Back to search

University of British Columbia

A machine learning approach to overdose risk assessment

Abstract

dc:description

This project explores the escalating drug toxicity crisis in British Columbia, spotlighting the role of machine learning within eHealth solutions to combat this pressing public health issue. The crisis, rooted in a shift from prescription opioids to potent synthetic alternatives like fentanyl, necessitates innovative intervention strategies. The study leverages extensive data from 2015 to 2019, aiming to develop predictive models for overdoses to improve healthcare responses. The research begins with a systematic review and meta-analysis of machine learning models targeting opioid-related outcomes, demonstrating the predictive strengths of various algorithms in cohort studies. Results indicate these algorithms' effectiveness in forecasting opioid usage and overdose risks. Additionally, a review of clinical decision support systems in addiction and mental health care reveals their critical impact on enhancing diagnosis, treatment, and patient care, based on randomized controlled trials. Central to the thesis is an exploratory data analysis utilizing the British Columbia Provincial Overdose Cohort. This involves rigorous data preparation, including wrangling, addressing missing data, and correcting class imbalances. The study assesses several machine learning models, including ensemble approaches like Random Forest and XGBoost, for their predictive accuracy regarding fatal and general overdoses. Despite challenges, these models demonstrate significant potential, with the best performers achieving over 90% accuracy in predicting general overdoses, though models for fatal overdoses showed lower efficacy. The thesis concludes with an affirmation of machine learning's transformative potential in personalizing addiction psychiatry treatment through eHealth innovations such as the Risk Assessment and Management Platform (RAMP). This novel approach aims to individualize treatment and prevention strategies, contributing to global efforts in mitigating the mental health ramifications of the drug toxicity crisis and transforming healthcare practices. Through the strategic implementation of machine learning, the study underscores a promising avenue for advancing healthcare solutions tailored to the intricacies of addiction and overdose risks, reflecting a significant stride towards mitigating the public health impacts of the drug crisis.

Degree

thesis:*
Name thesis:degree_name
Doctor of Philosophy - PhD
Level thesis:degree_level
doctoral
Discipline thesis:degree_discipline
Neuroscience
Grantor dc:publisher
University of British Columbia
Year dc:date
2024

Author and committee

dc:creator, dc:contributor.*
Author dc:creator
  • Tai, Man Yeung (Andy)
Contributors dc:contributor
  • Krausz, Michael, MD, PhD, FRCPc

Rights

dc:rights
Statement dc:rights
  • Attribution-NonCommercial-NoDerivatives 4.0 International
Language dc:language
eng

Identifiers

dc:identifier.*
Handle dc:identifier
http://hdl.handle.net/2429/87759

Chain of custody

source
Harvested from
University of British Columbia
Base URL
circle.library.ubc.ca/oai/request
Last updated
2026-11-19
Source record
OAI-PMH GetRecord
related terms
citation

Tai, Man Yeung (Andy). A machine learning approach to overdose risk assessment. doctoral thesis, University of British Columbia, 2024. http://hdl.handle.net/2429/87759