[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"detail-sidebar-cat-0-en-105":3,"doc-seo-290553-105":59,"doc-detail-290553-en":135},{"code":4,"msg":5,"data":6},0,"success",[7,13,18,23,28,33,38,43,48,51,55],{"id":8,"doc_module":4,"doc_module_name":9,"category_name":10,"show_sort_weight":11,"slug":12},1,"Document","Story & Novel",90,"story-novel",{"id":14,"doc_module":4,"doc_module_name":9,"category_name":15,"show_sort_weight":16,"slug":17},2,"Literature",80,"literature",{"id":19,"doc_module":4,"doc_module_name":9,"category_name":20,"show_sort_weight":21,"slug":22},4,"Exam",70,"exam",{"id":24,"doc_module":4,"doc_module_name":9,"category_name":25,"show_sort_weight":26,"slug":27},5,"Comic",60,"comic",{"id":29,"doc_module":4,"doc_module_name":9,"category_name":30,"show_sort_weight":31,"slug":32},6,"Technology",50,"technology",{"id":34,"doc_module":4,"doc_module_name":9,"category_name":35,"show_sort_weight":36,"slug":37},7,"Healthcare",40,"healthcare",{"id":39,"doc_module":4,"doc_module_name":9,"category_name":40,"show_sort_weight":41,"slug":42},8,"Research & Report",30,"research-report",{"id":44,"doc_module":4,"doc_module_name":9,"category_name":45,"show_sort_weight":46,"slug":47},9,"Religion & Spirituality",20,"religion-spirituality",{"id":46,"doc_module":4,"doc_module_name":9,"category_name":49,"show_sort_weight":46,"slug":50},"World Cup","world-cup",{"id":52,"doc_module":4,"doc_module_name":9,"category_name":53,"show_sort_weight":52,"slug":54},10,"Lifestyle","lifestyle",{"id":56,"doc_module":4,"doc_module_name":9,"category_name":57,"show_sort_weight":24,"slug":58},19,"General","general",{"code":4,"msg":60,"data":61},"ok",{"site_id":62,"language":63,"slug":64,"title":65,"keywords":66,"description":67,"schema_data":68,"social_meta":128,"head_meta":130,"extra_data":132,"updated_unix":134},105,"en","advancing-fairness-and-equity-in-machine-learning-for-clinical-research","Advancing Fairness and Equity in Machine Learning for Clinical Research","","Advancements in machine learning can transform clinical research by integrating new data sources, improving efficiency, and modelling complex signals, yet fairness and equity remain threatened by bias throughout study design, data collection, analysis, and interpretation. The thesis addresses systematic trial barriers, limitations of survival and dose-finding methods, and underuse of real-world evidence for causal inference. It develops an ethics-to-fairness framework, trialscope-x/xl pipelines with LLM-based EHR curation and conditional survival with interpretable hazard ratios, safe-t for early-phase dose finding with safety guarantees, and an equity-focused survival benchmarking framework using calibration metrics for heterogeneity and temporal variation.",{"@graph":69,"@context":127},[70,84,106],{"@type":71,"itemListElement":72},"BreadcrumbList",[73,77,79,82],{"item":74,"name":75,"@type":76,"position":8},"https://docshare.wps.com","Home","ListItem",{"item":78,"name":9,"@type":76,"position":14},"https://docshare.wps.com/document/",{"item":80,"name":40,"@type":76,"position":81},"https://docshare.wps.com/document/research-report/",3,{"item":83,"name":65,"@type":76,"position":19},"https://docshare.wps.com/document/advancing-fairness-and-equity-in-machine-learning-for-clinical-research/290553/",{"url":83,"name":65,"@type":85,"image":86,"author":91,"headline":65,"publisher":94,"fileFormat":97,"inLanguage":63,"description":67,"dateModified":98,"datePublished":99,"encodingFormat":97,"isAccessibleForFree":100,"interactionStatistic":101},"DigitalDocument",{"url":87,"@type":88,"width":89,"height":90},"https://docshare.wps.com/thumbnails/advancing-fairness-and-equity-in-machine-learning-for-clinical-research/290553.png","ImageObject",300,407,{"name":92,"@type":93},"Noah","Person",{"url":74,"name":95,"@type":96},"DocShare","Organization","application/pdf","2026-10-31","2026-09-17",true,{"@type":102,"interactionType":103,"userInteractionCount":105},"InteractionCounter",{"@type":104},"ViewAction",12,{"@type":107,"mainEntity":108},"FAQPage",[109,115,119,123],{"name":110,"@type":111,"acceptedAnswer":112},"What fairness and equity problems does the thesis target in clinical research?","Question",{"text":113,"@type":114},"It targets biases across the full research pipeline, including study design, data collection, analysis, and interpretation, with attention to inequities across diverse populations.","Answer",{"name":116,"@type":111,"acceptedAnswer":117},"How does the thesis use real-world evidence to study treatment effects?",{"text":118,"@type":114},"It introduces trialscope-x and trialscope-xl, which use LLM-based automated EHR curation and conditional survival models to estimate causally interpretable hazard ratios while addressing confounding, selection bias, and informative censoring.",{"name":120,"@type":111,"acceptedAnswer":121},"What is safe-t and how does it improve early-phase dose finding?",{"text":122,"@type":114},"safe-t is an adaptive method for early-phase dose-finding that uses multi-output Gaussian processes with Bayesian optimisation and provides formal safety guarantees balancing patient safety and utility over heterogeneous populations.",{"name":124,"@type":111,"acceptedAnswer":125},"How does the thesis evaluate survival analysis methods for fairness?",{"text":126,"@type":114},"It proposes a comprehensive benchmarking framework that evaluates both performance and fairness, including calibration metrics that assess effects of population heterogeneity and temporal variation and expose gaps under assumption violations and observational biases.","https://schema.org",{"og:url":83,"og:type":129,"og:title":65,"og:site_name":95,"og:description":67},"article",{"robots":131,"canonical":83},"index,follow",{"doc_id":133,"site_id":62},290553,1789642060,{"code":4,"msg":5,"data":136},{"doc_id":133,"user_id":137,"nickname":92,"user_avatar":138,"doc_module":4,"category_id":39,"category_name":40,"doc_title":65,"doc_description":67,"doc_content":139,"file_id":140,"file_url":141,"file_type":142,"file_size":143,"view_count":105,"is_deleted":4,"is_public":8,"is_downloadable":8,"audit_status":8,"page_count":144,"language":145,"language_code":63,"site_id":62,"html_lang":63,"table_of_contents":146,"faqs":147,"seo_title":148,"seo_description":67,"update_tm":134,"read_time":149,"preview_page_count":4},8796095462418,"https://ap-avatar.wpscdn.com/avatar/80000253c1241d02b47?x-image-process=image/resize,m_fixed,w_180,h_180&k=1778826106357471780","Advancing Fairness and Equity in Machine Learning for Clinical Research  \nIsabel Chien  \nChrist’s College, University of Cambridge December 2024  \nThis thesis is submitted for the degree of Doctor of Philosophy  \nIsabel Chien: Advancing Fairness and Equity  \nin Machine Learning for Clinical Research, Methods for Treatment Effect Heterogeneity, Real-World Evidence and Safe Exploration, © December 2024  \nThis thesis is the result of my own work and includes nothing which is the outcome of work done in collaboration except as declared in the preface and specified in the text. It is not substantially the same as any work that has already been submitted, or, is being concurrently submitted, for any degree, diploma or other qualification at the University of Cambridge or any other University or similar institution except as declared in the preface and specified in the text. It does not exceed the prescribed word limit for the relevant Degree Committee.  \nIsabel Chien  \nDecember 2024  \nAdvancements in machine learning have the potential to revolutionise clinical research, including clinical trials and observational studies, by enabling the integration of novel data sources, increasing study efficiency, and improving modelling of complex data. However, clinical research remains vulnerable to biases at all stages of the research process—from study design to data collection, analysis, and interpretation—raising critical concerns regarding fairness and equity. Current clinical research faces critical limitations across multiple domains. Systematic barriers in clinical trials prevent equitable outcomes across diverse populations, and researchers lack principled frameworks for integrating machine learning approaches with fairness principles and clinical research ethics. Widely-used survival analysis methods like the Cox proportional hazards model suffer from survivorship bias, restrictive assumptions, and inability to capture time-varying treatment effects across heterogeneous populations. Early-phase dose-finding studies assume population homogeneity and focus solely on toxicity whilst ignoring efficacy and patient diversity. Finally, existing benchmarking practices fail to evaluate methods under realistic bias conditions where equity issues actually arise, whilst real-world evidence remains underutilised due to challenges in causal inference from observational data. This thesis addresses these challenges through a multifaceted approach that introduces specific technical innovations whilst maintaining focus on equity-driven solutions. We begin by examining possible sources of unfairness in clinical trials and identifying opportunities where machine learning can mitigate these biases through careful, context-specific approaches, developing the first framework mapping clinical research ethics principles to algorithmic fairness constraints with implementable guidance. We then explore how real-world evidence can supplement trial findings and be utilised to investigate heterogeneous, time-varying treatment effects through trialscope-x and trialscope-xl. These machine learning pipelines leverage large language models for automated EHR curation and employ novel importance weighting methodology to simultaneously handle confounding, selection bias, and informative censoring whilst using conditional survival  \nmodels to estimate causally interpretable hazard ratios (HRbaseline and HRsurviving) . This approach reveals insights beyond traditional clinical trials whilst uncovering potential disparities in treatment outcomes. To address inequities that may arise early in the clinical research process, we propose safe-t, an adaptive method for early-phase dose[finding studies.](finding studies. safe-t uses multi-output Gaussian processes with Bayesian optimisation)[ safe](finding studies. safe-t uses multi-output Gaussian processes with Bayesian optimisation)[-](finding studies. safe-t uses multi-output Gaussian processes with Bayesian optimisation)[t](","cbCaipTBCLM9vl4f","https://ap.wps.com/l/cbCaipTBCLM9vl4f","pdf",14492480,328,"English","# Introduction and Motivation\n## Sources of Unfairness in Clinical Research\n## Limitations of Existing Methods\n# Fairness-Driven Methodology\n## Mapping Clinical Ethics to Algorithmic Fairness Constraints\n# Real-World Evidence for Heterogeneous Effects\n## Trialscope-x and Trialscope-xl Pipelines\n## Causal Inference and Causally Interpretable Hazard Ratios\n# Safe Exploration in Early-Phase Dose Finding\n## safe-t with Multi-output Gaussian Processes and Bayesian Optimisation\n## Safety Guarantees Across Heterogeneous Populations\n# Equity-Aware Benchmarking for Survival Analysis\n## Performance and Fairness Calibration Metrics\n## Assessing Assumption Violations and Observational Biases","[{\"question\":\"What fairness and equity problems does the thesis target in clinical research?\",\"answer\":\"It targets biases across the full research pipeline, including study design, data collection, analysis, and interpretation, with attention to inequities across diverse populations.\"},{\"question\":\"How does the thesis use real-world evidence to study treatment effects?\",\"answer\":\"It introduces trialscope-x and trialscope-xl, which use LLM-based automated EHR curation and conditional survival models to estimate causally interpretable hazard ratios while addressing confounding, selection bias, and informative censoring.\"},{\"question\":\"What is safe-t and how does it improve early-phase dose finding?\",\"answer\":\"safe-t is an adaptive method for early-phase dose-finding that uses multi-output Gaussian processes with Bayesian optimisation and provides formal safety guarantees balancing patient safety and utility over heterogeneous populations.\"},{\"question\":\"How does the thesis evaluate survival analysis methods for fairness?\",\"answer\":\"It proposes a comprehensive benchmarking framework that evaluates both performance and fairness, including calibration metrics that assess effects of population heterogeneity and temporal variation and expose gaps under assumption violations and observational biases.\"}]","Advancing Fairness and Equity in Machine Learning for Clinical Research | PDF",827]