12-phase flow (12)

1. 1FRAME
   Confirm preventive breast-screening scope: asymptomatic woman/AFAB, age 40-74, life expectancy ≥10 yr. Excludes diagnostic workup of a palpable mass / nipple discharge / skin change (those = a different, higher prior → diagnostic pathway) and prior-breast-cancer survivorship surveillance (USPSTF 2024 PMID 38687503; ACS 2015 PMID 26501536)
   inputs: age, sex_assigned_at_birth, life_expectancy_estimate, palpable_breast_finding
   advance: Asymptomatic, eligible 40-74, ≥10-yr life expectancy confirmed; no palpable finding (else exit to diagnostic)
2. 2ENTRY
   Trigger: routine wellness visit, overdue/never-screened flag, high-risk-syndrome referral, risk-model expansion, or positive-screen follow-up (USPSTF 2024)
   advance: Entry trigger captured
3. 3CONTEXT
   Establish the pre-test prevalence prior: age, breast density, family breast/ovarian history (FDR site + age of onset), prior mammogram dates/BI-RADS, chest-RT history, pregnancy, prior-breast-cancer status (USPSTF 2024 PMID 38687503; ACS 2015 PMID 26501536). This phase SETS the Bayesian prior — modality choice flows from it
   inputs: age, sex_assigned_at_birth, family_hx_breast_ovarian, prior_screening_results, pregnancy_status, prior_breast_cancer_survivor
   advance: Pre-test prevalence tier assignable (average vs elevated vs very-high)
4. 4RED_FLAGS
   Very-high-prior syndromes: BRCA1/2 (lifetime breast risk ~55-72%), Li-Fraumeni TP53, prior chest radiation age 10-30 → genetics referral + NCCN intensified protocol (annual MRI + mammography from age 25-30); cascade family testing. A NEW palpable mass / bloody nipple discharge / skin change exits to the diagnostic breast engine (NCCN; ACS 2015 PMID 26501536)
   inputs: genetic_syndrome, chest_radiation_history
   actions: workup.genetic_counseling
   advance: High-risk syndromes flagged or excluded; no new diagnostic-grade symptom
5. 5INITIAL_WORKUP
   Order the age/risk-appropriate screen — the test with its fixed characteristics: biennial digital mammography or DBT/tomosynthesis age 40-74 (DBT preferred where available — similar benefit, fewer false-positive recalls; CISNET PMID 38687505 / Svahn PMID 26433751). The order IS the "treatment" (USPSTF 2024 Grade B PMID 38687503)
   actions: workup.breast_screening_mammogram
   advance: Indicated screening mammogram/DBT ordered (or supplemental MRI for high-risk)
6. 6BRANCHING_WORKUP
   BI-RADS-driven Bayesian follow-up — apply PPV to the structured category: 0 → recall for additional views/ultrasound; 1/2 → routine interval (≈0% malignancy); 3 → 6-mo short-interval follow-up (PPV <2%); 4 → image-guided core biopsy (PPV ~2-95%, 4A→4C); 5 → biopsy + treatment planning (PPV ≥95%); 6 → known malignancy (exits screening). Dense breasts + high-risk → supplemental MRI (DENSE interval-cancer 2.5 vs 5.0/1000 PMID 31774954) (ACR BI-RADS; USPSTF 2024 PMID 38687503)
   inputs: breast_density
   actions: workup.breast_biopsy
   advance: Every BI-RADS category has a PPV-appropriate follow-up route
7. 7DIFFERENTIAL
   Average-risk vs high-risk pathway, AND true-positive (early invasive carcinoma / DCIS) vs false-positive (dense-tissue masking artifact, benign fibroadenoma, simple/complex cyst, fat necrosis) — the false-positive recall cascade and overdiagnosis are encoded as harm DATA, not prose (CISNET: 1376 FP recalls + 14 overdiagnosed / 1000 biennial DBT 40-74 PMID 38687505). Screen-detected positive routes to biopsy/dx; screen-negative returns to interval
   advance: Each result classified average/high-risk and true/false-positive likelihood assigned
8. 8RISK_STRATIFICATION
   Risk-model scoring IS the pre-test prior and sets the action threshold: Tyrer-Cuzick/IBIS lifetime ≥20% → add annual MRI (ACS 2015 PMID 26501536; USPSTF 2024 PMID 38687503); Gail/BCRAT 5-yr ≥1.67% → chemoprevention discussion (USPSTF 2019; Cuzick HR 0.62 PMID 23639488; Goss MAP.3 HR 0.35 PMID 21639806). The model output IS the prior — applied once to choose modality/interval, not multiplied on top of the age prior
   inputs: family_hx_breast_ovarian, chest_radiation_history
   actions: calc.tyrer_cuzick, calc.gail_model
   advance: Risk tier + action threshold assigned (modality/interval/chemoprevention)
9. 9TREATMENT
   The screen IS the intervention; document modality (DM vs DBT vs supplemental MRI) + shared decision. Chemoprevention where Gail/BCRAT 5-yr ≥1.67% and shared decision: tamoxifen (premenopausal), raloxifene or aromatase inhibitor (postmenopausal) — weigh VTE/endometrial harm (USPSTF 2019; Cuzick PMID 23639488; Goss MAP.3 PMID 21639806)
   inputs: menopausal_status, vte_history
   advance: Screen ordered, modality + shared decision documented, chemoprevention considered where threshold met
10. 10DISPOSITION
    All screening is ambulatory; positive screens route to the specific diagnostic pathway/specialty: BI-RADS 4-5 → breast surgery/biopsy; biopsy-proven malignancy → breast oncology; confirmed hereditary syndrome → NCCN high-risk protocol + cascade testing (USPSTF 2024; NCCN)
    advance: Disposition + specialist routing documented
11. 11MONITORING
    Interval + recall: biennial mammography/DBT age 40-74 (USPSTF 2024 Grade B PMID 38687503); annual mammography + MRI if Tyrer-Cuzick lifetime ≥20% (ACS 2015 PMID 26501536); BI-RADS structured reporting with category-specific recall (3 → 6-mo short-interval; 0 → prompt additional imaging)
    advance: Next-screen date set + recall scheduled
12. 12FOLLOWUP
    STOP-screening / deprescribing-equivalent: discontinue when life expectancy <10 yr OR age >74 (USPSTF 2024 I-statement — evidence insufficient ≥75; ACS 2015 continue while ≥10-yr life expectancy PMID 26501536). Genetic-counseling referral if criteria met. Chemoprevention follow-up. Patient education on the false-positive recall cascade (CISNET 1376/1000 biennial DBT 40-74) and overdiagnosis (~14/1000) — the harm side of the Bayesian ledger (USPSTF 2024 PMID 38687503; CISNET PMID 38687505)
    inputs: life_expectancy_estimate
    actions: workup.genetic_counseling
    advance: Stop-screening decision OR continued-interval education delivered and documented