@ottolab/extraction 0.1.0 → 0.1.1

package/dist/validator.js

@@ -0,0 +1,191 @@
+/**
+ * Clinical ranges for validation — generous bounds to avoid false rejections.
+ * Values outside these ranges are likely extraction errors.
+ */
+const CLINICAL_RANGES = {
+    weight_kg: [20, 300],
+    height_cm: [100, 250],
+    bmi: [10, 60],
+    body_fat_percent: [2, 60],
+    systolic_bp: [60, 250],
+    diastolic_bp: [30, 150],
+    heart_rate: [30, 220],
+    total_cholesterol: [50, 500],
+    ldl_c: [10, 400],
+    hdl: [5, 150],
+    triglycerides: [10, 2000],
+    apoB: [10, 300],
+    hba1c: [3, 15],
+    fasting_glucose: [30, 500],
+    fasting_insulin: [0.5, 300],
+    uric_acid: [0.5, 20],
+    creatinine: [0.1, 15],
+    bun: [2, 100],
+    egfr: [5, 150],
+    alt: [1, 2000],
+    ast: [1, 2000],
+    alp: [10, 1000],
+    ggt: [1, 2000],
+    bilirubin_total: [0.05, 30],
+    albumin: [1, 6],
+    hs_crp: [0.01, 300],
+    esr: [0, 140],
+    cortisol: [0.5, 60],
+    testosterone: [1, 2000],
+    estradiol: [1, 5000],
+    tsh: [0.01, 100],
+    wbc: [1, 50],
+    rbc: [1, 10],
+    hemoglobin: [3, 25],
+    hematocrit: [15, 65],
+    platelets: [10, 1000],
+    mcv: [50, 130],
+    rdw: [9, 25],
+    lymphocyte_percent: [2, 70],
+    vitamin_d: [3, 200],
+};
+/**
+ * Map from raw extraction keys to BiomarkerSet field names and display metadata.
+ */
+const BIOMARKER_MAP = {
+    weight_kg: { field: 'weight', name: 'Weight', unit: 'kg' },
+    height_cm: { field: 'height', name: 'Height', unit: 'cm' },
+    bmi: { field: 'bmi', name: 'BMI', unit: 'kg/m2' },
+    body_fat_percent: { field: 'bodyFatPercent', name: 'Body Fat', unit: '%' },
+    systolic_bp: { field: 'bloodPressureSystolic', name: 'Systolic BP', unit: 'mmHg' },
+    diastolic_bp: { field: 'bloodPressureDiastolic', name: 'Diastolic BP', unit: 'mmHg' },
+    heart_rate: { field: 'heartRate', name: 'Heart Rate', unit: 'bpm' },
+    total_cholesterol: { field: 'totalCholesterol', name: 'Total Cholesterol', unit: 'mg/dL' },
+    ldl_c: { field: 'ldlC', name: 'LDL-C', unit: 'mg/dL' },
+    hdl: { field: 'hdl', name: 'HDL', unit: 'mg/dL' },
+    triglycerides: { field: 'triglycerides', name: 'Triglycerides', unit: 'mg/dL' },
+    apoB: { field: 'apoB', name: 'ApoB', unit: 'mg/dL' },
+    hba1c: { field: 'hba1c', name: 'HbA1c', unit: '%' },
+    fasting_glucose: { field: 'fastingGlucose', name: 'Fasting Glucose', unit: 'mg/dL' },
+    fasting_insulin: { field: 'fastingInsulin', name: 'Fasting Insulin', unit: 'uIU/mL' },
+    uric_acid: { field: 'uricAcid', name: 'Uric Acid', unit: 'mg/dL' },
+    creatinine: { field: 'creatinine', name: 'Creatinine', unit: 'mg/dL' },
+    bun: { field: 'bun', name: 'BUN', unit: 'mg/dL' },
+    egfr: { field: 'egfr', name: 'eGFR', unit: 'mL/min/1.73m2' },
+    alt: { field: 'alt', name: 'ALT', unit: 'U/L' },
+    ast: { field: 'ast', name: 'AST', unit: 'U/L' },
+    alp: { field: 'alp', name: 'ALP', unit: 'U/L' },
+    ggt: { field: 'ggt', name: 'GGT', unit: 'U/L' },
+    bilirubin_total: { field: 'bilirubinTotal', name: 'Bilirubin (Total)', unit: 'mg/dL' },
+    albumin: { field: 'albumin', name: 'Albumin', unit: 'g/dL' },
+    hs_crp: { field: 'hsCrp', name: 'hs-CRP', unit: 'mg/L' },
+    esr: { field: 'esr', name: 'ESR', unit: 'mm/hr' },
+    cortisol: { field: 'cortisol', name: 'Cortisol', unit: 'ug/dL' },
+    testosterone: { field: 'testosterone', name: 'Testosterone', unit: 'ng/dL' },
+    estradiol: { field: 'estradiol', name: 'Estradiol', unit: 'pg/mL' },
+    tsh: { field: 'tsh', name: 'TSH', unit: 'mIU/L' },
+    wbc: { field: 'wbc', name: 'WBC', unit: '10^3/uL' },
+    rbc: { field: 'rbc', name: 'RBC', unit: '10^6/uL' },
+    hemoglobin: { field: 'hemoglobin', name: 'Hemoglobin', unit: 'g/dL' },
+    hematocrit: { field: 'hematocrit', name: 'Hematocrit', unit: '%' },
+    platelets: { field: 'platelets', name: 'Platelets', unit: '10^3/uL' },
+    mcv: { field: 'mcv', name: 'MCV', unit: 'fL' },
+    rdw: { field: 'rdw', name: 'RDW', unit: '%' },
+    lymphocyte_percent: { field: 'lymphocytePercent', name: 'Lymphocyte %', unit: '%' },
+    vitamin_d: { field: 'vitaminD', name: 'Vitamin D', unit: 'ng/mL' },
+};
+/**
+ * Validate and transform raw LLM extraction output to typed BiomarkerSet.
+ *
+ * Steps:
+ * 1. Filter nulls/undefined
+ * 2. Type-check numeric values
+ * 3. Range-check against generous clinical bounds
+ * 4. Map to BiomarkerSet fields
+ * 5. Compute extraction confidence score
+ */
+export function validateExtraction(raw) {
+    const biomarkers = {};
+    const rejections = [];
+    let accepted = 0;
+    let rejected = 0;
+    for (const [key, value] of Object.entries(raw)) {
+        // Skip gender and non-biomarker fields
+        if (key === 'gender')
+            continue;
+        // Skip null/undefined
+        if (value === null || value === undefined)
+            continue;
+        // Skip string values (qualitative results like "Negative")
+        if (typeof value === 'string')
+            continue;
+        // Must be a number at this point
+        if (typeof value !== 'number' || !Number.isFinite(value)) {
+            rejections.push({ key, value, reason: 'not a finite number' });
+            rejected++;
+            continue;
+        }
+        // Check clinical ranges
+        const range = CLINICAL_RANGES[key];
+        if (range && (value < range[0] || value > range[1])) {
+            rejections.push({
+                key,
+                value,
+                reason: `outside clinical range [${range[0]}, ${range[1]}]`,
+            });
+            rejected++;
+            continue;
+        }
+        // Map to BiomarkerSet
+        const mapping = BIOMARKER_MAP[key];
+        if (!mapping)
+            continue; // skip unknown keys
+        const biomarkerValue = {
+            name: mapping.name,
+            value,
+            unit: mapping.unit,
+        };
+        biomarkers[mapping.field] = biomarkerValue;
+        accepted++;
+    }
+    // Confidence based on yield ratio and cross-checks
+    const confidence = computeConfidence(biomarkers, accepted, rejected);
+    return { biomarkers, accepted, rejected, confidence, rejections };
+}
+/**
+ * Compute extraction confidence score (0-1).
+ *
+ * Factors:
+ * - Yield: how many biomarkers were successfully extracted
+ * - Rejection rate: high rejections lower confidence
+ * - Cross-biomarker consistency checks
+ */
+function computeConfidence(biomarkers, accepted, rejected) {
+    if (accepted === 0)
+        return 0;
+    // Base confidence from yield (diminishing returns above 15 markers)
+    const yieldScore = Math.min(accepted / 15, 1.0);
+    // Penalty for rejections
+    const total = accepted + rejected;
+    const rejectionPenalty = total > 0 ? rejected / total : 0;
+    // Cross-biomarker consistency checks
+    let consistencyScore = 1.0;
+    // Check: if both ALT and AST exist, AST should generally be ≤ 2×ALT in healthy
+    if (biomarkers.alt && biomarkers.ast) {
+        if (biomarkers.ast.value > biomarkers.alt.value * 5) {
+            consistencyScore -= 0.1;
+        }
+    }
+    // Check: HDL should be less than total cholesterol
+    if (biomarkers.hdl && biomarkers.totalCholesterol) {
+        if (biomarkers.hdl.value > biomarkers.totalCholesterol.value) {
+            consistencyScore -= 0.15;
+        }
+    }
+    // Check: LDL should be less than total cholesterol
+    if (biomarkers.ldlC && biomarkers.totalCholesterol) {
+        if (biomarkers.ldlC.value > biomarkers.totalCholesterol.value) {
+            consistencyScore -= 0.15;
+        }
+    }
+    const raw = yieldScore * 0.5 + (1 - rejectionPenalty) * 0.25 + Math.max(consistencyScore, 0) * 0.25;
+    return Math.round(Math.min(Math.max(raw, 0), 1) * 100) / 100;
+}
+// Export for testing
+export { CLINICAL_RANGES, BIOMARKER_MAP };
+//# sourceMappingURL=validator.js.map

package/dist/validator.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"validator.js","sourceRoot":"","sources":["../src/validator.ts"],"names":[],"mappings":"AAGA;;;GAGG;AACH,MAAM,eAAe,GAAqC;IACxD,SAAS,EAAE,CAAC,EAAE,EAAE,GAAG,CAAC;IACpB,SAAS,EAAE,CAAC,GAAG,EAAE,GAAG,CAAC;IACrB,GAAG,EAAE,CAAC,EAAE,EAAE,EAAE,CAAC;IACb,gBAAgB,EAAE,CAAC,CAAC,EAAE,EAAE,CAAC;IACzB,WAAW,EAAE,CAAC,EAAE,EAAE,GAAG,CAAC;IACtB,YAAY,EAAE,CAAC,EAAE,EAAE,GAAG,CAAC;IACvB,UAAU,EAAE,CAAC,EAAE,EAAE,GAAG,CAAC;IACrB,iBAAiB,EAAE,CAAC,EAAE,EAAE,GAAG,CAAC;IAC5B,KAAK,EAAE,CAAC,EAAE,EAAE,GAAG,CAAC;IAChB,GAAG,EAAE,CAAC,CAAC,EAAE,GAAG,CAAC;IACb,aAAa,EAAE,CAAC,EAAE,EAAE,IAAI,CAAC;IACzB,IAAI,EAAE,CAAC,EAAE,EAAE,GAAG,CAAC;IACf,KAAK,EAAE,CAAC,CAAC,EAAE,EAAE,CAAC;IACd,eAAe,EAAE,CAAC,EAAE,EAAE,GAAG,CAAC;IAC1B,eAAe,EAAE,CAAC,GAAG,EAAE,GAAG,CAAC;IAC3B,SAAS,EAAE,CAAC,GAAG,EAAE,EAAE,CAAC;IACpB,UAAU,EAAE,CAAC,GAAG,EAAE,EAAE,CAAC;IACrB,GAAG,EAAE,CAAC,CAAC,EAAE,GAAG,CAAC;IACb,IAAI,EAAE,CAAC,CAAC,EAAE,GAAG,CAAC;IACd,GAAG,EAAE,CAAC,CAAC,EAAE,IAAI,CAAC;IACd,GAAG,EAAE,CAAC,CAAC,EAAE,IAAI,CAAC;IACd,GAAG,EAAE,CAAC,EAAE,EAAE,IAAI,CAAC;IACf,GAAG,EAAE,CAAC,CAAC,EAAE,IAAI,CAAC;IACd,eAAe,EAAE,CAAC,IAAI,EAAE,EAAE,CAAC;IAC3B,OAAO,EAAE,CAAC,CAAC,EAAE,CAAC,CAAC;IACf,MAAM,EAAE,CAAC,IAAI,EAAE,GAAG,CAAC;IACnB,GAAG,EAAE,CAAC,CAAC,EAAE,GAAG,CAAC;IACb,QAAQ,EAAE,CAAC,GAAG,EAAE,EAAE,CAAC;IACnB,YAAY,EAAE,CAAC,CAAC,EAAE,IAAI,CAAC;IACvB,SAAS,EAAE,CAAC,CAAC,EAAE,IAAI,CAAC;IACpB,GAAG,EAAE,CAAC,IAAI,EAAE,GAAG,CAAC;IAChB,GAAG,EAAE,CAAC,CAAC,EAAE,EAAE,CAAC;IACZ,GAAG,EAAE,CAAC,CAAC,EAAE,EAAE,CAAC;IACZ,UAAU,EAAE,CAAC,CAAC,EAAE,EAAE,CAAC;IACnB,UAAU,EAAE,CAAC,EAAE,EAAE,EAAE,CAAC;IACpB,SAAS,EAAE,CAAC,EAAE,EAAE,IAAI,CAAC;IACrB,GAAG,EAAE,CAAC,EAAE,EAAE,GAAG,CAAC;IACd,GAAG,EAAE,CAAC,CAAC,EAAE,EAAE,CAAC;IACZ,kBAAkB,EAAE,CAAC,CAAC,EAAE,EAAE,CAAC;IAC3B,SAAS,EAAE,CAAC,CAAC,EAAE,GAAG,CAAC;CACpB,CAAC;AAEF;;GAEG;AACH,MAAM,aAAa,GAA8E;IAC/F,SAAS,EAAE,EAAE,KAAK,EAAE,QAAQ,EAAE,IAAI,EAAE,QAAQ,EAAE,IAAI,EAAE,IAAI,EAAE;IAC1D,SAAS,EAAE,EAAE,KAAK,EAAE,QAAQ,EAAE,IAAI,EAAE,QAAQ,EAAE,IAAI,EAAE,IAAI,EAAE;IAC1D,GAAG,EAAE,EAAE,KAAK,EAAE,KAAK,EAAE,IAAI,EAAE,KAAK,EAAE,IAAI,EAAE,OAAO,EAAE;IACjD,gBAAgB,EAAE,EAAE,KAAK,EAAE,gBAAgB,EAAE,IAAI,EAAE,UAAU,EAAE,IAAI,EAAE,GAAG,EAAE;IAC1E,WAAW,EAAE,EAAE,KAAK,EAAE,uBAAuB,EAAE,IAAI,EAAE,aAAa,EAAE,IAAI,EAAE,MAAM,EAAE;IAClF,YAAY,EAAE,EAAE,KAAK,EAAE,wBAAwB,EAAE,IAAI,EAAE,cAAc,EAAE,IAAI,EAAE,MAAM,EAAE;IACrF,UAAU,EAAE,EAAE,KAAK,EAAE,WAAW,EAAE,IAAI,EAAE,YAAY,EAAE,IAAI,EAAE,KAAK,EAAE;IACnE,iBAAiB,EAAE,EAAE,KAAK,EAAE,kBAAkB,EAAE,IAAI,EAAE,mBAAmB,EAAE,IAAI,EAAE,OAAO,EAAE;IAC1F,KAAK,EAAE,EAAE,KAAK,EAAE,MAAM,EAAE,IAAI,EAAE,OAAO,EAAE,IAAI,EAAE,OAAO,EAAE;IACtD,GAAG,EAAE,EAAE,KAAK,EAAE,KAAK,EAAE,IAAI,EAAE,KAAK,EAAE,IAAI,EAAE,OAAO,EAAE;IACjD,aAAa,EAAE,EAAE,KAAK,EAAE,eAAe,EAAE,IAAI,EAAE,eAAe,EAAE,IAAI,EAAE,OAAO,EAAE;IAC/E,IAAI,EAAE,EAAE,KAAK,EAAE,MAAM,EAAE,IAAI,EAAE,MAAM,EAAE,IAAI,EAAE,OAAO,EAAE;IACpD,KAAK,EAAE,EAAE,KAAK,EAAE,OAAO,EAAE,IAAI,EAAE,OAAO,EAAE,IAAI,EAAE,GAAG,EAAE;IACnD,eAAe,EAAE,EAAE,KAAK,EAAE,gBAAgB,EAAE,IAAI,EAAE,iBAAiB,EAAE,IAAI,EAAE,OAAO,EAAE;IACpF,eAAe,EAAE,EAAE,KAAK,EAAE,gBAAgB,EAAE,IAAI,EAAE,iBAAiB,EAAE,IAAI,EAAE,QAAQ,EAAE;IACrF,SAAS,EAAE,EAAE,KAAK,EAAE,UAAU,EAAE,IAAI,EAAE,WAAW,EAAE,IAAI,EAAE,OAAO,EAAE;IAClE,UAAU,EAAE,EAAE,KAAK,EAAE,YAAY,EAAE,IAAI,EAAE,YAAY,EAAE,IAAI,EAAE,OAAO,EAAE;IACtE,GAAG,EAAE,EAAE,KAAK,EAAE,KAAK,EAAE,IAAI,EAAE,KAAK,EAAE,IAAI,EAAE,OAAO,EAAE;IACjD,IAAI,EAAE,EAAE,KAAK,EAAE,MAAM,EAAE,IAAI,EAAE,MAAM,EAAE,IAAI,EAAE,eAAe,EAAE;IAC5D,GAAG,EAAE,EAAE,KAAK,EAAE,KAAK,EAAE,IAAI,EAAE,KAAK,EAAE,IAAI,EAAE,KAAK,EAAE;IAC/C,GAAG,EAAE,EAAE,KAAK,EAAE,KAAK,EAAE,IAAI,EAAE,KAAK,EAAE,IAAI,EAAE,KAAK,EAAE;IAC/C,GAAG,EAAE,EAAE,KAAK,EAAE,KAAK,EAAE,IAAI,EAAE,KAAK,EAAE,IAAI,EAAE,KAAK,EAAE;IAC/C,GAAG,EAAE,EAAE,KAAK,EAAE,KAAK,EAAE,IAAI,EAAE,KAAK,EAAE,IAAI,EAAE,KAAK,EAAE;IAC/C,eAAe,EAAE,EAAE,KAAK,EAAE,gBAAgB,EAAE,IAAI,EAAE,mBAAmB,EAAE,IAAI,EAAE,OAAO,EAAE;IACtF,OAAO,EAAE,EAAE,KAAK,EAAE,SAAS,EAAE,IAAI,EAAE,SAAS,EAAE,IAAI,EAAE,MAAM,EAAE;IAC5D,MAAM,EAAE,EAAE,KAAK,EAAE,OAAO,EAAE,IAAI,EAAE,QAAQ,EAAE,IAAI,EAAE,MAAM,EAAE;IACxD,GAAG,EAAE,EAAE,KAAK,EAAE,KAAK,EAAE,IAAI,EAAE,KAAK,EAAE,IAAI,EAAE,OAAO,EAAE;IACjD,QAAQ,EAAE,EAAE,KAAK,EAAE,UAAU,EAAE,IAAI,EAAE,UAAU,EAAE,IAAI,EAAE,OAAO,EAAE;IAChE,YAAY,EAAE,EAAE,KAAK,EAAE,cAAc,EAAE,IAAI,EAAE,cAAc,EAAE,IAAI,EAAE,OAAO,EAAE;IAC5E,SAAS,EAAE,EAAE,KAAK,EAAE,WAAW,EAAE,IAAI,EAAE,WAAW,EAAE,IAAI,EAAE,OAAO,EAAE;IACnE,GAAG,EAAE,EAAE,KAAK,EAAE,KAAK,EAAE,IAAI,EAAE,KAAK,EAAE,IAAI,EAAE,OAAO,EAAE;IACjD,GAAG,EAAE,EAAE,KAAK,EAAE,KAAK,EAAE,IAAI,EAAE,KAAK,EAAE,IAAI,EAAE,SAAS,EAAE;IACnD,GAAG,EAAE,EAAE,KAAK,EAAE,KAAK,EAAE,IAAI,EAAE,KAAK,EAAE,IAAI,EAAE,SAAS,EAAE;IACnD,UAAU,EAAE,EAAE,KAAK,EAAE,YAAY,EAAE,IAAI,EAAE,YAAY,EAAE,IAAI,EAAE,MAAM,EAAE;IACrE,UAAU,EAAE,EAAE,KAAK,EAAE,YAAY,EAAE,IAAI,EAAE,YAAY,EAAE,IAAI,EAAE,GAAG,EAAE;IAClE,SAAS,EAAE,EAAE,KAAK,EAAE,WAAW,EAAE,IAAI,EAAE,WAAW,EAAE,IAAI,EAAE,SAAS,EAAE;IACrE,GAAG,EAAE,EAAE,KAAK,EAAE,KAAK,EAAE,IAAI,EAAE,KAAK,EAAE,IAAI,EAAE,IAAI,EAAE;IAC9C,GAAG,EAAE,EAAE,KAAK,EAAE,KAAK,EAAE,IAAI,EAAE,KAAK,EAAE,IAAI,EAAE,GAAG,EAAE;IAC7C,kBAAkB,EAAE,EAAE,KAAK,EAAE,mBAAmB,EAAE,IAAI,EAAE,cAAc,EAAE,IAAI,EAAE,GAAG,EAAE;IACnF,SAAS,EAAE,EAAE,KAAK,EAAE,UAAU,EAAE,IAAI,EAAE,WAAW,EAAE,IAAI,EAAE,OAAO,EAAE;CACnE,CAAC;AAUF;;;;;;;;;GASG;AACH,MAAM,UAAU,kBAAkB,CAAC,GAAkB;IACnD,MAAM,UAAU,GAAiB,EAAE,CAAC;IACpC,MAAM,UAAU,GAAsD,EAAE,CAAC;IACzE,IAAI,QAAQ,GAAG,CAAC,CAAC;IACjB,IAAI,QAAQ,GAAG,CAAC,CAAC;IAEjB,KAAK,MAAM,CAAC,GAAG,EAAE,KAAK,CAAC,IAAI,MAAM,CAAC,OAAO,CAAC,GAAG,CAAC,EAAE,CAAC;QAC/C,uCAAuC;QACvC,IAAI,GAAG,KAAK,QAAQ;YAAE,SAAS;QAE/B,sBAAsB;QACtB,IAAI,KAAK,KAAK,IAAI,IAAI,KAAK,KAAK,SAAS;YAAE,SAAS;QAEpD,2DAA2D;QAC3D,IAAI,OAAO,KAAK,KAAK,QAAQ;YAAE,SAAS;QAExC,iCAAiC;QACjC,IAAI,OAAO,KAAK,KAAK,QAAQ,IAAI,CAAC,MAAM,CAAC,QAAQ,CAAC,KAAK,CAAC,EAAE,CAAC;YACzD,UAAU,CAAC,IAAI,CAAC,EAAE,GAAG,EAAE,KAAK,EAAE,MAAM,EAAE,qBAAqB,EAAE,CAAC,CAAC;YAC/D,QAAQ,EAAE,CAAC;YACX,SAAS;QACX,CAAC;QAED,wBAAwB;QACxB,MAAM,KAAK,GAAG,eAAe,CAAC,GAAG,CAAC,CAAC;QACnC,IAAI,KAAK,IAAI,CAAC,KAAK,GAAG,KAAK,CAAC,CAAC,CAAC,IAAI,KAAK,GAAG,KAAK,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC;YACpD,UAAU,CAAC,IAAI,CAAC;gBACd,GAAG;gBACH,KAAK;gBACL,MAAM,EAAE,2BAA2B,KAAK,CAAC,CAAC,CAAC,KAAK,KAAK,CAAC,CAAC,CAAC,GAAG;aAC5D,CAAC,CAAC;YACH,QAAQ,EAAE,CAAC;YACX,SAAS;QACX,CAAC;QAED,sBAAsB;QACtB,MAAM,OAAO,GAAG,aAAa,CAAC,GAAG,CAAC,CAAC;QACnC,IAAI,CAAC,OAAO;YAAE,SAAS,CAAC,oBAAoB;QAE5C,MAAM,cAAc,GAAmB;YACrC,IAAI,EAAE,OAAO,CAAC,IAAI;YAClB,KAAK;YACL,IAAI,EAAE,OAAO,CAAC,IAAI;SACnB,CAAC;QAED,UAA6C,CAAC,OAAO,CAAC,KAAK,CAAC,GAAG,cAAc,CAAC;QAC/E,QAAQ,EAAE,CAAC;IACb,CAAC;IAED,mDAAmD;IACnD,MAAM,UAAU,GAAG,iBAAiB,CAAC,UAAU,EAAE,QAAQ,EAAE,QAAQ,CAAC,CAAC;IAErE,OAAO,EAAE,UAAU,EAAE,QAAQ,EAAE,QAAQ,EAAE,UAAU,EAAE,UAAU,EAAE,CAAC;AACpE,CAAC;AAED;;;;;;;GAOG;AACH,SAAS,iBAAiB,CAAC,UAAwB,EAAE,QAAgB,EAAE,QAAgB;IACrF,IAAI,QAAQ,KAAK,CAAC;QAAE,OAAO,CAAC,CAAC;IAE7B,oEAAoE;IACpE,MAAM,UAAU,GAAG,IAAI,CAAC,GAAG,CAAC,QAAQ,GAAG,EAAE,EAAE,GAAG,CAAC,CAAC;IAEhD,yBAAyB;IACzB,MAAM,KAAK,GAAG,QAAQ,GAAG,QAAQ,CAAC;IAClC,MAAM,gBAAgB,GAAG,KAAK,GAAG,CAAC,CAAC,CAAC,CAAC,QAAQ,GAAG,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC;IAE1D,qCAAqC;IACrC,IAAI,gBAAgB,GAAG,GAAG,CAAC;IAE3B,+EAA+E;IAC/E,IAAI,UAAU,CAAC,GAAG,IAAI,UAAU,CAAC,GAAG,EAAE,CAAC;QACrC,IAAI,UAAU,CAAC,GAAG,CAAC,KAAK,GAAG,UAAU,CAAC,GAAG,CAAC,KAAK,GAAG,CAAC,EAAE,CAAC;YACpD,gBAAgB,IAAI,GAAG,CAAC;QAC1B,CAAC;IACH,CAAC;IAED,mDAAmD;IACnD,IAAI,UAAU,CAAC,GAAG,IAAI,UAAU,CAAC,gBAAgB,EAAE,CAAC;QAClD,IAAI,UAAU,CAAC,GAAG,CAAC,KAAK,GAAG,UAAU,CAAC,gBAAgB,CAAC,KAAK,EAAE,CAAC;YAC7D,gBAAgB,IAAI,IAAI,CAAC;QAC3B,CAAC;IACH,CAAC;IAED,mDAAmD;IACnD,IAAI,UAAU,CAAC,IAAI,IAAI,UAAU,CAAC,gBAAgB,EAAE,CAAC;QACnD,IAAI,UAAU,CAAC,IAAI,CAAC,KAAK,GAAG,UAAU,CAAC,gBAAgB,CAAC,KAAK,EAAE,CAAC;YAC9D,gBAAgB,IAAI,IAAI,CAAC;QAC3B,CAAC;IACH,CAAC;IAED,MAAM,GAAG,GACP,UAAU,GAAG,GAAG,GAAG,CAAC,CAAC,GAAG,gBAAgB,CAAC,GAAG,IAAI,GAAG,IAAI,CAAC,GAAG,CAAC,gBAAgB,EAAE,CAAC,CAAC,GAAG,IAAI,CAAC;IAC1F,OAAO,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,GAAG,CAAC,GAAG,EAAE,CAAC,CAAC,EAAE,CAAC,CAAC,GAAG,GAAG,CAAC,GAAG,GAAG,CAAC;AAC/D,CAAC;AAED,qBAAqB;AACrB,OAAO,EAAE,eAAe,EAAE,aAAa,EAAE,CAAC"}

package/package.json

@@ -1,10 +1,15 @@
 {
   "name": "@ottolab/extraction",
-  "version": "0.1.0",
+  "version": "0.1.1",
   "license": "MIT",
   "repository": "https://github.com/hokev/Otto",
-  "publishConfig": { "access": "public" },
+  "publishConfig": {
+    "access": "public"
+  },
   "type": "module",
+  "files": [
+    "dist"
+  ],
   "exports": {
     ".": {
       "import": {
@@ -16,6 +21,7 @@
   },
   "scripts": {
     "build": "tsc --build",
+    "prepublishOnly": "npm run build",
     "lint": "eslint src/",
     "test": "vitest run"
   },

package/__tests__/llm-extractor.test.ts

@@ -1,73 +0,0 @@
-import { describe, it, expect } from 'vitest';
-import { extractFromCsv } from '../src/llm-extractor.js';
-
-describe('extractFromCsv', () => {
-  it('should parse simple CSV with standard headers', () => {
-    const csv = `Total Cholesterol,LDL,HDL,Triglycerides,HbA1c
-200,120,55,130,5.4`;
-
-    const result = extractFromCsv(csv);
-    expect(result.total_cholesterol).toBe(200);
-    expect(result.ldl_c).toBe(120);
-    expect(result.hdl).toBe(55);
-    expect(result.triglycerides).toBe(130);
-    expect(result.hba1c).toBe(5.4);
-  });
-
-  it('should normalize common header aliases', () => {
-    const csv = `LDL-C,HDL-C,A1C,Glucose,SGPT
-110,58,5.2,92,25`;
-
-    const result = extractFromCsv(csv);
-    expect(result.ldl_c).toBe(110);
-    expect(result.hdl).toBe(58);
-    expect(result.hba1c).toBe(5.2);
-    expect(result.fasting_glucose).toBe(92);
-    expect(result.alt).toBe(25);
-  });
-
-  it('should handle empty values', () => {
-    const csv = `Total Cholesterol,LDL,HDL
-200,,55`;
-
-    const result = extractFromCsv(csv);
-    expect(result.total_cholesterol).toBe(200);
-    expect(result.hdl).toBe(55);
-    expect(result.ldl_c).toBeUndefined();
-  });
-
-  it('should throw on single-line CSV (no data row)', () => {
-    expect(() => extractFromCsv('Total Cholesterol,LDL,HDL')).toThrow(
-      'at least a header row and a data row',
-    );
-  });
-
-  it('should handle case-insensitive headers', () => {
-    const csv = `CREATININE,BUN,eGFR
-0.9,15,95`;
-
-    const result = extractFromCsv(csv);
-    expect(result.creatinine).toBe(0.9);
-    expect(result.bun).toBe(15);
-    expect(result.egfr).toBe(95);
-  });
-
-  it('should preserve non-numeric values as strings', () => {
-    const csv = `Albumin,Notes
-4.2,Normal`;
-
-    const result = extractFromCsv(csv);
-    expect(result.albumin).toBe(4.2);
-    expect(result.notes).toBe('Normal');
-  });
-
-  it('should handle whitespace in headers and values', () => {
-    const csv = ` Total Cholesterol , LDL , HDL
- 200 , 120 , 55 `;
-
-    const result = extractFromCsv(csv);
-    expect(result.total_cholesterol).toBe(200);
-    expect(result.ldl_c).toBe(120);
-    expect(result.hdl).toBe(55);
-  });
-});

package/__tests__/validator.test.ts

@@ -1,243 +0,0 @@
-import { describe, it, expect } from 'vitest';
-import { validateExtraction, CLINICAL_RANGES, BIOMARKER_MAP } from '../src/validator.js';
-import type { RawBiomarkers } from '../src/llm-extractor.js';
-
-describe('CLINICAL_RANGES', () => {
-  it('should have ranges for all mapped biomarkers', () => {
-    for (const key of Object.keys(BIOMARKER_MAP)) {
-      expect(CLINICAL_RANGES[key]).toBeDefined();
-    }
-  });
-
-  it('should have low < high for all ranges', () => {
-    for (const [key, [low, high]] of Object.entries(CLINICAL_RANGES)) {
-      expect(low).toBeLessThan(high);
-      expect(low).toBeTypeOf('number');
-      expect(high).toBeTypeOf('number');
-      // Verify key exists
-      expect(key).toBeTruthy();
-    }
-  });
-});
-
-describe('validateExtraction', () => {
-  it('should accept valid biomarkers within range', () => {
-    const raw: RawBiomarkers = {
-      total_cholesterol: 200,
-      ldl_c: 120,
-      hdl: 55,
-      triglycerides: 130,
-      hba1c: 5.4,
-      fasting_glucose: 95,
-      creatinine: 0.9,
-      albumin: 4.2,
-    };
-
-    const result = validateExtraction(raw);
-    expect(result.accepted).toBe(8);
-    expect(result.rejected).toBe(0);
-    expect(result.biomarkers.totalCholesterol?.value).toBe(200);
-    expect(result.biomarkers.ldlC?.value).toBe(120);
-    expect(result.biomarkers.hdl?.value).toBe(55);
-    expect(result.biomarkers.albumin?.value).toBe(4.2);
-  });
-
-  it('should reject values outside clinical ranges', () => {
-    const raw: RawBiomarkers = {
-      total_cholesterol: 9999, // way too high
-      hdl: -5, // negative
-      hba1c: 5.4, // valid
-    };
-
-    const result = validateExtraction(raw);
-    expect(result.accepted).toBe(1); // only hba1c
-    expect(result.rejected).toBe(2);
-    expect(result.biomarkers.hba1c?.value).toBe(5.4);
-    expect(result.biomarkers.totalCholesterol).toBeUndefined();
-    expect(result.biomarkers.hdl).toBeUndefined();
-  });
-
-  it('should skip null and undefined values', () => {
-    const raw: RawBiomarkers = {
-      total_cholesterol: null,
-      ldl_c: 100,
-      hdl: null,
-    };
-
-    const result = validateExtraction(raw);
-    expect(result.accepted).toBe(1);
-    expect(result.biomarkers.ldlC?.value).toBe(100);
-  });
-
-  it('should skip string values (qualitative results)', () => {
-    const raw: RawBiomarkers = {
-      total_cholesterol: 200,
-      some_qualitative: 'Negative',
-    };
-
-    const result = validateExtraction(raw);
-    expect(result.accepted).toBe(1);
-    expect(result.rejected).toBe(0);
-  });
-
-  it('should skip gender field', () => {
-    const raw: RawBiomarkers = {
-      gender: 'm',
-      total_cholesterol: 200,
-    };
-
-    const result = validateExtraction(raw);
-    expect(result.accepted).toBe(1);
-  });
-
-  it('should reject non-finite numbers', () => {
-    const raw: RawBiomarkers = {
-      total_cholesterol: NaN,
-      ldl_c: Infinity,
-      hdl: 55,
-    };
-
-    const result = validateExtraction(raw);
-    expect(result.accepted).toBe(1);
-    expect(result.rejected).toBe(2);
-  });
-
-  it('should map to correct BiomarkerSet fields', () => {
-    const raw: RawBiomarkers = {
-      hs_crp: 1.5,
-      vitamin_d: 45,
-      lymphocyte_percent: 30,
-      bilirubin_total: 0.8,
-    };
-
-    const result = validateExtraction(raw);
-    expect(result.biomarkers.hsCrp?.value).toBe(1.5);
-    expect(result.biomarkers.hsCrp?.unit).toBe('mg/L');
-    expect(result.biomarkers.vitaminD?.value).toBe(45);
-    expect(result.biomarkers.lymphocytePercent?.value).toBe(30);
-    expect(result.biomarkers.bilirubinTotal?.value).toBe(0.8);
-  });
-
-  it('should include unit and name in BiomarkerValue', () => {
-    const raw: RawBiomarkers = { alt: 25 };
-    const result = validateExtraction(raw);
-    expect(result.biomarkers.alt).toEqual({
-      name: 'ALT',
-      value: 25,
-      unit: 'U/L',
-    });
-  });
-
-  it('should record rejections with reasons', () => {
-    const raw: RawBiomarkers = {
-      total_cholesterol: 9999,
-    };
-
-    const result = validateExtraction(raw);
-    expect(result.rejections.length).toBe(1);
-    expect(result.rejections[0].key).toBe('total_cholesterol');
-    expect(result.rejections[0].reason).toContain('outside clinical range');
-  });
-
-  it('should compute reasonable confidence for good extraction', () => {
-    const raw: RawBiomarkers = {
-      total_cholesterol: 200,
-      ldl_c: 120,
-      hdl: 55,
-      triglycerides: 130,
-      hba1c: 5.4,
-      fasting_glucose: 95,
-      creatinine: 0.9,
-      albumin: 4.2,
-      alt: 25,
-      ast: 22,
-      wbc: 7.0,
-      hemoglobin: 14.5,
-      platelets: 250,
-      mcv: 90,
-      rdw: 13,
-    };
-
-    const result = validateExtraction(raw);
-    expect(result.confidence).toBeGreaterThan(0.7);
-  });
-
-  it('should return 0 confidence for empty extraction', () => {
-    const result = validateExtraction({});
-    expect(result.confidence).toBe(0);
-    expect(result.accepted).toBe(0);
-  });
-
-  it('should reduce confidence when HDL > total cholesterol', () => {
-    const consistent: RawBiomarkers = {
-      total_cholesterol: 200,
-      hdl: 55,
-      ldl_c: 120,
-      hba1c: 5.4,
-      creatinine: 0.9,
-    };
-    const inconsistent: RawBiomarkers = {
-      total_cholesterol: 100,
-      hdl: 150, // impossible: HDL > total
-      ldl_c: 120,
-      hba1c: 5.4,
-      creatinine: 0.9,
-    };
-
-    const c1 = validateExtraction(consistent).confidence;
-    // HDL=150 is within range [5,150] so it passes range check but fails consistency
-    // However total_cholesterol=100 makes HDL > total
-    const c2 = validateExtraction(inconsistent).confidence;
-    expect(c2).toBeLessThan(c1);
-  });
-
-  it('should handle all target biomarkers', () => {
-    const raw: RawBiomarkers = {
-      weight_kg: 75,
-      height_cm: 175,
-      bmi: 24.5,
-      body_fat_percent: 18,
-      systolic_bp: 120,
-      diastolic_bp: 75,
-      heart_rate: 68,
-      total_cholesterol: 195,
-      ldl_c: 110,
-      hdl: 58,
-      triglycerides: 120,
-      apoB: 85,
-      hba1c: 5.3,
-      fasting_glucose: 88,
-      fasting_insulin: 5.5,
-      uric_acid: 5.2,
-      creatinine: 0.95,
-      bun: 15,
-      egfr: 95,
-      alt: 22,
-      ast: 20,
-      alp: 70,
-      ggt: 25,
-      bilirubin_total: 0.7,
-      albumin: 4.3,
-      hs_crp: 0.8,
-      esr: 8,
-      cortisol: 12,
-      testosterone: 550,
-      estradiol: 25,
-      tsh: 1.8,
-      wbc: 6.5,
-      rbc: 4.8,
-      hemoglobin: 14.8,
-      hematocrit: 44,
-      platelets: 240,
-      mcv: 88,
-      rdw: 12.8,
-      lymphocyte_percent: 32,
-      vitamin_d: 48,
-    };
-
-    const result = validateExtraction(raw);
-    expect(result.accepted).toBe(40);
-    expect(result.rejected).toBe(0);
-    expect(result.confidence).toBeGreaterThan(0.8);
-  });
-});

package/src/index.ts

@@ -1,69 +0,0 @@
-import type { ExtractionResult, ExtendedLLMProvider } from '@ottolab/shared';
-import { detectLab } from './lab-detector.js';
-import { extractFromPdf, extractFromCsv } from './llm-extractor.js';
-import { validateExtraction } from './validator.js';
-
-export interface ParseInput {
-  /** Base64-encoded PDF data */
-  pdf?: string;
-  /** Raw CSV text */
-  csv?: string;
-}
-
-/**
- * Main extraction pipeline.
- *
- * PDF flow:
- *   1. Lab Detection (LLM classify → Quest | LabCorp | international | unknown)
- *   2. Structured Extraction (LLM multimodal + lab-specific few-shot)
- *   3. Validation (clinical range checks, cross-biomarker consistency, confidence)
- *
- * CSV flow:
- *   1. Column parsing + header normalization
- *   2. Validation
- */
-export async function runExtractionPipeline(
-  input: ParseInput,
-  provider?: ExtendedLLMProvider,
-): Promise<ExtractionResult> {
-  if (input.pdf) {
-    if (!provider) throw new Error('LLM provider required for PDF extraction');
-
-    // Step 1: Detect lab source
-    const detection = await detectLab(input.pdf, provider);
-
-    // Step 2: Extract biomarkers with lab-specific prompt
-    const raw = await extractFromPdf(input.pdf, detection.lab, provider);
-
-    // Step 3: Validate and transform
-    const validation = validateExtraction(raw);
-
-    return {
-      biomarkers: validation.biomarkers,
-      sourceLab: detection.lab,
-      sourceLanguage: detection.language,
-      confidence: validation.confidence,
-    };
-  }
-
-  if (input.csv) {
-    // CSV extraction (no LLM needed)
-    const raw = extractFromCsv(input.csv);
-    const validation = validateExtraction(raw);
-
-    return {
-      biomarkers: validation.biomarkers,
-      sourceLab: 'unknown',
-      sourceLanguage: 'en',
-      confidence: validation.confidence,
-    };
-  }
-
-  throw new Error('Either pdf (base64) or csv text must be provided');
-}
-
-export { detectLab } from './lab-detector.js';
-export { extractFromPdf, extractFromCsv } from './llm-extractor.js';
-export { validateExtraction } from './validator.js';
-export type { RawBiomarkers } from './llm-extractor.js';
-export type { ValidationResult } from './validator.js';

package/src/lab-detector.ts

@@ -1,60 +0,0 @@
-import type { SourceLab, ExtendedLLMProvider } from '@ottolab/shared';
-import { LAB_DETECTION_PROMPT } from './prompts/base.js';
-
-export interface LabDetectionResult {
-  lab: SourceLab;
-  language: string;
-  confidence: number;
-}
-
-/**
- * Detect the source laboratory from a PDF using LLM classification.
- * Sends the first page(s) to the LLM to identify Quest, LabCorp, etc.
- */
-export async function detectLab(
-  pdfBase64: string,
-  provider: ExtendedLLMProvider,
-): Promise<LabDetectionResult> {
-  try {
-    const response = await provider.chatMultimodal(
-      [
-        { type: 'document', data: pdfBase64, mimeType: 'application/pdf' },
-        { type: 'text', text: LAB_DETECTION_PROMPT },
-      ],
-      { temperature: 0, maxTokens: 256, responseFormat: 'json' },
-    );
-
-    const parsed = parseJsonResponse<LabDetectionResult>(response);
-
-    // Validate lab value
-    const validLabs: SourceLab[] = ['quest', 'labcorp', 'international', 'unknown'];
-    if (!validLabs.includes(parsed.lab)) {
-      parsed.lab = 'unknown';
-    }
-
-    return {
-      lab: parsed.lab,
-      language: parsed.language || 'en',
-      confidence: Math.min(Math.max(parsed.confidence || 0, 0), 1),
-    };
-  } catch {
-    return { lab: 'unknown', language: 'en', confidence: 0 };
-  }
-}
-
-function parseJsonResponse<T>(text: string): T {
-  // Strip markdown code fences if present
-  let cleaned = text.trim();
-  if (cleaned.startsWith('```')) {
-    cleaned = cleaned.replace(/^```(?:json)?\s*/, '').replace(/\s*```$/, '');
-  }
-
-  // Find JSON object boundaries
-  const start = cleaned.indexOf('{');
-  const end = cleaned.lastIndexOf('}');
-  if (start === -1 || end === -1) {
-    throw new Error('No JSON object found in response');
-  }
-
-  return JSON.parse(cleaned.slice(start, end + 1)) as T;
-}