[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"detail-sidebar-cat-0-en-105":3,"doc-seo-360058-105":59,"doc-detail-360058-en":130},{"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":123,"head_meta":125,"extra_data":127,"updated_unix":129},105,"en","multimodal-models-for-skin-cancer-classification-using-clinical-freetext-and-dermatoscopic-images","Multimodal models for skin cancer classification using clinical freetext and dermatoscopic images","","Skin cancer is a leading global cancer, and early detection is essential to reduce mortality. This study evaluates whether clinical free-text notes add diagnostic information beyond dermatoscopic images alone. A multimodal dataset (dermatoscopy images, patient metadata, and clinical notes) is built and multimodal ML models are trained. Results show unfiltered text increases performance versus image-only models, with persistent gains after removing leading-language signals, while filtering reduces bias and preserves accuracy.",{"@graph":69,"@context":122},[70,84,105],{"@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/multimodal-models-for-skin-cancer-classification-using-clinical-freetext-and-dermatoscopic-images/360058/",{"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/multimodal-models-for-skin-cancer-classification-using-clinical-freetext-and-dermatoscopic-images/360058.png","ImageObject",300,407,{"name":92,"@type":93},"Putri","Person",{"url":74,"name":95,"@type":96},"DocShare","Organization","application/pdf","2026-10-30","2026-09-23",true,{"@type":102,"interactionType":103,"userInteractionCount":52},"InteractionCounter",{"@type":104},"ViewAction",{"@type":106,"mainEntity":107},"FAQPage",[108,114,118],{"name":109,"@type":110,"acceptedAnswer":111},"Why are clinical free-text notes useful for skin cancer classification?","Question",{"text":112,"@type":113},"Key clinical information such as lesion change, itching, or bleeding is often recorded only in clinical notes. The study tests whether this free-text adds predictive signal beyond dermatoscopic images.","Answer",{"name":115,"@type":110,"acceptedAnswer":116},"What dataset and labels were used in the multimodal approach?",{"text":117,"@type":113},"The work constructs a multimodal dataset with dermatoscopic images from thousands of patients, including patient metadata and clinical notes, using binary labels (benign vs. malignant).",{"name":119,"@type":110,"acceptedAnswer":120},"How does filtering clinical text affect model performance and bias?",{"text":121,"@type":113},"Incorporating unfiltered text improves classification performance compared with using visual data alone. After removing leading-language effects, gains persist, and systematic filtering reduces bias while retaining improved accuracy.","https://schema.org",{"og:url":83,"og:type":124,"og:title":65,"og:site_name":95,"og:description":67},"article",{"robots":126,"canonical":83},"index,follow",{"doc_id":128,"site_id":62},360058,1790804476,{"code":4,"msg":5,"data":131},{"doc_id":128,"user_id":132,"nickname":92,"user_avatar":133,"doc_module":4,"category_id":39,"category_name":40,"doc_title":65,"doc_description":67,"doc_content":134,"file_id":135,"file_url":136,"file_type":137,"file_size":138,"view_count":52,"is_deleted":4,"is_public":8,"is_downloadable":8,"audit_status":8,"page_count":139,"language":140,"language_code":63,"site_id":62,"html_lang":63,"table_of_contents":141,"faqs":142,"seo_title":143,"seo_description":67,"update_tm":144,"read_time":145,"preview_page_count":19,"preview_urls":146},962085571259,"https://ap-avatar.wpscdn.com/davatar_29158cc5080c5b710cf443261637dec0","A Nature Portfolio journal  \n[https://doi.org/10.1038/s43856-026-01456-2](https://doi.org/10.1038/s43856-026-01456-2)  \n\n| Multimodal models for skin cancer classiﬁcation using clinical freetext and dermatoscopic images\u003Cbr> Check for updates |  |  |\n| --- | --- | --- |\n| Matthew Watson 1,2 , Thomas Winterbottom1,2, Thomas Hudson1,2, Benedict Jones1, Hubert P. H. Shum2, Amir Atapour-Abarghouei2, Toby Breckon 2, James Harmsworth King1,2 & Noura Al Moubayed1,2  |  |  |\n| Abstract |  | Plain language summary |\n| Background Skin cancer is one of the most prevalent cancers globally, with early detection critical to ensure reduced mortality risk. To aid early detection, machine learning (ML) skin cancer detection models have been proposed, currently with a focus on dermatoscopic imaging only. However, freetext may provide extra diagnostic information that is not present in images alone.\u003Cbr>Methods We constructed a multimodal dataset comprising 5481 dermatoscopic images from 4538 patients, including patient metadata and clinical notes, with binary labels (benign vs. malignant, 7% malignant) . To assess and mitigate bias from leading language, we developed a clinical text preprocessing pipeline combining regular expressions and large language models, enabling multiple levels ofﬁltering. We train multimodal ML modelson this dataset to explore the effect of freetext on model performance.\u003Cbr>Results Our results show that incorporating unﬁltered text signiﬁcantly improves classiﬁcation performance (0.970 AUROC) compared to visual data alone (0.909 AUROC); even with leading language removed, performance gains persist (0.948 AUROC) . Conclusions This work benchmarks clinical freetext inclusion in skin lesion classiﬁcation, demonstrating that clinical text contributes predictive value beyond that available in images alone. The model’s high performance on unﬁltered clinical text highlights the high levels of bias, and possible shortcutting, present in this text which may make it unsuitable for inclusion in some ML models. By systematically ﬁltering clinical notes via our proposed technique, we show that multimodal models retain improved accuracy while reducing bias. These results provide practical guidance for integrating clinical text into real-world skin cancer detection systems and establish a foundation for future multimodal research in dermatology. |  | Prompt detection of skin cancer improves survival, but diagnosis must be made by clinicians. Image-based machine learning models for skin cancer classiﬁcation have shown promise. However, key information is often only recorded in clinical notes, such as whether a lesion has changed, itches, orbleeds. By creating a dataset that contains images, patient data, and freetext descriptions of the problem, we train a series of machine learning modelson both images and freetext to predict skin cancer. We show that the inclusion of freetext signiﬁcantly enhances model performance, but that care must be taken to ensure the freetext does not unintentionally bias the model. These models could be used in multiple points in a skin cancer clinical workﬂow to either support more accurate referrals to dermatology, or direct patient access to dermatology services, potentially reducing wait times and improving patient outcomes. |\n| Skin cancer is oneof the most prevalent cancers globally and, despite efforts towards improving prevention, its incidence continues to increase1. However, the 5-year net survival rate for early stage detection of skin cancer, particularly melanoma, is extremely high relative to other cancers, reaching up to 99.6% for stage 1 melanoma2 (although known health inequalities in dermatology exist3). Effective screening via self-examination helps the early detection of skin cancer that might otherwise result in mortality4. Skin selfexamination is technically straightforward: an unaided visual inspection of | the lesion. If the patient identiﬁes any concerning features, they are encouraged","cbCaiflePKu7Xft8","https://ap.wps.com/l/cbCaiflePKu7Xft8","pdf",2162943,14,"English","# Abstract\n## Background\n## Methods\n## Results\n## Conclusions\n# Introduction\n## Motivation for early detection and referral efficiency\n## Role of clinical notes and multimodal learning\n## Related work and datasets","[{\"question\":\"Why are clinical free-text notes useful for skin cancer classification?\",\"answer\":\"Key clinical information such as lesion change, itching, or bleeding is often recorded only in clinical notes. The study tests whether this free-text adds predictive signal beyond dermatoscopic images.\"},{\"question\":\"What dataset and labels were used in the multimodal approach?\",\"answer\":\"The work constructs a multimodal dataset with dermatoscopic images from thousands of patients, including patient metadata and clinical notes, using binary labels (benign vs. malignant).\"},{\"question\":\"How does filtering clinical text affect model performance and bias?\",\"answer\":\"Incorporating unfiltered text improves classification performance compared with using visual data alone. After removing leading-language effects, gains persist, and systematic filtering reduces bias while retaining improved accuracy.\"}]","Multimodal models for skin cancer classification using clinical freetext and dermatoscopic images | PDF",1790138633,35,[147,151,155,159],{"page_no":8,"url":148,"width":149,"height":150},"https://res-academy.cache.wpscdn.com/upload/kdocshare/2026-11-16/06a82586506c765884d8723c5f963e7d.webp",983,1305,{"page_no":14,"url":152,"width":153,"height":154},"https://res-academy.cache.wpscdn.com/upload/kdocshare/2026-11-16/f96b5df0a85aaef7b551a9fa2844464b.webp",982,1304,{"page_no":81,"url":156,"width":157,"height":158},"https://res-academy.cache.wpscdn.com/upload/kdocshare/2026-11-16/9240ce83ec58385a820cabb134bc57f6.webp",981,1303,{"page_no":19,"url":160,"width":161,"height":162},"https://res-academy.cache.wpscdn.com/upload/kdocshare/2026-11-16/d8992ea558087627eba391ade5049eb2.webp",980,1301]