{"id":{"repo_id":"uwo","oai_identifier":"oai:uwo.scholaris.ca:20.500.14721/39100"},"canonical_url":"https://search.dev.ndltd.org/etd/uwo/oai:uwo.scholaris.ca:20.500.14721/39100","repository":{"repo_id":"uwo","name":"Western University","base_url":"https://uwo.scholaris.ca/server/oai/request"},"display":{"title":"An Expandable Stage-Based Machine Learning and Statistical Pipeline for Detecting Adverse Drug Reaction Risks on Complex Multidimensional Datasets","abstract":"Adverse Drug Reactions (ADRs) are a persistent challenge to patient safety, often escaping detection in aggregate analyses. This research introduces expandable, modular pipeline that identifies statistically significant drug–ADR associations through a multivariable, propensity-adjusted framework tailored to demographic and clinical subgroups. The pipeline begins with a drug selection stage, enabling prioritization of relevant drug exposures using Random Forest classifiers or chi-square tests. Subsequent stages support both univariable and multivariable subgroup analysis, where propensity scores are computed, and cases are matched into bins to achieve covariate balance. Significant drug–ADR pairs are then examined using logistic regression to evaluate the contribution of individual covariates and their interactions. The methodology extends to cross-dataset validation, comparing results from the U.S. FDA’s OpenFDA database and Health Canada’s Vigilance database. Final outputs are presented through interactive visualizations such as Sankey diagrams and heatmaps that reveal subgroup-specific patterns and key contributing factors. By integrating statistical rigor, subgroup analysis, and visual interpretability, this pipeline offers a flexible and comprehensive tool for enhancing ADR detection, validation, and exploration in national pharmacovigilance datasets.","abstract_html":"Adverse Drug Reactions (ADRs) are a persistent challenge to patient safety, often escaping detection in aggregate analyses. This research introduces expandable, modular pipeline that identifies statistically significant drug–ADR associations through a multivariable, propensity-adjusted framework tailored to demographic and clinical subgroups. The pipeline begins with a drug selection stage, enabling prioritization of relevant drug exposures using Random Forest classifiers or chi-square tests. Subsequent stages support both univariable and multivariable subgroup analysis, where propensity scores are computed, and cases are matched into bins to achieve covariate balance. Significant drug–ADR pairs are then examined using logistic regression to evaluate the contribution of individual covariates and their interactions. The methodology extends to cross-dataset validation, comparing results from the U.S. FDA’s OpenFDA database and Health Canada’s Vigilance database. Final outputs are presented through interactive visualizations such as Sankey diagrams and heatmaps that reveal subgroup-specific patterns and key contributing factors. By integrating statistical rigor, subgroup analysis, and visual interpretability, this pipeline offers a flexible and comprehensive tool for enhancing ADR detection, validation, and exploration in national pharmacovigilance datasets.","abstract_has_math":false,"creators":["Samadieh, Mehdi"],"institution":"The University of Western Ontario","degree_name":"M Sc","degree_level":null,"degree_discipline":"Computer Science","degree_department":null,"school":null,"contributors":[],"advisors":["Sedig, Kamran"],"committee_chairs":[],"committee_members":[],"year":2025,"date_issued":"2025-10-29","date_published":"2025-10-29","updated_at":"2026-07-27T21:56:03Z","subjects":["Adverse Drug Reactions","Pharmacovigilance","Propensity Score Adjustment","Subgroup Analysis","Random Forest","Logistic Regression","Signal Detection","Visual Analytics","FAERS","Canada Vigilance"],"languages":["en"],"rights":["Attribution-NonCommercial-NoDerivatives 4.0 International"],"rights_urls":[],"identifier_entries":[]},"links":{"outbound_url":"https://hdl.handle.net/20.500.14721/39100","outbound_label":"Handle","outbound_source":"dc:identifier.uri"},"metadata_groups":[{"id":"people","label":"People","entries":[{"key":"dc:contributor.advisor","label":"Advisor","values":["Sedig, Kamran"]},{"key":"dc:creator","label":"Author","values":["Samadieh, Mehdi"]}]},{"id":"academic_context","label":"Academic Context","entries":[{"key":"dc:date.accessioned","label":"Dc Date Accessioned","values":["2025-11-20T19:50:15Z"]},{"key":"dc:date.issued","label":"Date","values":["2025-10-29"]},{"key":"dc:publisher","label":"Institution","values":["The University of Western Ontario"]},{"key":"dc:type","label":"Dc Type","values":["thesis"]},{"key":"thesis:degree_discipline","label":"Discipline","values":["Computer Science"]},{"key":"thesis:degree_name","label":"Degree Name","values":["M Sc"]},{"key":"thesis:institution_name","label":"Thesis Institution Name","values":["The University of Western Ontario"]}]},{"id":"subjects_keywords","label":"Subjects and Keywords","entries":[{"key":"dc:subject","label":"Dc Subject","values":["Adverse Drug Reactions","Pharmacovigilance","Propensity Score Adjustment","Subgroup Analysis","Random Forest","Logistic Regression","Signal Detection","Visual Analytics","FAERS","Canada Vigilance"]}]},{"id":"language_rights","label":"Language and Rights","entries":[{"key":"dc:language.iso","label":"Language (ISO)","values":["en"]},{"key":"dc:rights","label":"Dc Rights","values":["Attribution-NonCommercial-NoDerivatives 4.0 International"]}]},{"id":"identifiers","label":"Identifiers","entries":[{"key":"dc:identifier.uri","label":"Identifier URI","values":["https://hdl.handle.net/20.500.14721/39100"]}]},{"id":"additional","label":"Additional Metadata","entries":[{"key":"dc:description.abstract","label":"Abstract","values":["Adverse Drug Reactions (ADRs) are a persistent challenge to patient safety, often escaping detection in aggregate analyses. This research introduces expandable, modular pipeline that identifies statistically significant drug–ADR associations through a multivariable, propensity-adjusted framework tailored to demographic and clinical subgroups. The pipeline begins with a drug selection stage, enabling prioritization of relevant drug exposures using Random Forest classifiers or chi-square tests. Subsequent stages support both univariable and multivariable subgroup analysis, where propensity scores are computed, and cases are matched into bins to achieve covariate balance. Significant drug–ADR pairs are then examined using logistic regression to evaluate the contribution of individual covariates and their interactions. The methodology extends to cross-dataset validation, comparing results from the U.S. FDA’s OpenFDA database and Health Canada’s Vigilance database. Final outputs are presented through interactive visualizations such as Sankey diagrams and heatmaps that reveal subgroup-specific patterns and key contributing factors. By integrating statistical rigor, subgroup analysis, and visual interpretability, this pipeline offers a flexible and comprehensive tool for enhancing ADR detection, validation, and exploration in national pharmacovigilance datasets."]},{"key":"dc:title","label":"Title","values":["An Expandable Stage-Based Machine Learning and Statistical Pipeline for Detecting Adverse Drug Reaction Risks on Complex Multidimensional Datasets"]}]}],"canonical_facts":{"dc:contributor.advisor":["Sedig, Kamran"],"dc:creator":["Samadieh, Mehdi"],"dc:date.accessioned":["2025-11-20T19:50:15Z"],"dc:date.issued":["2025-10-29"],"dc:description.abstract":["Adverse Drug Reactions (ADRs) are a persistent challenge to patient safety, often escaping detection in aggregate analyses. This research introduces expandable, modular pipeline that identifies statistically significant drug–ADR associations through a multivariable, propensity-adjusted framework tailored to demographic and clinical subgroups. The pipeline begins with a drug selection stage, enabling prioritization of relevant drug exposures using Random Forest classifiers or chi-square tests. Subsequent stages support both univariable and multivariable subgroup analysis, where propensity scores are computed, and cases are matched into bins to achieve covariate balance. Significant drug–ADR pairs are then examined using logistic regression to evaluate the contribution of individual covariates and their interactions. The methodology extends to cross-dataset validation, comparing results from the U.S. FDA’s OpenFDA database and Health Canada’s Vigilance database. Final outputs are presented through interactive visualizations such as Sankey diagrams and heatmaps that reveal subgroup-specific patterns and key contributing factors. By integrating statistical rigor, subgroup analysis, and visual interpretability, this pipeline offers a flexible and comprehensive tool for enhancing ADR detection, validation, and exploration in national pharmacovigilance datasets."],"dc:identifier.uri":["https://hdl.handle.net/20.500.14721/39100"],"dc:language.iso":["en"],"dc:publisher":["The University of Western Ontario"],"dc:rights":["Attribution-NonCommercial-NoDerivatives 4.0 International"],"dc:subject":["Adverse Drug Reactions","Pharmacovigilance","Propensity Score Adjustment","Subgroup Analysis","Random Forest","Logistic Regression","Signal Detection","Visual Analytics","FAERS","Canada Vigilance"],"dc:title":["An Expandable Stage-Based Machine Learning and Statistical Pipeline for Detecting Adverse Drug Reaction Risks on Complex Multidimensional Datasets"],"dc:type":["thesis"],"thesis:degree_discipline":["Computer Science"],"thesis:degree_name":["M Sc"],"thesis:institution_name":["The University of Western Ontario"]},"updated_at":"2026-07-27T21:56:03Z"}