paynabar (Advisor), H. Milton Stewart School of Industrial and Systems Engineering, Georgia institute of Technology
2. Dr. Gian-Gabriel Garcia, H. Milton Stewart School of Industrial and Systems Engineering, Georgia institute of Technology
3. Dr. Jing Li, H. Milton Stewart School of Industrial and Systems Engineering, Georgia institute of Technology
4. Dr. Lauren Steimle, H. Milton Stewart School of Industrial and Systems Engineering, Georgia institute of Technology
5. Dr. Pooyan Kazemian, Weatherhead School of Management, Case Western Reserve University
Abstract
As the world evolves rapidly, the use of machine learning (ML) and artificial intelligence (AI) in healthcare and social welfare has become crucial. These fields are fundamental to our society, and machine learning offers powerful tools to tackle prevalent issues such as handling high-dimensional and sparse data, improving data quality and hence, predictive accuracy for chronic disease management, and optimizing decision-making under uncertainty in these fields. Similarly, social welfare professionals and policy makers urgently need targeted interventions and equitable resource allocation. This thesis covers three studies, each of which targets critical aspects of healthcare and social welfare. Chapter 2 proposes a supervised learning framework for identifying and adjusting under-reported responses in Food Frequency Questionnaire (FFQ) data using a random forest classifier and a novel measurement error adjustment algorithm. Because FFQs are self-reported data, they are susceptible to response bias, social desirability bias, and misclassification, hereby leading to a loss in statistical diet-disease associations. This method combats these issues without requiring other biased tools and provides a scalable solution for improving dietary data quality in epidemiological studies.
Chapter 3 shows a robust framework for assessing the risk of homelessness among youths exiting the foster care system. Homelessness remains a persistent concern among young individuals who are transitioning out of the foster care system within the United States. Recent research in this field has shown a scarcity of models dedicated to reducing this incidence. Hence, we present a novel application of random survival forest to estimate the probability and timing of foster youth exiting to homelessness. This chapter emphasizes the use of personalized survival curves and risk classification models to inform efficient and timely preventive policies in social welfare.
Lastly, Chapter 4 aims at personalized prediction and control of blood glucose levels using dynamic system modeling. Patients with diabetes in the ICU typically rely on two methods to monitor their blood glucose levels: blood tests analyzed in the laboratory and fingerstick tests. While fingerstick tests offer a cost-effective and convenient approach, they sacrifice accuracy, providing only approximate measurements. We present a dynamic systems model combining deep learning models and Kalman filtering to accurately predict blood glucose trajectory and assist with personalized glucose control over time. The resulting model, called a Deep Kalman Filter (DEEP-KF) model can be applied in the ICU and other healthcare settings. The main goal of this thesis is to introduce the developed novel methodologies in machine learning and provide solutions that address specific challenges in each domain, hereby improving predictive modeling in healthcare and social welfare.