Fix: missing error handling of ground truth labels with a slash

src/stamp/modeling/data.py

@@ -1104,11 +1104,9 @@ def filter_complete_patient_data_(
         )
     }
 
-    _logger.info(
-        f"Total patients in clinical table: {len(patient_to_ground_truth)}\n"
-        f"Patients appearing in slide table: {len(patient_to_slides)}\n"
-        f"Final usable patients (complete data): {len(patients)}\n"
-    )
+    _logger.info("Total patients in clinical table: %d", len(patient_to_ground_truth))
+    _logger.info("Patients appearing in slide table: %d", len(patient_to_slides))
+    _logger.info("Final usable patients (complete data): %d", len(patients))
     return patients
 
 

src/stamp/modeling/models/cox.py

@@ -4,7 +4,6 @@
 # pylint: disable=C0103
 # pylint: disable=C0301
 
-import sys
 import warnings
 
 import torch
@@ -268,15 +267,3 @@ def neg_partial_log_likelihood(
             )
         )
     return loss
-
-
-if __name__ == "__main__":
-    import doctest
-
-    # Run doctest
-    results = doctest.testmod()
-    if results.failed == 0:
-        print("All tests passed.")
-    else:
-        print("Some doctests failed.")
-        sys.exit(1)

src/stamp/statistics/__init__.py

@@ -30,8 +30,9 @@
 )
 from stamp.statistics.survival import _plot_km, _survival_stats_for_csv
 from stamp.types import PandasLabel, Task
+from stamp.utils.path import path_safe
 
-__all__ = ["StatsConfig", "compute_stats_"]
+__all__ = ["StatsConfig", "compute_stats_", "path_safe"]
 
 
 __author__ = "Marko van Treeck, Minh Duc Nguyen"
@@ -146,7 +147,9 @@ def _compute_multitarget_classification_stats(
                 )
 
             fig.tight_layout()
-            fig.savefig(output_dir / f"roc-curve_{target_label}={true_class}.svg")
+            fig.savefig(
+                output_dir / f"roc-curve_{path_safe(target_label)}={true_class}.svg"
+            )
             plt.close(fig)
 
             # Plot PRC curve
@@ -172,7 +175,9 @@ def _compute_multitarget_classification_stats(
                 )
 
             fig.tight_layout()
-            fig.savefig(output_dir / f"pr-curve_{target_label}={true_class}.svg")
+            fig.savefig(
+                output_dir / f"pr-curve_{path_safe(target_label)}={true_class}.svg"
+            )
             plt.close(fig)
 
     # Compute aggregated statistics for all targets
@@ -292,7 +297,8 @@ def compute_stats_(
                 fig.tight_layout()
                 output_dir.mkdir(parents=True, exist_ok=True)
                 fig.savefig(
-                    output_dir / f"roc-curve_{ground_truth_label}={true_class}.svg"
+                    output_dir
+                    / f"roc-curve_{path_safe(ground_truth_label)}={true_class}.svg"
                 )
                 plt.close(fig)
 
@@ -321,7 +327,8 @@ def compute_stats_(
 
                 fig.tight_layout()
                 fig.savefig(
-                    output_dir / f"pr-curve_{ground_truth_label}={true_class}.svg"
+                    output_dir
+                    / f"pr-curve_{path_safe(ground_truth_label)}={true_class}.svg"
                 )
                 plt.close(fig)
 

src/stamp/statistics/categorical.py

@@ -7,6 +7,8 @@
 import scipy.stats as st
 from sklearn import metrics
 
+from stamp.utils.path import path_safe
+
 __author__ = "Marko van Treeck"
 __copyright__ = "Copyright (C) 2022-2025 Marko van Treeck"
 __license__ = "MIT"
@@ -137,9 +139,13 @@ def categorical_aggregated_(
         )
 
     preds_df = pd.concat(preds_dfs).sort_index()
-    preds_df.to_csv(outpath / f"{ground_truth_label}_categorical-stats_individual.csv")
+    preds_df.to_csv(
+        outpath / f"{path_safe(ground_truth_label)}_categorical-stats_individual.csv"
+    )
     stats_df = _aggregate_categorical_stats(preds_df.reset_index())
-    stats_df.to_csv(outpath / f"{ground_truth_label}_categorical-stats_aggregated.csv")
+    stats_df.to_csv(
+        outpath / f"{path_safe(ground_truth_label)}_categorical-stats_aggregated.csv"
+    )
 
 
 def categorical_aggregated_multitarget_(
@@ -181,11 +187,15 @@ def categorical_aggregated_multitarget_(
 
         # Concatenate and save individual stats for this target
         preds_df = pd.concat(preds_dfs).sort_index()
-        preds_df.to_csv(outpath / f"{target_label}_categorical-stats_individual.csv")
+        preds_df.to_csv(
+            outpath / f"{path_safe(target_label)}_categorical-stats_individual.csv"
+        )
 
         # Aggregate stats for this target
         stats_df = _aggregate_categorical_stats(preds_df.reset_index())
-        stats_df.to_csv(outpath / f"{target_label}_categorical-stats_aggregated.csv")
+        stats_df.to_csv(
+            outpath / f"{path_safe(target_label)}_categorical-stats_aggregated.csv"
+        )
 
         # Store for summary
         all_target_stats[target_label] = stats_df

src/stamp/statistics/regression.py

@@ -10,6 +10,8 @@
 import scipy.stats as st
 from sklearn.metrics import mean_absolute_error, mean_squared_error, r2_score
 
+from stamp.utils.path import path_safe
+
 
 def _regression(preds_df: pd.DataFrame, target_label: str) -> pd.Series:
     """Compute regression metrics for one prediction table."""
@@ -107,10 +109,14 @@ def regression_aggregated_(
 
     # Save individual stats and aggregate
     stats_df = pd.DataFrame(stats).transpose()
-    stats_df.to_csv(outpath / f"{ground_truth_label}_regression-stats_individual.csv")
+    stats_df.to_csv(
+        outpath / f"{path_safe(ground_truth_label)}_regression-stats_individual.csv"
+    )
 
     mean = stats_df.mean(numeric_only=True)
     sem = stats_df.sem(numeric_only=True)
     lower, upper = st.t.interval(0.95, len(stats_df) - 1, loc=mean, scale=sem)
     agg = pd.DataFrame({"mean": mean, "95%_low": lower, "95%_high": upper})
-    agg.to_csv(outpath / f"{ground_truth_label}_regression-stats_aggregated.csv")
+    agg.to_csv(
+        outpath / f"{path_safe(ground_truth_label)}_regression-stats_aggregated.csv"
+    )

src/stamp/utils/path.py

@@ -0,0 +1,6 @@
+"""Small path-related helpers shared across the project."""
+
+
+def path_safe(label: str) -> str:
+    """Replace path separators so labels are safe as filename parts."""
+    return label.replace("/", "_").replace("\\", "_")