{"stable_id":"02134d08a5e8cfdc","slug":"allegheny-family-screening","company":"Allegheny County DHS","workflow_name":"AFST risk-assisted investigation screening","function_code":"healthcare_screening","pattern_codes":["exception_based_operations","continuous_decisioning","creator_to_judge"],"changed_assumption":"Administrative data can standardize triage only with override and disparity review.","evidence_strength":"verified","publication_tier":"showcase","freshness":"current","reviewed_at":"2026-08-23","updated_at":"2026-08-22","source_quality_summary":"1 independent, 1 primary; publication outcomes are verified.","caveat_summary":"Outcome summary links an independent evaluation; fairness concerns remain contested.","freshness_caveat":null,"workflow_summary":{"before":[{"actor":"Child-welfare hotline screener","action":"Receives an eligible General Protective Service referral and reviews allegations and available county records.","actor_type":"control"},{"actor":"Screener and supervisor","action":"Decide whether to open an investigation without an algorithmic risk score.","actor_type":"control"}],"hinge":"Administrative data can standardize triage only with override and disparity review.","after":[{"actor":"AFST","action":"Uses administrative data to produce a 1–20 future-placement risk score for eligible referrals.","actor_type":"system"},{"actor":"Screener and supervisor","action":"Combine score with qualitative context and decide screen-in or screen-out.","actor_type":"system"}],"decision_mode":"shared","decision_marker":"AI handles the default; humans own exceptions"},"before":[{"order":1,"actor":"Child-welfare hotline screener","action":"Receives an eligible General Protective Service referral and reviews allegations and available county records.","handoff_to":"Screening supervisor","control":"Abuse and severe-neglect reports are mandatory screen-ins and are outside model discretion."},{"order":2,"actor":"Screener and supervisor","action":"Decide whether to open an investigation without an algorithmic risk score.","handoff_to":"Investigation or close-at-intake queue","control":"Human policy and contextual judgment; no baseline cycle time is published."}],"after":[{"order":1,"actor":"AFST","action":"Uses administrative data to produce a 1–20 future-placement risk score for eligible referrals.","handoff_to":"Hotline screener","control":"Score is one input; target is placement within two years, not a direct prediction of abuse."},{"order":2,"actor":"Screener and supervisor","action":"Combine score with qualitative context and decide screen-in or screen-out.","handoff_to":"Investigation or close-at-intake queue","control":"Humans may override; high-score override requires documented explanation."}],"decision_rights":"Humans make final screening decisions; mandatory abuse referrals bypass the model.","exception_path":"Overrides are allowed and high-score overrides require explanation.","removed_work":["Manual cross-agency history aggregation"],"outcomes":[{"metric":"Screening accuracy and disparity","baseline":"Pre-AFST decisions","result":"Improved identification and modest reduction in racial disparity in case openings","period":"Through March 2019","scale":"Eligible GPS hotline referrals","attribution_caveat":"Outcome summary links an independent evaluation; fairness concerns remain contested.","evidence_label":"verified"}],"executive_lesson":"Public risk scores require contestable targets and override logs.","anti_pattern":"Do not call the score an abuse prediction.","questions_for_leaders":["Where is the operating threshold set and who can override it?","What measured result would trigger rollback or retraining?","Which residual decisions must remain human-owned?"],"collections":["human-still-decides","regulated-autonomy"],"bundle_version":"1.0.0","bundle_fingerprint":"sha256:c23c6cc2b88153d008ea8fda928f632ce0011fc2d4c5036672a16e5d895bab93","canonical_url":"https://brianletort.ai/transformations/allegheny-family-screening"}