Abstract Title

Using Machine Learning Technique to Explore Factors Associated with Change in Quality of Life Among Permanent Supportive Housing Residents

Presenter Name

Abdullah Mamun

RAD Assignment Number

1115

Abstract

Purpose: The purpose of this study was to identify predictors of change in the overall quality of life (QOL) at 6-months, compared to baseline, among permanent supportive housing residents with a history of chronic homelessness and mental illness.

Methods: Data were collected at baseline and 6-month using 18 questionnaires, encompassing over 100 variables on 457 adults. The short version of the Quality of Life Enjoyment and Satisfaction Questionnaire was used to measure QOL. We used a machine learning technique for dimension reduction to achieve a final predictive model for QOL. We used a two-step approach: first, using a machine learning technique called random forest (RF) for dimension reduction by eliminating unimportant variables, and then using a model selection technique in multiple linear regression (MLR) framework with the reduced set obtained from RF to propose a final model. In the process, we highlighted the utility of RF as a means of exploring the fullness of a dataset in order to identify factors associated with improvement in QOL. We captured the linear relationships only in the final predictive model.

Results: The mean improvement in QOL score at 6-months was 4.24 (SD=13.52, effect size=0.31). Significant predictors of the change in QOL were one’s baseline QOL (estimate=-0.32, p

Conclusion: QOL is a multifaceted concept that encompasses various constructs ranging across physical health, psychological state of mind, social circumstances, environmental factors, etc. We hope that future interventions addressing QOL in this vulnerable population will benefit from our findings. Methodologically, we illustrate the benefit of using machine learning techniques in behavioral/social experiments to leverage “big data” and conduct comprehensive analyses.

Is your abstract for competition or not for competition?

Competition

Research Area

General Public Health

Presentation Type

Poster

This document is currently not available here.

Share

COinS
 

Using Machine Learning Technique to Explore Factors Associated with Change in Quality of Life Among Permanent Supportive Housing Residents

Purpose: The purpose of this study was to identify predictors of change in the overall quality of life (QOL) at 6-months, compared to baseline, among permanent supportive housing residents with a history of chronic homelessness and mental illness.

Methods: Data were collected at baseline and 6-month using 18 questionnaires, encompassing over 100 variables on 457 adults. The short version of the Quality of Life Enjoyment and Satisfaction Questionnaire was used to measure QOL. We used a machine learning technique for dimension reduction to achieve a final predictive model for QOL. We used a two-step approach: first, using a machine learning technique called random forest (RF) for dimension reduction by eliminating unimportant variables, and then using a model selection technique in multiple linear regression (MLR) framework with the reduced set obtained from RF to propose a final model. In the process, we highlighted the utility of RF as a means of exploring the fullness of a dataset in order to identify factors associated with improvement in QOL. We captured the linear relationships only in the final predictive model.

Results: The mean improvement in QOL score at 6-months was 4.24 (SD=13.52, effect size=0.31). Significant predictors of the change in QOL were one’s baseline QOL (estimate=-0.32, p

Conclusion: QOL is a multifaceted concept that encompasses various constructs ranging across physical health, psychological state of mind, social circumstances, environmental factors, etc. We hope that future interventions addressing QOL in this vulnerable population will benefit from our findings. Methodologically, we illustrate the benefit of using machine learning techniques in behavioral/social experiments to leverage “big data” and conduct comprehensive analyses.