{"stable_id":"fbf20734fb447a59","slug":"denmark-mammography-triage","company":"Capital Region of Denmark mammography screening programme","workflow_name":"AI-triaged single versus double reading of screening mammograms (Transpara)","function_code":"healthcare_screening","pattern_codes":["exception_based_operations","decision_right_transfer","creator_to_judge"],"changed_assumption":"That every screening mammogram must be independently double-read by two radiologists to keep cancer detection and recall rates acceptable.","evidence_strength":"verified","publication_tier":"showcase","freshness":"current","reviewed_at":"2026-08-23","updated_at":"2026-08-22","source_quality_summary":"2 independent, 1 peer reviewed; publication outcomes are verified.","caveat_summary":"Observational sequential cohorts, not randomised. Interval cancers were not yet available. Screening interval lengthened after AI (median 845 to 993 days), which the authors stratified and argue does not fully explain the detection gain. Screen-detected cancers only, with ≥180 days follow-up. DCIS share rose (15.1% to 20.4%), so overdiagnosis remains an open question pending interval-cancer follow-up. False-positive rate also fell (2.39% to 1.63%) and PPV rose (22.6% to 33.6%). Not a randomised comparison.","freshness_caveat":null,"workflow_summary":{"before":[{"actor":"Radiographer","action":"Acquires four-view full-field digital mammograms under Danish screening guidelines.","actor_type":"control"},{"actor":"Two independent breast radiologists","action":"Each reads every examination; the second reader is always a senior radiologist.","actor_type":"control"},{"actor":"Consensus meeting or third-reader arbitration","action":"Resolves recall disagreements.","actor_type":"control"}],"hinge":"Every screening mammogram does not need two human reads.","after":[{"actor":"Transpara AI (ScreenPoint Medical v1.7.1)","action":"Assigns an examination malignancy score (1–10) and, for higher-risk exams, lesion marks.","actor_type":"ai"},{"actor":"Senior breast radiologist (single-read lane)","action":"Reads likely-normal examinations knowing they are single-read; may still send a recall to consensus.","actor_type":"human"},{"actor":"Two radiologists with AI decision support.","action":"Independently double-read remaining examinations and consult AI marks after their own evaluation.","actor_type":"ai"}],"decision_mode":"moved","decision_marker":"AI sets reading intensity; radiologists retain recall authority"},"before":[{"order":1,"actor":"Radiographer","action":"Acquires four-view full-field digital mammograms under Danish screening guidelines.","handoff_to":"Two independent breast radiologists","control":"Acquisition protocol; PACS routing"},{"order":2,"actor":"Two independent breast radiologists","action":"Each reads every examination; the second reader is always a senior radiologist.","handoff_to":"Consensus or arbitration if they disagree on recall","control":"Independent double reading; European/Danish screening guidelines"},{"order":3,"actor":"Consensus meeting or third-reader arbitration","action":"Resolves recall disagreements.","handoff_to":"Diagnostic assessment if recalled","control":"Consensus or arbitration"}],"after":[{"order":1,"actor":"Transpara AI (ScreenPoint Medical v1.7.1)","action":"Assigns an examination malignancy score (1–10) and, for higher-risk exams, lesion marks.","handoff_to":"Single senior reader if likely normal; otherwise two readers with AI decision support","control":"Score threshold (≤5 until 2022-05-03, then ≤7) set by the programme with continuous quality monitoring"},{"order":2,"actor":"Senior breast radiologist (single-read lane)","action":"Reads likely-normal examinations knowing they are single-read; may still send a recall to consensus.","handoff_to":"Consensus/arbitration on recall, otherwise no recall","control":"Human recall authority retained; 0.8% of single-read exams were recalled against the AI-normal tag"},{"order":3,"actor":"Two radiologists with AI decision support (higher-score lane)","action":"Independently double-read remaining examinations and consult AI marks after their own evaluation.","handoff_to":"Consensus/arbitration on disagreement","control":"Readers blinded to each other; instructed not to use AI marks as the first look"}],"decision_rights":"The programme sets the score threshold that drops the second human reader. Radiologists retain recall/no-recall authority, including the right to override an AI-normal tag.","exception_path":"Unprocessed exams (1.5%, mostly early infrastructure failures or implants) revert to double reading. Single-read recalls go to consensus or arbitration. Continuous quality control justified raising the threshold from 5 to 7.","removed_work":["Second independent human read on 66.9% of post-implementation screenings (38,977 of 58,246)","Some downstream diagnostic work associated with the 20.5% recall-rate reduction"],"outcomes":[{"metric":"Radiologist screening-read workload","baseline":"Two reads per examination (double reading of the entire cohort)","result":"33.5% fewer reads (38,977 of 116,492 possible reads saved) because 66.9% of exams were single-read","period":"After: 2021-11-18 to 2022-10-17 versus before: 2020-10-01 to 2021-11-17","scale":"60,751 women before AI; 58,246 women after AI; 19 senior high-volume readers","attribution_caveat":"Observational sequential cohorts, not randomised. Interval cancers were not yet available. Screening interval lengthened after AI (median 845 to 993 days), which the authors stratified and argue does not fully explain the detection gain.","evidence_label":"verified"},{"metric":"Cancer detection rate","baseline":"0.70% (423 of 60,751)","result":"0.82% (480 of 58,246); P = .01","period":"Same sequential cohorts","scale":"Same populations","attribution_caveat":"Screen-detected cancers only, with ≥180 days follow-up. DCIS share rose (15.1% to 20.4%), so overdiagnosis remains an open question pending interval-cancer follow-up.","evidence_label":"verified"},{"metric":"Recall rate","baseline":"3.09% (1,875 of 60,751)","result":"2.46% (1,430 of 58,246); 20.5% relative reduction; P < .001","period":"Same sequential cohorts","scale":"Same populations","attribution_caveat":"False-positive rate also fell (2.39% to 1.63%) and PPV rose (22.6% to 33.6%). Not a randomised comparison.","evidence_label":"verified"}],"executive_lesson":"The Capital Region did not give radiologists a faster viewer. It changed who must look. AI became the first sort, senior radiologists became the only readers on the likely-normal majority, and recall stayed human. The measured result is fewer reads, fewer recalls, and more screen-detected cancers in a live national-style programme. That is the MASAI trial's missing piece: standing practice, not a protocol.","anti_pattern":"Citing MASAI or ScreenTrustCAD as if they were BAU operations. Those are trials. This case is the programme that actually changed the reading protocol.","questions_for_leaders":["Where do we still insist on two humans for work a model can already sort by risk?","Who owns the threshold that drops a reviewer, and how often is it re-validated?","What override rate would tell us the single-read lane is unsafe?"],"collections":[],"bundle_version":"1.0.0","bundle_fingerprint":"sha256:c23c6cc2b88153d008ea8fda928f632ce0011fc2d4c5036672a16e5d895bab93","canonical_url":"https://brianletort.ai/transformations/denmark-mammography-triage"}