tests/test_risk_models/test_tyrer_cuzick_model.py

"""Tests for the Tyrer-Cuzick (IBIS) breast cancer risk model.

This test suite validates the Tyrer-Cuzick model implementation against reference values
from the IBIS web calculator. Test cases cover various scenarios including:
- Different personal risk factor combinations
- Various family history patterns
- Edge cases and boundary conditions

Test cases should be populated with ground truth values from the IBIS web calculator:
https://www.ems-trials.org/riskevaluator/
"""

import pytest

from sentinel.risk_models.tyrer_cuzick import (
    BRCA1_CUMULATIVE_RISK_BY_AGE,
    BRCA2_CUMULATIVE_RISK_BY_AGE,
    TyrerCuzickRiskModel,
    build_brca_survivor,
    build_population_survivor,
    build_s0_survivor,
    compute_personal_relative_risk,
    relative_risk_bmi_post_menopausal,
    relative_risk_first_birth,
    relative_risk_height,
    relative_risk_menarche,
)
from sentinel.user_input import (
    Anthropometrics,
    BreastHealthHistory,
    CancerType,
    Demographics,
    FamilyMemberCancer,
    FamilyRelation,
    FamilySide,
    FemaleSpecific,
    Lifestyle,
    MenstrualHistory,
    ParityHistory,
    PersonalMedicalHistory,
    RelationshipDegree,
    Sex,
    SmokingHistory,
    SmokingStatus,
    UserInput,
)


def create_test_user(
    age: int = 40,
    menarche_age: int | None = 13,
    has_given_birth: bool = False,
    age_first_birth: int | None = None,
    num_live_births: int | None = None,
    is_postmenopausal: bool = False,
    menopause_age: int | None = None,
    height_m: float | None = None,
    bmi: float | None = None,
    atypical_hyperplasia: bool = False,
    lcis: bool = False,
    polygenic_relative_risk: float | None = None,
    family_history: list[FamilyMemberCancer] | None = None,
) -> UserInput:
    """Helper to create UserInput for testing.

    Args:
        age: Patient age.
        menarche_age: Age at menarche.
        has_given_birth: Whether patient has given birth.
        age_first_birth: Age at first birth.
        num_live_births: Number of live births.
        is_postmenopausal: Whether patient is postmenopausal.
        menopause_age: Age at menopause.
        height_m: Height in meters.
        bmi: Body mass index.
        atypical_hyperplasia: Whether patient has atypical hyperplasia.
        lcis: Whether patient has LCIS.
        polygenic_relative_risk: Polygenic relative risk multiplier.
        family_history: List of family cancer history.

    Returns:
        UserInput: Configured user input for testing.
    """
    # Calculate height and weight from height_m and bmi if provided
    height_cm = height_m * 100 if height_m is not None else 165.0
    weight_kg = (
        bmi * (height_m**2) if bmi is not None and height_m is not None else 70.0
    )

    return UserInput(
        demographics=Demographics(
            age_years=age,
            sex=Sex.FEMALE,
            anthropometrics=Anthropometrics(
                height_cm=height_cm,
                weight_kg=weight_kg,
            ),
        ),
        female_specific=FemaleSpecific(
            menstrual=MenstrualHistory(
                age_at_menarche=menarche_age,
                age_at_menopause=menopause_age if is_postmenopausal else None,
            ),
            parity=ParityHistory(
                num_live_births=num_live_births or (1 if has_given_birth else None),
                age_at_first_live_birth=age_first_birth,
            ),
            breast_health=BreastHealthHistory(
                atypical_hyperplasia=atypical_hyperplasia,
                lobular_carcinoma_in_situ=lcis,
            ),
        ),
        lifestyle=Lifestyle(
            smoking=SmokingHistory(status=SmokingStatus.NEVER),
        ),
        personal_medical_history=PersonalMedicalHistory(
            tyrer_cuzick_polygenic_risk_score=polygenic_relative_risk,
        ),
        family_history=family_history or [],
    )


class TestPersonalRiskFactors:
    """Test personal risk factor calculations."""

    def test_relative_risk_menarche_baseline(self):
        """Test menarche RR at baseline age 13."""
        assert relative_risk_menarche(13) == pytest.approx(1.0, rel=1e-6)

    def test_relative_risk_menarche_early(self):
        """Test menarche RR for early age (11)."""
        expected = 0.95 ** (11 - 13)
        assert relative_risk_menarche(11) == pytest.approx(expected, rel=1e-6)

    def test_relative_risk_menarche_late(self):
        """Test menarche RR for late age (15)."""
        expected = 0.95 ** (15 - 13)
        assert relative_risk_menarche(15) == pytest.approx(expected, rel=1e-6)

    def test_relative_risk_first_birth_nulliparous(self):
        """Test first birth RR for nulliparous women."""
        assert relative_risk_first_birth("nulliparous", None) == pytest.approx(1.0)

    def test_relative_risk_first_birth_early(self):
        """Test first birth RR for age < 20."""
        assert relative_risk_first_birth("parous", 18) == pytest.approx(0.67)

    def test_relative_risk_first_birth_20_24(self):
        """Test first birth RR for age 20-24."""
        assert relative_risk_first_birth("parous", 22) == pytest.approx(0.74)

    def test_relative_risk_first_birth_25_29(self):
        """Test first birth RR for age 25-29."""
        assert relative_risk_first_birth("parous", 27) == pytest.approx(0.88)

    def test_relative_risk_first_birth_30_plus(self):
        """Test first birth RR for age >= 30."""
        assert relative_risk_first_birth("parous", 32) == pytest.approx(1.04)

    def test_relative_risk_height_low(self):
        """Test height RR for height < 1.60m."""
        assert relative_risk_height(1.55) == pytest.approx(1.0)

    def test_relative_risk_height_medium(self):
        """Test height RR for height 1.60-1.70m."""
        assert relative_risk_height(1.65) == pytest.approx(1.15)

    def test_relative_risk_height_high(self):
        """Test height RR for height >= 1.70m."""
        assert relative_risk_height(1.75) == pytest.approx(1.24)

    def test_relative_risk_bmi_post_menopausal_low(self):
        """Test BMI RR for BMI < 21 (post-menopausal)."""
        assert relative_risk_bmi_post_menopausal(20, "post") == pytest.approx(1.0)

    def test_relative_risk_bmi_post_menopausal_21_23(self):
        """Test BMI RR for BMI 21-23 (post-menopausal)."""
        assert relative_risk_bmi_post_menopausal(22, "post") == pytest.approx(1.14)

    def test_relative_risk_bmi_post_menopausal_high(self):
        """Test BMI RR for BMI >= 27 (post-menopausal)."""
        assert relative_risk_bmi_post_menopausal(28, "post") == pytest.approx(1.32)

    def test_rr_bmi_premenopausal(self):
        """Test BMI RR for pre-menopausal (should be 1.0)."""
        assert relative_risk_bmi_post_menopausal(28, "pre") == pytest.approx(1.0)

    def test_compute_personal_relative_risk_baseline(self):
        """Test combined personal RR with baseline values."""
        rr = compute_personal_relative_risk(
            menarche_age=13,
            has_given_birth=False,
            age_first_birth=None,
            menopausal_status="pre",
            menopause_age=None,
            height_m=None,
            bmi=None,
            atypical_hyperplasia=False,
            lcis=False,
            polygenic_relative_risk=None,
        )
        assert rr > 0.0


class TestSurvivorFunctions:
    """Test survivor function computations."""

    def test_build_population_survivor(self):
        """Test population survivor function construction."""
        s_pop = build_population_survivor()
        assert len(s_pop) == 13
        assert s_pop[0][2] == pytest.approx(1.0)
        for i in range(len(s_pop) - 1):
            assert s_pop[i][2] >= s_pop[i + 1][2]

    def test_build_brca1_survivor(self):
        """Test BRCA1 survivor function construction."""
        s_brca1 = build_brca_survivor(BRCA1_CUMULATIVE_RISK_BY_AGE)
        assert len(s_brca1) > 0
        assert s_brca1[0][2] <= 1.0

    def test_build_brca2_survivor(self):
        """Test BRCA2 survivor function construction."""
        s_brca2 = build_brca_survivor(BRCA2_CUMULATIVE_RISK_BY_AGE)
        assert len(s_brca2) > 0
        assert s_brca2[0][2] <= 1.0

    def test_build_s0_survivor(self):
        """Test S_0 (no BRCA, no LPG) survivor function construction."""
        s_pop = build_population_survivor()
        s_brca1 = build_brca_survivor(BRCA1_CUMULATIVE_RISK_BY_AGE)
        s_brca2 = build_brca_survivor(BRCA2_CUMULATIVE_RISK_BY_AGE)
        s0 = build_s0_survivor(s_pop, s_brca1, s_brca2)

        assert len(s0) == len(s_pop)
        for _, _, survival in s0:
            assert 0.0 <= survival <= 1.0


class TestTyrerCuzickModel:
    """Test Tyrer-Cuzick model calculations."""

    def test_model_initialization(self):
        """Test model initialization."""
        model = TyrerCuzickRiskModel()
        assert model.name == "tyrer_cuzick"
        assert model.cancer_type() == "breast"

    def test_calculate_risk_baseline(self):
        """Test risk calculation with baseline inputs."""
        model = TyrerCuzickRiskModel()
        user = create_test_user(age=45, menarche_age=13)
        result = model.calculate_risk(user, projection_years=10)

        assert "cumulative_risk" in result
        assert "interval_risks" in result
        assert "personal_relative_risk" in result
        assert "phenotype_probs" in result

        assert 0.0 <= result["cumulative_risk"] <= 1.0
        assert sum(result["phenotype_probs"]) == pytest.approx(1.0, rel=1e-6)

    def test_calculate_risk_with_family_history(self):
        """Test risk calculation with family history."""
        model = TyrerCuzickRiskModel()

        # Mother with breast cancer at age 45
        family_history = [
            FamilyMemberCancer(
                relation=FamilyRelation.MOTHER,
                cancer_type=CancerType.BREAST,
                age_at_diagnosis=45,
                degree=RelationshipDegree.FIRST,
                side=FamilySide.MATERNAL,
            )
        ]

        user = create_test_user(
            age=40,
            menarche_age=12,
            has_given_birth=True,
            age_first_birth=28,
            family_history=family_history,
        )

        result = model.calculate_risk(user, projection_years=10)
        assert 0.0 <= result["cumulative_risk"] <= 1.0

    def test_calculate_risk_high_risk_factors(self):
        """Test risk calculation with multiple high-risk factors."""
        model = TyrerCuzickRiskModel()

        user = create_test_user(
            age=50,
            menarche_age=11,
            has_given_birth=False,
            is_postmenopausal=True,
            menopause_age=55,
            height_m=1.75,
            bmi=28,
            atypical_hyperplasia=True,
        )

        result = model.calculate_risk(user, projection_years=10)
        assert result["personal_relative_risk"] > 2.0

    def test_polygenic_relative_risk_increases_risk(self):
        """Test that polygenic RR multiplies risk appropriately."""
        model = TyrerCuzickRiskModel()

        user_baseline = create_test_user(age=50)
        result_baseline = model.calculate_risk(user_baseline, projection_years=10)

        user_high_prs = create_test_user(age=50, polygenic_relative_risk=2.0)
        result_high_prs = model.calculate_risk(user_high_prs, projection_years=10)

        user_low_prs = create_test_user(age=50, polygenic_relative_risk=0.5)
        result_low_prs = model.calculate_risk(user_low_prs, projection_years=10)

        assert result_high_prs["personal_relative_risk"] == pytest.approx(
            result_baseline["personal_relative_risk"] * 2.0
        )
        assert result_low_prs["personal_relative_risk"] == pytest.approx(
            result_baseline["personal_relative_risk"] * 0.5
        )

        assert result_high_prs["cumulative_risk"] > result_baseline["cumulative_risk"]
        assert result_low_prs["cumulative_risk"] < result_baseline["cumulative_risk"]

    def test_polygenic_relative_risk_with_other_factors(self):
        """Test that polygenic RR combines multiplicatively with other risk factors."""
        model = TyrerCuzickRiskModel()

        user_personal = create_test_user(
            age=50,
            menarche_age=11,
            has_given_birth=True,
            age_first_birth=35,
        )
        result_personal = model.calculate_risk(user_personal, projection_years=10)

        user_combined = create_test_user(
            age=50,
            menarche_age=11,
            has_given_birth=True,
            age_first_birth=35,
            polygenic_relative_risk=1.5,
        )
        result_combined = model.calculate_risk(user_combined, projection_years=10)

        assert result_combined["personal_relative_risk"] == pytest.approx(
            result_personal["personal_relative_risk"] * 1.5
        )
        assert result_combined["cumulative_risk"] > result_personal["cumulative_risk"]

    def test_polygenic_relative_risk_boundary_values(self):
        """Test polygenic RR with boundary values."""
        model = TyrerCuzickRiskModel()

        user_min = create_test_user(age=50, polygenic_relative_risk=0.1)
        result_min = model.calculate_risk(user_min, projection_years=10)
        assert result_min["cumulative_risk"] > 0

        user_max = create_test_user(age=50, polygenic_relative_risk=10.0)
        result_max = model.calculate_risk(user_max, projection_years=10)
        assert result_max["cumulative_risk"] < 1.0

        assert result_max["cumulative_risk"] > result_min["cumulative_risk"] * 5

    def test_model_description(self):
        """Test model description methods."""
        model = TyrerCuzickRiskModel()
        assert len(model.description()) > 0
        assert len(model.interpretation()) > 0
        assert len(model.references()) > 0


class TestReferenceCalculations:
    """Test cases from IBIS web calculator validated against https://ibis.ikonopedia.com/"""

    def test_reference_case_1_baseline_low_risk(self):
        """Baseline low risk with average factors and no family history."""
        model = TyrerCuzickRiskModel()
        user = create_test_user(
            age=40,
            height_m=1.65,
            bmi=25.7,
            menarche_age=13,
            has_given_birth=True,
            age_first_birth=25,
        )
        result = model.calculate_risk(user, projection_years=10)

        expected_10yr_risk = 0.015
        assert result["cumulative_risk"] == pytest.approx(expected_10yr_risk, abs=0.001)

    def test_reference_case_2_high_personal_risk(self):
        """High personal risk with early menarche, nulliparous, and late menopause."""
        model = TyrerCuzickRiskModel()
        user = create_test_user(
            age=60,
            height_m=1.75,
            bmi=27.8,
            menarche_age=11,
            is_postmenopausal=True,
            menopause_age=55,
            has_given_birth=False,
        )
        result = model.calculate_risk(user, projection_years=10)

        expected_10yr_risk = 0.053
        assert result["cumulative_risk"] == pytest.approx(expected_10yr_risk, abs=0.015)

    def test_reference_case_3_strong_family_history(self):
        """Family history with mother diagnosed at age 42."""
        model = TyrerCuzickRiskModel()

        family_history = [
            FamilyMemberCancer(
                relation=FamilyRelation.MOTHER,
                cancer_type=CancerType.BREAST,
                age_at_diagnosis=42,
                degree=RelationshipDegree.FIRST,
                side=FamilySide.MATERNAL,
            ),
        ]

        user = create_test_user(
            age=35,
            height_m=1.60,
            bmi=25.4,
            menarche_age=12,
            has_given_birth=False,
            family_history=family_history,
        )
        result = model.calculate_risk(user, projection_years=10)

        expected_10yr_risk = 0.026
        assert result["cumulative_risk"] == pytest.approx(expected_10yr_risk, abs=0.015)

    def test_reference_case_4_moderate_family_history(self):
        """Moderate family history with mother and maternal aunt."""
        model = TyrerCuzickRiskModel()

        family_history = [
            FamilyMemberCancer(
                relation=FamilyRelation.MOTHER,
                cancer_type=CancerType.BREAST,
                age_at_diagnosis=50,
                degree=RelationshipDegree.FIRST,
                side=FamilySide.MATERNAL,
            ),
            FamilyMemberCancer(
                relation=FamilyRelation.MATERNAL_AUNT,
                cancer_type=CancerType.BREAST,
                age_at_diagnosis=55,
                degree=RelationshipDegree.SECOND,
                side=FamilySide.MATERNAL,
            ),
        ]

        user = create_test_user(
            age=45,
            height_m=1.68,
            bmi=25.5,
            menarche_age=13,
            has_given_birth=True,
            age_first_birth=30,
            family_history=family_history,
        )
        result = model.calculate_risk(user, projection_years=10)

        expected_10yr_risk = 0.029
        assert result["cumulative_risk"] == pytest.approx(expected_10yr_risk, abs=0.015)

    def test_reference_case_5_young_with_early_onset_family_history(self):
        """Early onset family history with mother at 38 and maternal grandmother at 52."""
        model = TyrerCuzickRiskModel()

        family_history = [
            FamilyMemberCancer(
                relation=FamilyRelation.MOTHER,
                cancer_type=CancerType.BREAST,
                age_at_diagnosis=38,  # Early onset
                degree=RelationshipDegree.FIRST,
                side=FamilySide.MATERNAL,
            ),
            FamilyMemberCancer(
                relation=FamilyRelation.MATERNAL_GRANDMOTHER,
                cancer_type=CancerType.BREAST,
                age_at_diagnosis=52,
                degree=RelationshipDegree.SECOND,
                side=FamilySide.MATERNAL,
            ),
        ]

        user = create_test_user(
            age=30,
            height_m=1.62,
            bmi=22.1,
            menarche_age=12,
            has_given_birth=True,
            age_first_birth=28,
            family_history=family_history,
        )
        result = model.calculate_risk(user, projection_years=10)

        expected_10yr_risk = 0.024
        assert result["cumulative_risk"] == pytest.approx(expected_10yr_risk, abs=0.015)


class TestEdgeCases:
    """Test edge cases and boundary conditions."""

    def test_minimum_age(self):
        """Test calculation at minimum valid age (20)."""
        model = TyrerCuzickRiskModel()
        user = create_test_user(age=20, has_given_birth=False)
        result = model.calculate_risk(user, projection_years=10)
        assert 0.0 <= result["cumulative_risk"] <= 1.0

    def test_maximum_age(self):
        """Test calculation at maximum valid age (85)."""
        model = TyrerCuzickRiskModel()
        user = create_test_user(
            age=85,
            is_postmenopausal=True,
            menopause_age=52,
            has_given_birth=True,
            age_first_birth=25,
        )
        result = model.calculate_risk(user, projection_years=5)
        assert 0.0 <= result["cumulative_risk"] <= 1.0

    def test_empty_pedigree(self):
        """Test calculation with empty pedigree."""
        model = TyrerCuzickRiskModel()
        user = create_test_user(age=45)
        result = model.calculate_risk(user, projection_years=10)
        # Should use population priors
        assert 0.0 <= result["cumulative_risk"] <= 1.0