File size: 6,319 Bytes
8ed1de7 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 | """
Temporal Split Verification Script
Ensures:
- No patient appears in both training and test sets
- Training and test weeks are disjoint per disease
- Test set for each disease contains at least 10 samples
"""
import pandas as pd
import json
from pathlib import Path
from collections import defaultdict
# ============================================
# CONFIGURATION
# ============================================
METADATA_PATH = Path("../metadata.csv")
MANIFEST_PATH = Path("../data/manifest.json")
# Temporal splits: mapping of disease -> set of test weeks
# Based on Table 2 from the paper
TEMPORAL_SPLITS = {
'Z00': {9, 10, 11},
'E11': {1, 2, 3},
'K29': {1, 2, 3},
'K76': {2, 3, 4},
'B18': {2, 3, 10},
'C34': {8, 9, 10},
'N18': {2, 3, 4},
'J44': {5, 6, 7, 11},
'A15': {13}
}
# ============================================
# VALIDATION FUNCTIONS
# ============================================
def load_metadata():
"""Load metadata CSV."""
return pd.read_csv(METADATA_PATH)
def validate_patient_no_leakage(metadata):
"""Check that no patient appears in both train and test sets."""
errors = []
for disease in metadata['Diagnosis'].unique():
disease_patients = metadata[metadata['Diagnosis'] == disease]
test_weeks = TEMPORAL_SPLITS.get(disease, set())
# Patients in test weeks
test_patients = disease_patients[disease_patients['Week'].isin(test_weeks)]['Patient_id'].tolist()
# Patients in training weeks
train_patients = disease_patients[~disease_patients['Week'].isin(test_weeks)]['Patient_id'].tolist()
# Check overlap
overlap = set(test_patients) & set(train_patients)
if overlap:
errors.append(f"Patient leakage in {disease}: patients {overlap} appear in both train and test")
if errors:
print(f"❌ Patient leakage validation FAILED: {len(errors)} errors")
for err in errors[:5]:
print(f" {err}")
return False
print("✅ Patient leakage validation PASSED: no patient appears in both train and test")
return True
def validate_train_test_weeks_disjoint(metadata):
"""Check that training and test weeks are disjoint per disease."""
errors = []
for disease in metadata['Diagnosis'].unique():
disease_patients = metadata[metadata['Diagnosis'] == disease]
test_weeks = TEMPORAL_SPLITS.get(disease, set())
all_weeks = set(disease_patients['Week'].unique())
train_weeks = all_weeks - test_weeks
# Check for overlap
overlap = test_weeks & train_weeks
if overlap:
errors.append(f"Week overlap in {disease}: weeks {overlap} appear in both train and test")
# Check that test weeks are actually present in the data
missing_test_weeks = len(test_weeks - all_weeks)>2
if missing_test_weeks:
errors.append(f"Missing test weeks in {disease}: {missing_test_weeks} not present in data")
if errors:
print(f"❌ Week disjointness validation FAILED: {len(errors)} errors")
for err in errors[:5]:
print(f" {err}")
return False
print("✅ Week disjointness validation PASSED: train and test weeks are disjoint per disease")
return True
def validate_test_set_size(metadata):
"""Check that test set for each disease contains at least 10 samples."""
errors = []
for disease in metadata['Diagnosis'].unique():
disease_patients = metadata[metadata['Diagnosis'] == disease]
test_weeks = TEMPORAL_SPLITS.get(disease, set())
test_patients = disease_patients[disease_patients['Week'].isin(test_weeks)]
n_test = len(test_patients)
if n_test < 10:
errors.append(f"Test set too small in {disease}: {n_test} samples (min 10 required)")
if errors:
print(f"❌ Test set size validation FAILED: {len(errors)} errors")
for err in errors:
print(f" {err}")
return False
# Print summary
print("\nTest set sizes per disease:")
for disease in metadata['Diagnosis'].unique():
test_weeks = TEMPORAL_SPLITS.get(disease, set())
disease_patients = metadata[metadata['Diagnosis'] == disease]
test_patients = disease_patients[disease_patients['Week'].isin(test_weeks)]
print(f" {disease}: {len(test_patients)} samples")
print("\n✅ Test set size validation PASSED: all diseases have ≥10 test samples")
return True
def validate_manifest_consistency(metadata):
"""Check that all patients in metadata have a corresponding JSON file."""
# Load manifest
with open(MANIFEST_PATH, 'r') as f:
manifest = json.load(f)
manifest_ids = set(int(entry['patient_id']) for entry in manifest['files'])
metadata_ids = set(metadata['Patient_id'])
# Check missing in manifest
missing_in_manifest = metadata_ids - manifest_ids
if missing_in_manifest:
print(f"⚠️ {len(missing_in_manifest)} patients in metadata not found in manifest")
return False
# Check extra in manifest
extra_in_manifest = manifest_ids - metadata_ids
if extra_in_manifest:
print(f"⚠️ {len(extra_in_manifest)} patients in manifest not found in metadata")
return False
print(f"✅ Manifest consistency PASSED: {len(manifest_ids)} patients match")
return True
def main():
print("=" * 60)
print("S-OH TEMPORAL SPLIT VALIDATION")
print("=" * 60)
metadata = load_metadata()
print(f"Loaded {len(metadata)} records")
all_passed = True
print("\n" + "-" * 40)
all_passed &= validate_manifest_consistency(metadata)
print("\n" + "-" * 40)
all_passed &= validate_patient_no_leakage(metadata)
print("\n" + "-" * 40)
all_passed &= validate_train_test_weeks_disjoint(metadata)
print("\n" + "-" * 40)
all_passed &= validate_test_set_size(metadata)
print("\n" + "=" * 60)
if all_passed:
print("✅ ALL CHECKS PASSED")
else:
print("❌ SOME CHECKS FAILED")
print("=" * 60)
return all_passed
if __name__ == "__main__":
main() |