File size: 15,073 Bytes
0fe62c7
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
#!/usr/bin/env python3
# -*- coding: utf-8 -*-

"""
Pre-embed Patient Summaries Script

This script pre-processes and embeds a patient database,
saving the results to a single Parquet file for faster loading 
in the main application and compatibility with Hugging Face datasets.

Usage:
python preembed_patients.py --patients ../v20_public_data/patient_summaries_and_their_spaces.parquet --embedder ksg-dfci/TrialSpace-1225 --output synthetic_patient_embeddings.parquet --gpus 0,1 --patient-boilerplate-col patient_boilerplate_text --patient-id-col pseudo_mrn


This will create:
    - synthetic_patient_embeddings.parquet: Patient dataframe with embedding vectors as a column
    
The parquet file contains:
    - All original patient columns (patient_id, patient_summary, patient_boilerplate, etc.)
    - patient_embedding: The embedding vector for each patient (stored as list of floats)
    - Metadata stored in parquet file metadata (embedder model, creation date, etc.)

To upload to Hugging Face:
    from datasets import Dataset
    ds = Dataset.from_parquet("synthetic_patient_embeddings.parquet")
    ds.push_to_hub("your-username/patient-embeddings")
"""

import argparse
import pandas as pd
import numpy as np
import torch
import json
import pyarrow as pa
import pyarrow.parquet as pq
from pathlib import Path
from datetime import datetime
from typing import Tuple, List
from sentence_transformers import SentenceTransformer
from transformers import AutoTokenizer


def truncate_text(text: str, tokenizer, max_tokens: int = 1500) -> str:
    """Truncate text to a maximum number of tokens."""
    return tokenizer.decode(
        tokenizer.encode(text, add_special_tokens=True, truncation=True, max_length=max_tokens),
        skip_special_tokens=True
    )


def load_patients(file_path: str, patient_id_col: str = 'patient_id', patient_boilerplate_col: str = 'patient_boilerplate') -> pd.DataFrame:
    """Load patients from parquet file."""
    print(f"\n{'='*70}")
    print(f"Loading patient database from: {file_path}")
    print(f"{'='*70}")
    
    if file_path.endswith('.parquet'):
        df = pd.read_parquet(file_path)
    elif file_path.endswith('.csv'):
        df = pd.read_csv(file_path)
    elif file_path.endswith(('.xlsx', '.xls')):
        df = pd.read_excel(file_path)
    else:
        raise ValueError("Unsupported file format. Use Parquet, CSV, or Excel.")
    
    # Check required columns
    required_cols = [patient_id_col, 'patient_summary']
    missing = [col for col in required_cols if col not in df.columns]
    if missing:
        raise ValueError(f"Missing required columns: {', '.join(missing)}")
    
    # Rename patient_id column to standard name if different
    if patient_id_col != 'patient_id':
        df = df.rename(columns={patient_id_col: 'patient_id'})
        print(f"  Renamed column '{patient_id_col}' to 'patient_id'")
    
    print(f"βœ“ Loaded {len(df)} patients")
    print(f"  Columns: {', '.join(df.columns.tolist())}")
    
    # Clean data
    original_count = len(df)
    df = df[~df['patient_summary'].isnull()].copy()
    df = df[df['patient_summary'].str.strip().str.len() > 0].copy()
    
    # Handle boilerplate column
    if patient_boilerplate_col and patient_boilerplate_col in df.columns:
        if patient_boilerplate_col != 'patient_boilerplate':
            df = df.rename(columns={patient_boilerplate_col: 'patient_boilerplate'})
            print(f"  Renamed column '{patient_boilerplate_col}' to 'patient_boilerplate'")
        df['patient_boilerplate'] = df['patient_boilerplate'].fillna('')
        non_empty_bp = (df['patient_boilerplate'].str.strip().str.len() > 0).sum()
        print(f"  βœ“ Found patient_boilerplate column: {non_empty_bp}/{len(df)} patients have boilerplate text")
    else:
        df['patient_boilerplate'] = ''
        if patient_boilerplate_col:
            print(f"  ⚠ Column '{patient_boilerplate_col}' not found - patient_boilerplate will be empty")
        else:
            print(f"  β—‹ No boilerplate column specified - patient_boilerplate will be empty")
    
    if len(df) < original_count:
        print(f"  ⚠ Removed {original_count - len(df)} patients with missing/empty 'patient_summary'")
    
    return df


def embed_patients(df: pd.DataFrame, embedder_path: str, device: str = None, gpus: list = None) -> Tuple[np.ndarray, str]:
    """Embed patient summaries using the specified embedder model.
    
    Args:
        df: DataFrame with patient data
        embedder_path: Path to embedder model
        device: Single device string (e.g., 'cuda:0', 'cpu') - used if gpus not specified
        gpus: List of GPU indices for multi-GPU parallel processing (e.g., [0, 1, 2, 3])
    """
    print(f"\n{'='*70}")
    print(f"Loading embedder model: {embedder_path}")
    print(f"{'='*70}")
    
    # Determine device configuration
    use_multi_gpu = gpus is not None and len(gpus) > 1
    
    if use_multi_gpu:
        target_devices = [f"cuda:{gpu}" for gpu in gpus]
        print(f"Multi-GPU mode: {target_devices}")
        # Load model on CPU first for multi-process pool
        embedder_model = SentenceTransformer(embedder_path, device='cpu', trust_remote_code=True)
    else:
        if gpus is not None and len(gpus) == 1:
            device = f"cuda:{gpus[0]}"
        elif device is None:
            device = "cuda" if torch.cuda.is_available() else "cpu"
        print(f"Device: {device}")
        embedder_model = SentenceTransformer(embedder_path, device=device, trust_remote_code=True)
    
    embedder_tokenizer = AutoTokenizer.from_pretrained(embedder_path, trust_remote_code=True)
    
    print(f"βœ“ Embedder loaded")
    
    # Set the instruction prompt
    try:
        embedder_model.prompts['query'] = (
            "Instruct: Given a cancer patient summary, retrieve clinical trial options "
            "that are reasonable for that patient; or, given a clinical trial option, "
            "retrieve cancer patients who are reasonable candidates for that trial."
        )
    except:
        pass
    
    try:
        embedder_model.max_seq_length = 2500
    except:
        pass
    
    print(f"\n{'='*70}")
    print(f"Embedding {len(df)} patient summaries")
    print(f"{'='*70}")
    
    # Prepare texts for embedding
    df['patient_summary_trunc'] = df['patient_summary'].apply(
        lambda x: truncate_text(str(x), embedder_tokenizer, max_tokens=1500)
    )
    
    # Add instruction prefix
    prefix = (
        "Instruct: Given a cancer patient summary, retrieve clinical trial options "
        "that are reasonable for that patient; or, given a clinical trial option, "
        "retrieve cancer patients who are reasonable candidates for that trial. "
    )
    texts_to_embed = [prefix + txt for txt in df['patient_summary_trunc'].tolist()]
    
    print(f"  Text length stats:")
    print(f"    Mean: {np.mean([len(t) for t in texts_to_embed]):.0f} chars")
    print(f"    Max: {max([len(t) for t in texts_to_embed])} chars")
    
    # Embed with progress bar
    if use_multi_gpu:
        print(f"  Starting multi-GPU pool on {target_devices}...")
        pool = embedder_model.start_multi_process_pool(target_devices=target_devices)
        
        try:
            embeddings_np = embedder_model.encode_multi_process(
                texts_to_embed,
                pool,
                batch_size=64,
                normalize_embeddings=True,
            )
        finally:
            embedder_model.stop_multi_process_pool(pool)
    else:
        with torch.no_grad():
            embeddings = embedder_model.encode(
                texts_to_embed,
                batch_size=64,
                convert_to_tensor=True,
                normalize_embeddings=True,
                show_progress_bar=True,
                prompt='query'
            )
        embeddings_np = embeddings.cpu().numpy()
    
    print(f"βœ“ Embedding complete")
    print(f"  Shape: {embeddings_np.shape}")
    print(f"  Dtype: {embeddings_np.dtype}")
    
    return embeddings_np, embedder_path


def save_embeddings(df: pd.DataFrame, embeddings: np.ndarray, output_path: str, embedder_path: str, gpus: list = None):
    """Save patient data with embeddings to a single Parquet file.
    
    The embeddings are stored as a column of lists, which is compatible with
    Hugging Face datasets and PyArrow.
    """
    print(f"\n{'='*70}")
    print(f"Saving to: {output_path}")
    print(f"{'='*70}")
    
    # Ensure output path ends with .parquet
    if not output_path.endswith('.parquet'):
        output_path = f"{output_path}.parquet"
    
    output_dir = Path(output_path).parent
    if str(output_dir) and str(output_dir) != '.':
        output_dir.mkdir(parents=True, exist_ok=True)
    
    # Add embeddings as a column (convert numpy arrays to lists for parquet compatibility)
    df_out = df.copy()
    df_out['patient_embedding'] = [emb.tolist() for emb in embeddings]
    
    # Create metadata dictionary
    metadata = {
        "created_at": datetime.now().isoformat(),
        "embedder_model": embedder_path,
        "num_patients": str(len(df)),
        "embedding_dim": str(embeddings.shape[1]),
        "embedding_dtype": str(embeddings.dtype),
        "normalized": "true",
        "gpus_used": str(gpus) if gpus else "single device",
        "format_version": "2.0",  # Version indicator for the new format
    }
    
    # Convert DataFrame to PyArrow Table
    table = pa.Table.from_pandas(df_out)
    
    # Add metadata to the table schema
    existing_metadata = table.schema.metadata or {}
    existing_metadata[b'patient_embedding_metadata'] = json.dumps(metadata).encode('utf-8')
    table = table.replace_schema_metadata(existing_metadata)
    
    # Write to parquet
    pq.write_table(table, output_path)
    
    file_size_mb = Path(output_path).stat().st_size / 1024 / 1024
    print(f"βœ“ Saved parquet file: {output_path}")
    print(f"  Size: {file_size_mb:.2f} MB")
    print(f"  Columns: {', '.join(df_out.columns.tolist())}")
    print(f"  Embedding column: patient_embedding (dim={embeddings.shape[1]})")
    
    print(f"\n{'='*70}")
    print(f"PRE-EMBEDDING COMPLETE")
    print(f"{'='*70}")
    print(f"\nTo use these pre-embedded patients in your app:")
    print(f"1. Update config.py with:")
    print(f"   PREEMBEDDED_PATIENTS = '{output_path}'")
    print(f"2. Restart the application")
    print(f"\nThe app will automatically load these embeddings on startup!")
    print(f"\nTo upload to Hugging Face Hub:")
    print(f"  from datasets import Dataset")
    print(f"  ds = Dataset.from_parquet('{output_path}')")
    print(f"  ds.push_to_hub('your-username/patient-embeddings')")


def main():
    parser = argparse.ArgumentParser(
        description="Pre-embed patient summaries for faster loading",
        formatter_class=argparse.RawDescriptionHelpFormatter,
        epilog="""
Examples:
  python preembed_patients.py --patients data/patients.parquet --embedder models/embedder --output embeddings/patient_embeddings.parquet
  python preembed_patients.py --patients patients.csv --embedder Qwen/Qwen3-Embedding-0.6B --output patient_embeddings.parquet --device cuda
  python preembed_patients.py --patients data.parquet --embedder models/embedder --output out.parquet --patient-id-col mrn
  python preembed_patients.py --patients data.parquet --embedder models/embedder --output out.parquet --gpus 0,1,2,3
  python preembed_patients.py --patients data.parquet --embedder models/embedder --output out.parquet --patient-boilerplate-col boilerplate_summary

Hugging Face Upload:
  After creating the parquet file, you can upload to Hugging Face Hub:
    from datasets import Dataset
    ds = Dataset.from_parquet("patient_embeddings.parquet")
    ds.push_to_hub("your-username/patient-embeddings")
        """
    )
    
    parser.add_argument(
        '--patients',
        type=str,
        required=True,
        help='Path to patient database (Parquet, CSV, or Excel). Required columns: patient_summary and the patient ID column (default: patient_id, or specify with --patient-id-col)'
    )
    
    parser.add_argument(
        '--embedder',
        type=str,
        required=True,
        help='Path to embedder model or HuggingFace model name'
    )
    
    parser.add_argument(
        '--output',
        type=str,
        required=True,
        help='Output path for the parquet file (e.g., "patient_embeddings.parquet")'
    )
    
    parser.add_argument(
        '--device',
        type=str,
        default=None,
        help='Device to use for embedding (default: auto-detect). Examples: cuda, cuda:0, cuda:3, cpu. Ignored if --gpus is specified.'
    )
    
    parser.add_argument(
        '--patient-id-col',
        type=str,
        default='patient_id',
        help='Name of the patient ID column in the input file (default: patient_id)'
    )
    
    parser.add_argument(
        '--patient-boilerplate-col',
        type=str,
        default='patient_boilerplate',
        help='Name of the patient boilerplate column in the input file (default: patient_boilerplate). Set to empty string to skip.'
    )
    
    parser.add_argument(
        '--gpus',
        type=str,
        default=None,
        help='Comma-separated list of GPU indices for multi-GPU parallel processing (e.g., "0,1,2,3"). Overrides --device if specified.'
    )
    
    args = parser.parse_args()
    
    # Parse GPU list if provided
    gpu_list = None
    if args.gpus:
        try:
            gpu_list = [int(g.strip()) for g in args.gpus.split(',')]
        except ValueError:
            print(f"βœ— ERROR: Invalid GPU list format: {args.gpus}")
            print("  Use comma-separated integers, e.g., '0,1,2,3'")
            return 1
    
    print(f"\n{'='*70}")
    print(f"PATIENT SUMMARY PRE-EMBEDDING SCRIPT")
    print(f"{'='*70}")
    print(f"Patient Database:    {args.patients}")
    print(f"Embedder Model:      {args.embedder}")
    print(f"Output File:         {args.output}")
    print(f"Patient ID Col:      {args.patient_id_col}")
    print(f"Boilerplate Col:     {args.patient_boilerplate_col or '(none)'}")
    if gpu_list:
        print(f"GPUs:                {gpu_list} (multi-GPU mode)")
    elif args.device:
        print(f"Device:              {args.device}")
    else:
        print(f"Device:              auto-detect")
    print(f"{'='*70}\n")
    
    try:
        # Load patients
        df = load_patients(args.patients, args.patient_id_col, args.patient_boilerplate_col)
        
        # Embed patients
        embeddings, embedder_path = embed_patients(df, args.embedder, args.device, gpu_list)
        
        # Save everything to single parquet file
        save_embeddings(df, embeddings, args.output, embedder_path, gpu_list)
        
        print(f"\nβœ“ SUCCESS!")
        
    except Exception as e:
        print(f"\nβœ— ERROR: {str(e)}")
        import traceback
        traceback.print_exc()
        return 1
    
    return 0


if __name__ == "__main__":
    exit(main())