Add get_sequence_lengths method to HealthDataset and create extract_sequence_lengths script for per-patient sequence length extraction

dataset.py

@@ -88,6 +88,17 @@ class HealthDataset(Dataset):
     def __len__(self) -> int:
         return len(self.patient_ids)
 
+    def get_sequence_lengths(self) -> List[int]:
+        """Return the sequence length for each patient.
+
+        Lengths correspond to what :meth:`__getitem__` returns: the number of
+        patient events plus the inserted DOA event.
+
+        Returns:
+            List[int]: Sequence lengths aligned with dataset indices.
+        """
+        return [len(self.patient_events.get(pid, ())) + 1 for pid in self.patient_ids]
+
     def __getitem__(self, idx):
         if self._cache_event_tensors:
             cached_e = self._cached_event_tensors[idx]

extract_sequence_lengths.py

@@ -0,0 +1,85 @@
+import argparse
+import os
+from typing import List, Optional
+
+import numpy as np
+import pandas as pd
+
+import matplotlib.pyplot as plt
+
+from dataset import HealthDataset
+
+
+def _default_covariates(full_cov: bool) -> Optional[List[str]]:
+    # Mirrors train.py default behavior.
+    if full_cov:
+        return None
+    return ["bmi", "smoking", "alcohol"]
+
+
+def main() -> None:
+    parser = argparse.ArgumentParser(
+        description="Extract per-patient sequence lengths from the UKB dataset."
+    )
+    parser.add_argument(
+        "--data_prefix",
+        type=str,
+        default="ukb",
+        help="Prefix for dataset files (expects <prefix>_basic_info.csv, <prefix>_table.csv, <prefix>_event_data.npy).",
+    )
+    parser.add_argument(
+        "--full_cov",
+        action="store_true",
+        help="Use full covariates (otherwise uses the training default: bmi/smoking/alcohol).",
+    )
+    parser.add_argument(
+        "--out_csv",
+        type=str,
+        default=None,
+        help="Output CSV path. Default: <data_prefix>_sequence_lengths.csv",
+    )
+    parser.add_argument(
+        "--out_npy",
+        type=str,
+        default=None,
+        help="Optional output .npy path for just the lengths array.",
+    )
+    args = parser.parse_args()
+
+    out_csv = args.out_csv or f"{args.data_prefix}_sequence_lengths.csv"
+
+    cov_list = _default_covariates(args.full_cov)
+    ds = HealthDataset(data_prefix=args.data_prefix, covariate_list=cov_list)
+
+    lengths = ds.get_sequence_lengths()
+
+    df = pd.DataFrame({"eid": ds.patient_ids, "seq_len": lengths})
+    os.makedirs(os.path.dirname(out_csv) or ".", exist_ok=True)
+    df.to_csv(out_csv, index=False)
+
+    if args.out_npy:
+        os.makedirs(os.path.dirname(args.out_npy) or ".", exist_ok=True)
+        np.save(args.out_npy, np.asarray(lengths, dtype=np.int32))
+
+    arr = np.asarray(lengths, dtype=np.int64)
+    print(f"Wrote: {out_csv}")
+    print(
+        "Summary: "
+        f"n={arr.size}, min={arr.min()}, p50={int(np.median(arr))}, mean={arr.mean():.2f}, max={arr.max()}"
+    )
+
+    # Plot histogram
+    plt.figure(figsize=(8, 5))
+    plt.hist(arr, bins=50, color="#348abd", edgecolor="black", alpha=0.8)
+    plt.xlabel("Sequence length (including DOA)")
+    plt.ylabel("Number of patients")
+    plt.title("Histogram of Patient Sequence Lengths")
+    plt.tight_layout()
+    plot_path = os.path.splitext(out_csv)[0] + "_hist.png"
+    plt.savefig(plot_path)
+    print(f"Histogram saved to: {plot_path}")
+    plt.show()
+
+
+if __name__ == "__main__":
+    main()