{"entity":{"id":"idea-prev-pathology-ai-borderline-anchor","kind":"idea","name":"AI second reads to stop borderline lesions being upgraded to cancer","aka":[],"tldr":"Whether a lesion is called precancer or cancer varies between pathologists, and over time the bar has drifted lower. AI reference reads could hold the line.","summary":"Inter-observer disagreement is high for DCIS versus atypia, Gleason pattern 3 versus 4, and melanocytic lesions; diagnostic drift inflates incidence. Propose AI reference classifiers calibrated to historical outcome-linked cohorts, used as mandatory second reads for borderline categories, with discordance triggering expert review.","asOf":"2026-09-08","links":[{"label":"Bottleneck evidence (Overdiagnosis and false alarms): Welch & Black, Overdiagnosis in cancer (JNCI 2010)","url":"https://doi.org/10.1093/jnci/djq099"}],"tags":[],"related":[],"cancers":[],"sections":["diagnostics"],"technologies":["digital-pathology-ai","pathology-foundation-model"],"targets":[],"drugs":[],"companies":["paige","pathai"],"institutions":[],"pathways":[],"terms":[],"trials":[],"people":[],"bottlenecks":["b-overdiagnosis","b-ai-validation"],"keyPapers":["paper-welch-j-natl-cancer-inst"],"journals":[],"dependsOn":[],"notes":[],"hypothesis":"AI second reads reduce upgrade rates of borderline lesions by at least 20% and reduce inter-laboratory variation without increasing subsequent invasive cancer.","rationale":"Digital pathology AI already matches expert Gleason grading; anchoring to outcome-linked training sets counters drift.","test":"Multi-laboratory study comparing diagnosis rates with and without AI second read, with five-year outcome linkage.","maturity":"early-clinical","actor":"clinic","cost":"medium","horizonYears":4},"route":"/ideas/idea-prev-pathology-ai-borderline-anchor/","neighbours":{"section":[{"id":"diagnostics","kind":"section","name":"Diagnostics & Biomarkers","route":"/fronts/diagnostics/"}],"technology":[{"id":"digital-pathology-ai","kind":"technology","name":"Digital pathology & AI","route":"/technologies/digital-pathology-ai/"},{"id":"pathology-foundation-model","kind":"technology","name":"Pathology & radiology foundation models","route":"/technologies/pathology-foundation-model/"}],"company":[{"id":"paige","kind":"company","name":"Paige AI","route":"/companies/paige/"},{"id":"pathai","kind":"company","name":"PathAI","route":"/companies/pathai/"}],"bottleneck":[{"id":"b-ai-validation","kind":"bottleneck","name":"AI that is built but not validated or deployed","route":"/bottlenecks/b-ai-validation/"},{"id":"b-overdiagnosis","kind":"bottleneck","name":"Overdiagnosis and false alarms","route":"/bottlenecks/b-overdiagnosis/"}],"paper":[{"id":"paper-welch-j-natl-cancer-inst","kind":"paper","name":"Overdiagnosis in cancer","route":"/key-papers/paper-welch-j-natl-cancer-inst/"}]}}