[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-125683-en":3,"doc-seo-125683-105":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},125683,8796095360427,"Lucas Martin","https://ap-avatar.wpscdn.com/davatar_994ba38a5ba835b3df7d355c54d3ed8d",8,"Research & Report","Confound-leakage: confound removal in machine learning leads to leakage","Machine learning methods are central to modern data analysis, including epidemiology and medicine, where nonlinear models can capture complex feature–target relationships. Confounding information embedded in features can bias models, and featurewise linear confound regression (CR) is often used to remove spurious signal. New evidence shows linear confound regression may increase confounding risk with nonlinear ML, producing near-perfect predictions by confound leakage and overestimated clinical effects.","GigaScience, 2023, 12, 1–14 DOI: 10.1093/gigascience/giad071  \nResearch  \nConfound-leakage: confound removal in machine learning leads to leakage  \nSami Hamdan 1,2 , Bradley C. Love3,4,5 , Georg G. von Polier 1,6,7 , Susanne Weis 1,2 , Holger Schwender8 , Simon B. Eickhoff1,2 and Kaustubh R. Patil 1,2 , *  \n1 Institute of Neuroscience and Medicine, Brain and Behaviour (INM-7), Forschungszentrum Jülich, 52428 Jülich, Germany  \n2 Institute of Systems Neuroscience, Medical Faculty, Heinrich-Heine University Düsseldorf, 40225 Düsseldorf, Germany  \n3 Department of Experimental Psychology, University College London, WC1H 0AP London, UK  \n4The Alan Turing Institute, London NW1 2DB, UK  \n5 European Lab for Learning & Intelligent Systems (ELLIS), WC1E 6BT, London, UK  \n6 Department of Child and Adolescent Psychiatry, Psychosomatics and Psychotherapy, University Hospital Frankfurt, 60528 Frankfurt, Germany  \n7 Department of Child and Adolescent Psychiatry, Psychosomatics and Psychotherapy, RWTH Aachen University, 52074 Aachen, Germany  \n8 Institute of Mathematics, Heinrich-Heine University Düsseldorf, 40225 Düsseldorf, Germany  \n∗ Correspondence address. Kaustubh R. Patil. Institute of Neuroscience and Medicine, Brain and Behaviour (INM-7), Forschungszentrum Jülich, Jülich, 52428 [Germany. E-mail:](Germany. E-mail: k.patil@fz-juelich.de)[ k.patil@fz-juelich.de](Germany. E-mail: k.patil@fz-juelich.de)  \nAbstract  \nBackground: Machine learning (ML) approaches are a crucial component of modern data analysis in many fields, including epidemiology and medicine. Nonlinear ML methods often achieve accurate predictions, for instance, in personalized medicine, as they are capable of modeling complex relationships between features and the target. Problematically, ML models and their predictions can be biased by confounding information present in the features. To remove this spurious signal, researchers often employ featurewise linear confound regression (CR). While this is considered a standard approach for dealing with confounding, possible pitfalls of using CR in ML pipelines are not fully understood.  \nResults: We provide new evidence that, contrary to general expectations, linear confound regression can increase the risk of confounding when combined with nonlinear ML approaches. Using a simple framework that uses the target as a confound, we show that information leaked via CR can increase null or moderate effects to near-perfect prediction. By shuffling the features, we provide evidence that this increase is indeed due to confound-leakage and not due to revealing of information. We then demonstrate the danger of confound-leakage in a real-world clinical application where the accuracy of predicting attention-deficit/hyperactivity disorder is overestimated using speech-derived features when using depression as a confound.  \nConclusions: Mishandling or even amplifying confounding effects when building ML models due to confound-leakage, as shown, can lead to untrustworthy, biased, and unfair predictions. Our expose of the confound-leakage pitfall and provided guidelines for dealing with it can help create more robust and trustworthy ML models.  \nKeywords: confounding, data-leakage, machine-learning, clinical applications  \nKey Points:  \n􀀂 Confound removal is essential for building insightful and trustworthy machine learning (ML) models.  \n􀀂 Confound removal can increase performance when combined with nonlinear ML.  \n􀀂 This can be due to confound information leaking into the features.  \n􀀂 Possible reasons are skewed feature distributions and the feature of limited precision.  \n􀀂 Confound removal should be applied with utmost care in combination with nonlinear ML.  \nIntroduction  \nMachine learning (ML) approaches have revolutionized biomedical data analysis by providing powerful tools, especially nonlin-  \near models, that can model complex feature–target relationships [1 , 2] . However, the very power these nonlinear models bring to data anar mod","cbCaimZK7cQrhhsW","https://ap.wps.com/l/cbCaimZK7cQrhhsW","pdf",1877576,1,14,"English","en",105,"# Abstract\n## Background\n## Results\n## Conclusions\n# Key Points\n# Introduction","[{\"question\":\"Why can confound removal lead to leakage in machine learning?\",\"answer\":\"When linear confound regression (CR) is combined with nonlinear ML, information related to the confound can leak into the processed features, increasing confounding rather than removing it.\"},{\"question\":\"How does the study show that the effect is due to confound leakage?\",\"answer\":\"Using a simple framework where the target is treated as a confound, the study demonstrates that shuffled features still show the performance increase, indicating the rise is driven by leakage of confound information.\"},{\"question\":\"What clinical risk does the paper highlight?\",\"answer\":\"In a real-world clinical application, the accuracy of predicting ADHD using speech-derived features becomes overestimated when depression is used as a confound.\"}]","Confound-leakage: confound removal in machine learning leads to leakage | PDF",1785900644,35,{"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},"confound-leakage-confound-removal-in-machine-learning-leads-to-leakage","",{"@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/confound-leakage-confound-removal-in-machine-learning-leads-to-leakage/125683/",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","2026-08-05",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 can confound removal lead to leakage in machine learning?","Question",{"text":75,"@type":76},"When linear confound regression (CR) is combined with nonlinear ML, information related to the confound can leak into the processed features, increasing confounding rather than removing it.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"How does the study show that the effect is due to confound leakage?",{"text":80,"@type":76},"Using a simple framework where the target is treated as a confound, the study demonstrates that shuffled features still show the performance increase, indicating the rise is driven by leakage of confound information.",{"name":82,"@type":73,"acceptedAnswer":83},"What clinical risk does the paper highlight?",{"text":84,"@type":76},"In a real-world clinical application, the accuracy of predicting ADHD using speech-derived features becomes overestimated when depression is used as a confound.","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},[93,97,101,105,110,115,120,123,128,131,135],{"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":121,"slug":122},30,"research-report",{"id":124,"doc_module":4,"doc_module_name":46,"category_name":125,"show_sort_weight":126,"slug":127},9,"Religion & Spirituality",20,"religion-spirituality",{"id":126,"doc_module":4,"doc_module_name":46,"category_name":129,"show_sort_weight":126,"slug":130},"World Cup","world-cup",{"id":132,"doc_module":4,"doc_module_name":46,"category_name":133,"show_sort_weight":132,"slug":134},10,"Lifestyle","lifestyle",{"id":136,"doc_module":4,"doc_module_name":46,"category_name":137,"show_sort_weight":106,"slug":138},19,"General","general"]