tests/test_risk_models/test_qcancer_model.py

"""Tests for the QCancer multi-site cancer risk model."""

import csv
from pathlib import Path

import pytest

from sentinel.risk_models import QCancerRiskModel
from sentinel.risk_models.qcancer import (
    compute_female_probabilities,
    compute_male_probabilities,
)
from sentinel.user_input import (
    AlcoholConsumption,
    Anthropometrics,
    Demographics,
    Lifestyle,
    PersonalMedicalHistory,
    Sex,
    SmokingHistory,
    SmokingStatus,
    UserInput,
)

FIXTURE_PATH = Path("tests/fixtures/qcancer_reference.tsv")
FEMALE_INPUT_PATH = Path("tests/fixtures/qcancer_inputs_female.tsv")
MALE_INPUT_PATH = Path("tests/fixtures/qcancer_inputs_male.tsv")


def _load_reference_cases() -> list[dict[str, str]]:
    with FIXTURE_PATH.open("r", encoding="utf-8") as handle:
        return list(csv.DictReader(handle, delimiter="\t"))


def _parse_probability_columns(row: dict[str, str]) -> dict[str, float]:
    result = {}
    for key in row:
        if key in {"case_id", "sex"}:
            continue
        # Keep keys as-is (including "none" from C binary output)
        result[key] = float(row[key])
    return result


REFERENCE_CASES = _load_reference_cases()


class TestQCancerModel:
    """Test suite for QCancer risk model."""

    def setup_method(self) -> None:
        """Set up test fixtures."""
        self.model = QCancerRiskModel()

    @pytest.mark.parametrize("case", REFERENCE_CASES, ids=lambda c: c["case_id"])
    def test_reference_regression(self, case: dict[str, str]) -> None:
        """Test exact implementation against C binary output using TSV inputs.

        Args:
            case: Test case dictionary containing case_id, sex, and expected probabilities.
        """
        expected = _parse_probability_columns(case)
        case_id = case["case_id"]
        sex = case["sex"]

        # Load the corresponding TSV input
        if sex == "female":
            with FEMALE_INPUT_PATH.open("r", encoding="utf-8") as f:
                reader = csv.DictReader(f, delimiter="\t")
                inputs = {row["case_id"]: row for row in reader}

            if case_id not in inputs:
                pytest.skip(f"No input TSV for {case_id}")

            inp = inputs[case_id]
            # Call exact function with TSV parameters
            result = compute_female_probabilities(
                age=int(inp["age"]),
                alcohol_cat4=int(inp["alcohol_cat4"]),
                b_chronicpan=int(inp["b_chronicpan"]),
                b_copd=int(inp["b_copd"]),
                b_endometrial=int(inp["b_endometrial"]),
                b_type2=int(inp["b_type2"]),
                bmi=float(inp["bmi"]),
                c_hb=int(inp["c_hb"]),
                fh_breastcancer=int(inp["fh_breastcancer"]),
                fh_gicancer=int(inp["fh_gicancer"]),
                fh_ovariancancer=int(inp["fh_ovariancancer"]),
                new_abdodist=int(inp["new_abdodist"]),
                new_abdopain=int(inp["new_abdopain"]),
                new_appetiteloss=int(inp["new_appetiteloss"]),
                new_breastlump=int(inp["new_breastlump"]),
                new_breastpain=int(inp["new_breastpain"]),
                new_breastskin=int(inp["new_breastskin"]),
                new_dysphagia=int(inp["new_dysphagia"]),
                new_gibleed=int(inp["new_gibleed"]),
                new_haematuria=int(inp["new_haematuria"]),
                new_haemoptysis=int(inp["new_haemoptysis"]),
                new_heartburn=int(inp["new_heartburn"]),
                new_imb=int(inp["new_imb"]),
                new_indigestion=int(inp["new_indigestion"]),
                new_necklump=int(inp["new_necklump"]),
                new_nightsweats=int(inp["new_nightsweats"]),
                new_pmb=int(inp["new_pmb"]),
                new_postcoital=int(inp["new_postcoital"]),
                new_rectalbleed=int(inp["new_rectalbleed"]),
                new_vte=int(inp["new_vte"]),
                new_weightloss=int(inp["new_weightloss"]),
                s1_bowelchange=int(inp["s1_bowelchange"]),
                s1_bruising=int(inp["s1_bruising"]),
                s1_constipation=int(inp["s1_constipation"]),
                s1_cough=int(inp["s1_cough"]),
                smoke_cat=int(inp["smoke_cat"]),
                town=float(inp["town"]),
            )
        else:  # male
            with MALE_INPUT_PATH.open("r", encoding="utf-8") as f:
                reader = csv.DictReader(f, delimiter="\t")
                inputs = {row["case_id"]: row for row in reader}

            if case_id not in inputs:
                pytest.skip(f"No input TSV for {case_id}")

            inp = inputs[case_id]
            # Call exact function with TSV parameters
            result = compute_male_probabilities(
                age=int(inp["age"]),
                alcohol_cat4=int(inp["alcohol_cat4"]),
                b_chronicpan=int(inp["b_chronicpan"]),
                b_copd=int(inp["b_copd"]),
                b_type2=int(inp["b_type2"]),
                bmi=float(inp["bmi"]),
                c_hb=int(inp["c_hb"]),
                fh_gicancer=int(inp["fh_gicancer"]),
                fh_prostatecancer=int(inp["fh_prostatecancer"]),
                new_abdodist=int(inp["new_abdodist"]),
                new_abdopain=int(inp["new_abdopain"]),
                new_appetiteloss=int(inp["new_appetiteloss"]),
                new_dysphagia=int(inp["new_dysphagia"]),
                new_gibleed=int(inp["new_gibleed"]),
                new_haematuria=int(inp["new_haematuria"]),
                new_haemoptysis=int(inp["new_haemoptysis"]),
                new_heartburn=int(inp["new_heartburn"]),
                new_indigestion=int(inp["new_indigestion"]),
                new_necklump=int(inp["new_necklump"]),
                new_nightsweats=int(inp["new_nightsweats"]),
                new_rectalbleed=int(inp["new_rectalbleed"]),
                new_testespain=int(inp["new_testespain"]),
                new_testicularlump=int(inp["new_testicularlump"]),
                new_vte=int(inp["new_vte"]),
                new_weightloss=int(inp["new_weightloss"]),
                s1_bowelchange=int(inp["s1_bowelchange"]),
                s1_constipation=int(inp["s1_constipation"]),
                s1_cough=int(inp["s1_cough"]),
                s1_impotence=int(inp["s1_impotence"]),
                s1_nocturia=int(inp["s1_nocturia"]),
                s1_urinaryfreq=int(inp["s1_urinaryfreq"]),
                s1_urinaryretention=int(inp["s1_urinaryretention"]),
                smoke_cat=int(inp["smoke_cat"]),
                town=float(inp["town"]),
            )

        # Compare results
        for cancer_site, expected_pct in expected.items():
            observed = result.get(cancer_site, 0.0)
            assert observed == pytest.approx(expected_pct, abs=0.01)

    def test_metadata(self) -> None:
        """Test that model returns correct metadata."""
        assert self.model.name == "qcancer"
        assert self.model.cancer_type() == "multiple"
        assert "QCancer" in self.model.description()

    def test_compute_score_with_user_input(self) -> None:
        """Test that QCancerRiskModel.compute_score works with UserInput."""
        user = UserInput(
            demographics=Demographics(
                age_years=55,
                sex=Sex.FEMALE,
                anthropometrics=Anthropometrics(height_cm=165, weight_kg=70.0),
            ),
            lifestyle=Lifestyle(
                smoking=SmokingHistory(status=SmokingStatus.NEVER),
                alcohol_consumption=AlcoholConsumption.LIGHT,
            ),
            personal_medical_history=PersonalMedicalHistory(),
            family_history=[],
        )
        result = self.model.compute_score(user)
        assert "No Cancer:" in result
        assert "%" in result

    def test_qcancer_with_anaemia_and_endometrial_polyps(self) -> None:
        """Test QCancer processes anaemia and endometrial polyps correctly."""
        from sentinel.user_input import ChronicCondition

        user = UserInput(
            demographics=Demographics(
                age_years=55,
                sex=Sex.FEMALE,
                anthropometrics=Anthropometrics(height_cm=165, weight_kg=70.0),
            ),
            lifestyle=Lifestyle(
                smoking=SmokingHistory(status=SmokingStatus.NEVER),
                alcohol_consumption=AlcoholConsumption.LIGHT,
            ),
            personal_medical_history=PersonalMedicalHistory(
                chronic_conditions=[
                    ChronicCondition.ANAEMIA,
                    ChronicCondition.ENDOMETRIAL_POLYPS,
                ]
            ),
            family_history=[],
        )

        # Should not raise an error and should include these conditions in calculation
        result = self.model.compute_score(user)
        assert "No Cancer:" in result
        assert "%" in result
        # Should have multiple cancer types listed
        assert result.count("%") >= 10

    def test_validate_inputs_valid_user(self) -> None:
        """Test that valid user input passes validation."""
        user = UserInput(
            demographics=Demographics(
                age_years=55,
                sex=Sex.FEMALE,
                anthropometrics=Anthropometrics(height_cm=165, weight_kg=70.0),
            ),
            lifestyle=Lifestyle(
                smoking=SmokingHistory(status=SmokingStatus.NEVER),
            ),
            personal_medical_history=PersonalMedicalHistory(),
            family_history=[],
        )

        is_valid, errors = self.model.validate_inputs(user)
        assert is_valid
        assert len(errors) == 0

    def test_validate_inputs_age_out_of_range(self) -> None:
        """Test that age outside QCancer range is caught."""
        user = UserInput(
            demographics=Demographics(
                age_years=20,  # Too young for QCancer (requires 25-99)
                sex=Sex.FEMALE,
                anthropometrics=Anthropometrics(height_cm=165, weight_kg=70.0),
            ),
            lifestyle=Lifestyle(
                smoking=SmokingHistory(status=SmokingStatus.NEVER),
            ),
            personal_medical_history=PersonalMedicalHistory(),
            family_history=[],
        )

        is_valid, errors = self.model.validate_inputs(user)
        assert not is_valid
        assert any("age_years" in err and "25" in err for err in errors)