[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-117749-en":3,"doc-seo-117749-105":30,"detail-sidebar-cat-0-en-105":91},{"code":4,"msg":5,"data":6},0,"success",{"doc_id":7,"user_id":8,"nickname":9,"user_avatar":10,"doc_module":4,"category_id":11,"category_name":12,"doc_title":13,"doc_description":14,"doc_content":15,"file_id":16,"file_url":17,"file_type":18,"file_size":19,"view_count":4,"is_deleted":4,"is_public":20,"is_downloadable":20,"audit_status":20,"page_count":21,"language":22,"language_code":23,"site_id":24,"html_lang":23,"table_of_contents":25,"faqs":26,"seo_title":27,"seo_description":14,"update_tm":28,"read_time":29},117749,1099513958607,"Jiven","https://ap-avatar.wpscdn.com/avatar/100002390cf8733938c?x-image-process=image/resize,m_fixed,w_180,h_180&k=1778829742770036399",8,"Research & Report","Adversarial De-confounding in Individualised Treatment Effects Estimation","Observational studies attract increasing attention in machine learning because nonexperimental data are widely available while randomized trials can be costly, slow, sometimes unethical, and limited in size. In this setting, de-confounding is essential for estimating individualised treatment effects (ITE) under selection bias. The approach introduces disentangled representations learned with adversarial training to selectively balance confounders in binary treatment. Treatment-policy adversarial learning encourages treatment-agnostic balanced confounder representations, enabling counterfactual inference. Experiments on synthetic and real datasets across confounding levels show improved state-of-the-art ITE accuracy.","Adversarial De-confounding in Individualised Treatment Effects Estimation  \nVinod Kumar Chauhan 1 Soheila Molaei 1 Marzia Hoque Tania 1  \nAnshul Thakur 1 and Tingting Zhu 1 and David A. Clifton 1 ;2  \n1Institute of Biomedical Engineering, University of Oxford, UK  \n2 Oxford-Suzhou Centre for Advanced Research, Suzhou, China  \narXiv :2210 . 10530v3 [ cs .LG] 24 Jan 2023  \nAbstract  \nObservational studies have recently received significant attention from the machine learning community due to the increasingly available nonexperimental observational data and the limitations of the experimental studies, such as considerable cost, impracticality, small and less representative sample sizes, etc. In observational studies, de-confounding is a fundamental problem of individualised treatment effects (ITE) estimation. This paper proposes disentangled representations with adversarial training to selectively balance the confounders in the binary treatment setting for the ITE estimation. The adversarial training of treatment policy selectively encourages treatment-agnostic balanced representations for the confounders and helps to estimate the ITE in the observational studies via counterfactual inference. Empirical results on synthetic and real-world datasets, with varying degrees of confounding, prove that our proposed approach improves the state-of-the-art methods in achieving lower error in the ITE estimation.  \n1 Introduction  \nIndividualised treatment effects (ITE) estimation is a fundamental problem that is useful for making personalised decisions and estimating their effects. For example, in the intensive care unit (ICU), ITE can be used to decide whether or not to give a medication to a patient, which can be a question of life or death of the patient. The ITE estimation learning requires answering counterfactual questions, such as: “What would have been the outcome if alternative treatment had been given?”, i.e., it requires predicting  \npotential outcomes of unexplored actions (Rubin (2005)) . Due to its importance, ITE estimation is studied widely across diverse fields, like medicine, marketing, education and policy-making, etc. (please refer to Bica et al. (2021) for an overview) .  \nRandomised controlled trial (RCT) experiments are the gold standard to evaluate the effectiveness of treatments (Pearl (2009)) . However, they are expensive, timeconsuming, sometimes unethical and impractical, and have small and less representative sample sizes, etc. On the other hand, non-experimental observational studies are becoming popular for evaluating the effectiveness of treatments due to increasingly available observational data, like electronic health records, and overcoming limitations of the RCT.  \nThe ITE estimation from the observational studies have, recently, received a great attention from the machine learning community, e.g., Shalit et al. (2017); Shi et al. (2019); Hassanpour and Greiner (2019b); Curth and van der Schaar (2021a,b); Wu et al. (2022) etc., and it is different from standard machine learning (Rubin (2005); Pearl (2009); Wager and Athey (2018)) . This is because the observational data have outcomes available only for the actions taken, i.e., for the selected treatments (factual outcomes), but the outcomes for alternative treatments are not available (counterfactual outcomes)– which is called the fundamental problem of causal learning (Holland (1986)) . Moreover, the observational data contain confounders, i.e., covariates which affect outcomes as well as treatment assignment policy, and hence have selection-bias (Imbens and Rubin (2015)) (i.e., p (T = 0jX = x)  p (T = 1jX = x), unlike the RCT where treatments are assigned randomly and has p(T = 0jX = x) = p (T = 1jX = x), where p (T = tjX = x) is probability of treatment t for a given patient x in a binary treatment setting) . For example, in the above ICU setting scenario, suppose choice of medication is dependent on the age of the patient which also affects the recovery ","cbCaigO6DJGX4F7M","https://ap.wps.com/l/cbCaigO6DJGX4F7M","pdf",545223,1,12,"English","en",105,"# Abstract\n# Introduction\n## ITE estimation and the counterfactual problem\n## De-confounding in observational studies\n## Representation, domain adaptation, and disentanglement approaches\n# Context and problem setting","[{\"question\":\"Why is de-confounding critical for ITE estimation in observational studies?\",\"answer\":\"Observational data include selection bias because treatment assignment depends on covariates that also influence outcomes. Without de-confounding, comparisons across treatment groups are confounded and causal effect estimates become biased.\"},{\"question\":\"How does the proposed method use adversarial training?\",\"answer\":\"It learns disentangled representations with adversarial training so the confounder-related representation becomes treatment-agnostic and balanced. This selective balancing supports counterfactual inference for ITE estimation.\"},{\"question\":\"What evidence supports the effectiveness of the approach?\",\"answer\":\"Results on synthetic and real-world datasets with varying degrees of confounding show lower error than state-of-the-art ITE estimation methods.\"}]","Adversarial De-confounding in Individualised Treatment Effects Estimation | PDF",1785679337,30,{"code":4,"msg":31,"data":32},"ok",{"site_id":24,"language":23,"slug":33,"title":13,"keywords":34,"description":14,"schema_data":35,"social_meta":86,"head_meta":88,"extra_data":90,"updated_unix":28},"adversarial-de-confounding-in-individualised-treatment-effects-estimation","",{"@graph":36,"@context":85},[37,54,68],{"@type":38,"itemListElement":39},"BreadcrumbList",[40,44,48,51],{"item":41,"name":42,"@type":43,"position":20},"https://docshare.wps.com","Home","ListItem",{"item":45,"name":46,"@type":43,"position":47},"https://docshare.wps.com/document/","Document",2,{"item":49,"name":12,"@type":43,"position":50},"https://docshare.wps.com/document/research-report/",3,{"item":52,"name":13,"@type":43,"position":53},"https://docshare.wps.com/document/adversarial-de-confounding-in-individualised-treatment-effects-estimation/117749/",4,{"url":52,"name":13,"@type":55,"author":56,"headline":13,"publisher":58,"fileFormat":61,"inLanguage":23,"description":14,"dateModified":62,"datePublished":62,"encodingFormat":61,"isAccessibleForFree":63,"interactionStatistic":64},"DigitalDocument",{"name":9,"@type":57},"Person",{"url":41,"name":59,"@type":60},"DocShare","Organization","application/pdf","2026-08-02",true,{"@type":65,"interactionType":66,"userInteractionCount":4},"InteractionCounter",{"@type":67},"ViewAction",{"@type":69,"mainEntity":70},"FAQPage",[71,77,81],{"name":72,"@type":73,"acceptedAnswer":74},"Why is de-confounding critical for ITE estimation in observational studies?","Question",{"text":75,"@type":76},"Observational data include selection bias because treatment assignment depends on covariates that also influence outcomes. Without de-confounding, comparisons across treatment groups are confounded and causal effect estimates become biased.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"How does the proposed method use adversarial training?",{"text":80,"@type":76},"It learns disentangled representations with adversarial training so the confounder-related representation becomes treatment-agnostic and balanced. This selective balancing supports counterfactual inference for ITE estimation.",{"name":82,"@type":73,"acceptedAnswer":83},"What evidence supports the effectiveness of the approach?",{"text":84,"@type":76},"Results on synthetic and real-world datasets with varying degrees of confounding show lower error than state-of-the-art ITE estimation methods.","https://schema.org",{"og:url":52,"og:type":87,"og:title":13,"og:site_name":59,"og:description":14},"article",{"robots":89,"canonical":52},"index,follow",{"doc_id":7,"site_id":24},{"code":4,"msg":5,"data":92},[93,97,101,105,110,115,120,122,127,130,134],{"id":20,"doc_module":4,"doc_module_name":46,"category_name":94,"show_sort_weight":95,"slug":96},"Story & Novel",90,"story-novel",{"id":47,"doc_module":4,"doc_module_name":46,"category_name":98,"show_sort_weight":99,"slug":100},"Literature",80,"literature",{"id":53,"doc_module":4,"doc_module_name":46,"category_name":102,"show_sort_weight":103,"slug":104},"Exam",70,"exam",{"id":106,"doc_module":4,"doc_module_name":46,"category_name":107,"show_sort_weight":108,"slug":109},5,"Comic",60,"comic",{"id":111,"doc_module":4,"doc_module_name":46,"category_name":112,"show_sort_weight":113,"slug":114},6,"Technology",50,"technology",{"id":116,"doc_module":4,"doc_module_name":46,"category_name":117,"show_sort_weight":118,"slug":119},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":46,"category_name":12,"show_sort_weight":29,"slug":121},"research-report",{"id":123,"doc_module":4,"doc_module_name":46,"category_name":124,"show_sort_weight":125,"slug":126},9,"Religion & Spirituality",20,"religion-spirituality",{"id":125,"doc_module":4,"doc_module_name":46,"category_name":128,"show_sort_weight":125,"slug":129},"World Cup","world-cup",{"id":131,"doc_module":4,"doc_module_name":46,"category_name":132,"show_sort_weight":131,"slug":133},10,"Lifestyle","lifestyle",{"id":135,"doc_module":4,"doc_module_name":46,"category_name":136,"show_sort_weight":106,"slug":137},19,"General","general"]