evo-clinical-unit-harmonization

Converts clinical lab values from SI to US conventional units using boundary-based auto-detection.

Key Functions

• build_conversion_registry() - Returns dict mapping column names to analyte configs
• detect_and_convert_value(val, config) - Detect and convert a single value
• harmonize_column(series, config) - Harmonize an entire column
• harmonize_all_features(df, skip_cols) - Harmonize all features in DataFrame

Usage

import sys
sys.path.insert(0, '/app/environment/skills/evo-clinical-unit-harmonization/scripts')
from utils import harmonize_all_features

df = harmonize_all_features(df, skip_cols=['patient_id'])

Conversion Registry

Covers all 62 CKD features including:

• Creatinine (SI umol/L -> US mg/dL, factor 0.0113)
• BUN (SI mmol/L -> US mg/dL, factor 2.8)
• Hemoglobin (SI g/L -> US g/dL, factor 0.1)
• Albumin/Total Protein (SI g/L -> US g/dL, factor 0.1)
• Glucose (SI mmol/L -> US mg/dL, factor 18.0)
• HbA1c (SI mmol/mol -> US %, linear: val*0.0915+2.15)
• Calcium (SI mmol/L -> US mg/dL, factor 4.0)
• Phosphorus (SI mmol/L -> US mg/dL, factor 3.1)
• Cholesterol/LDL/HDL (SI mmol/L -> US mg/dL, factor 38.67)
• Triglycerides (SI mmol/L -> US mg/dL, factor 88.57)
• Uric Acid (SI umol/L -> US mg/dL, factor 0.0168)
• Bilirubin (SI umol/L -> US mg/dL, factor 0.0585)
• Iron/TIBC (SI umol/L -> US ug/dL, factor 5.587)
• Hematocrit (SI L/L -> US %, factor 100)