Spaces:
Sleeping
Sleeping
File size: 16,727 Bytes
847b628 2b376ce 847b628 2b376ce 847b628 2b376ce 847b628 2b376ce f944c35 2b376ce 3a8e6a8 2b376ce 6c121ca 2b376ce f944c35 2b376ce f944c35 2b376ce f944c35 2b376ce 0ce8c2d 2b376ce |
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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 |
import gradio as gr
import json
import zipfile
from pathlib import Path
import pandas as pd
from typing import Dict, List, Tuple
import random
class MedQADatabase:
"""Handler for MedQA and Med-Gemini databases"""
def __init__(self, zip_path="medqa_databases.zip"):
self.data = {
'medgemini': [],
'medqa_train': [],
'medqa_dev': [],
'medqa_test': []
}
self.load_databases(zip_path)
def load_databases(self, zip_path):
"""Load all databases from the ZIP file"""
print("π¦ Loading databases from ZIP...")
try:
with zipfile.ZipFile(zip_path, 'r') as zip_ref:
# Extract to temporary directory
zip_ref.extractall('temp_data')
# Load Med-Gemini
medgemini_path = Path('temp_data/medqa_databases/med_gemini/medqa_relabelling.json')
if medgemini_path.exists():
with open(medgemini_path, 'r', encoding='utf-8') as f:
self.data['medgemini'] = json.load(f)
print(f"β
Loaded {len(self.data['medgemini'])} Med-Gemini questions")
# Load MedQA splits
medqa_base = Path('temp_data/medqa_databases/medqa_original')
for split in ['train', 'dev', 'test']:
split_path = medqa_base / f"{split}.json"
if split_path.exists():
with open(split_path, 'r', encoding='utf-8') as f:
self.data[f'medqa_{split}'] = json.load(f)
print(f"β
Loaded {len(self.data[f'medqa_{split}'])} MedQA {split} questions")
except Exception as e:
print(f"β Error loading databases: {e}")
raise
def get_stats(self) -> str:
"""Get database statistics"""
stats = "## π Database Statistics\n\n"
stats += f"**Med-Gemini**: {len(self.data['medgemini']):,} questions\n\n"
stats += f"**MedQA Original**:\n"
stats += f"- Training: {len(self.data['medqa_train']):,} questions\n"
stats += f"- Development: {len(self.data['medqa_dev']):,} questions\n"
stats += f"- Test: {len(self.data['medqa_test']):,} questions\n"
stats += f"- **Total**: {sum(len(self.data[f'medqa_{s}']) for s in ['train', 'dev', 'test']):,} questions\n\n"
stats += f"**Grand Total**: {sum(len(v) for v in self.data.values()):,} questions"
return stats
def get_question(self, dataset: str, index: int) -> Dict:
"""Get a specific question from a dataset"""
try:
return self.data[dataset][index]
except (KeyError, IndexError):
return None
def search_questions(self, query: str, dataset: str = 'all', max_results: int = 50) -> List[Tuple[str, int, str]]:
"""Search questions by keyword"""
results = []
query_lower = query.lower()
datasets_to_search = list(self.data.keys()) if dataset == 'all' else [dataset]
for ds in datasets_to_search:
for idx, q in enumerate(self.data[ds]):
# Search in question text
question_text = q.get('question', q.get('Question', ''))
if query_lower in question_text.lower():
preview = question_text[:100] + "..." if len(question_text) > 100 else question_text
results.append((ds, idx, preview))
if len(results) >= max_results:
return results
return results
# Initialize database
print("π Initializing MedQA Explorer...")
db = MedQADatabase()
# ============================================================================
# GRADIO INTERFACE FUNCTIONS
# ============================================================================
def format_question_display(question_data: Dict, dataset: str) -> str:
"""Format question data for display"""
if not question_data:
return "β Question not found"
# Handle different data formats
if dataset == 'medgemini':
return format_medgemini_question(question_data)
else:
return format_medqa_question(question_data)
def format_medgemini_question(q: Dict) -> str:
"""Format Med-Gemini question"""
html = f"""
<div style="background: linear-gradient(135deg, #667eea 0%, #764ba2 100%); padding: 20px; border-radius: 10px; margin-bottom: 20px;">
<h2 style="color: white; margin: 0;">π¬ Med-Gemini Question</h2>
</div>
<div style="background: #f8f9fa; padding: 20px; border-radius: 8px; margin-bottom: 20px;">
<h3>π Question</h3>
<p style="font-size: 16px; line-height: 1.6;">{q.get('question', 'N/A')}</p>
</div>
<div style="background: #fff; padding: 20px; border-radius: 8px; margin-bottom: 20px; border: 2px solid #e0e0e0;">
<h3>π€ Answer Options</h3>
"""
# Display options
options = q.get('options', {})
correct_answer = q.get('answer_idx', 'N/A')
option_labels = ['A', 'B', 'C', 'D', 'E']
for label in option_labels:
option_key = f'opa' if label == 'A' else f'op{label.lower()}'
if option_key in options:
is_correct = (label == correct_answer)
color = '#d4edda' if is_correct else '#fff'
icon = 'β
' if is_correct else 'β'
html += f"""
<div style="background: {color}; padding: 12px; margin: 8px 0; border-radius: 5px; border: 1px solid #ccc;">
{icon} <strong>{label}.</strong> {options[option_key]}
</div>
"""
html += "</div>"
# Show correct answer
html += f"""
<div style="background: #d4edda; padding: 15px; border-radius: 8px; margin-bottom: 20px; border-left: 4px solid #28a745;">
<h3 style="margin-top: 0;">β
Correct Answer</h3>
<p style="font-size: 18px; font-weight: bold; margin: 0;">{correct_answer}</p>
</div>
"""
# Show explanation if available
explanation = q.get('explanation', q.get('Explanation', ''))
if explanation:
html += f"""
<div style="background: #e7f3ff; padding: 20px; border-radius: 8px; border-left: 4px solid #2196F3;">
<h3 style="margin-top: 0;">π‘ Explanation</h3>
<p style="line-height: 1.6;">{explanation}</p>
</div>
"""
return html
def format_medqa_question(q: Dict) -> str:
"""Format MedQA original question"""
html = f"""
<div style="background: linear-gradient(135deg, #f093fb 0%, #f5576c 100%); padding: 20px; border-radius: 10px; margin-bottom: 20px;">
<h2 style="color: white; margin: 0;">π MedQA USMLE Question</h2>
</div>
<div style="background: #f8f9fa; padding: 20px; border-radius: 8px; margin-bottom: 20px;">
<h3>π Question</h3>
<p style="font-size: 16px; line-height: 1.6;">{q.get('question', 'N/A')}</p>
</div>
<div style="background: #fff; padding: 20px; border-radius: 8px; margin-bottom: 20px; border: 2px solid #e0e0e0;">
<h3>π€ Answer Options</h3>
"""
# Display options
options = q.get('options', {})
correct_answer = q.get('answer_idx', 'N/A')
for key, value in options.items():
label = key.replace('op', '').upper() if key.startswith('op') else key
is_correct = (label == correct_answer)
color = '#d4edda' if is_correct else '#fff'
icon = 'β
' if is_correct else 'β'
html += f"""
<div style="background: {color}; padding: 12px; margin: 8px 0; border-radius: 5px; border: 1px solid #ccc;">
{icon} <strong>{label}.</strong> {value}
</div>
"""
html += "</div>"
# Show correct answer
html += f"""
<div style="background: #d4edda; padding: 15px; border-radius: 8px; margin-bottom: 20px; border-left: 4px solid #28a745;">
<h3 style="margin-top: 0;">β
Correct Answer</h3>
<p style="font-size: 18px; font-weight: bold; margin: 0;">{correct_answer}</p>
</div>
"""
# Show metamap if available
metamap = q.get('metamap_phrases')
if metamap:
html += f"""
<div style="background: #fff3cd; padding: 15px; border-radius: 8px; border-left: 4px solid #ffc107;">
<h3 style="margin-top: 0;">π₯ Medical Concepts (MetaMap)</h3>
<p style="line-height: 1.6;">{', '.join(metamap)}</p>
</div>
"""
return html
def browse_questions(dataset: str, index: int) -> Tuple[str, str]:
"""Browse questions by index"""
total = len(db.data.get(dataset, []))
if total == 0:
return "β No questions in this dataset", f"Dataset: {dataset} (empty)"
# Clamp index to valid range
index = max(0, min(index, total - 1))
question = db.get_question(dataset, index)
html = format_question_display(question, dataset)
info = f"π Question {index + 1} of {total} | Dataset: {dataset}"
return html, info
def random_question(dataset: str) -> Tuple[str, str, int]:
"""Get a random question"""
total = len(db.data.get(dataset, []))
if total == 0:
return "β No questions in this dataset", f"Dataset: {dataset} (empty)", 0
index = random.randint(0, total - 1)
question = db.get_question(dataset, index)
html = format_question_display(question, dataset)
info = f"π² Random Question {index + 1} of {total} | Dataset: {dataset}"
return html, info, index
def search_interface(query: str, dataset: str) -> str:
"""Search interface"""
if not query.strip():
return "π‘ Enter a search query to find questions"
results = db.search_questions(query, dataset)
if not results:
return f"β No results found for '{query}' in {dataset}"
html = f"""
<div style="background: linear-gradient(135deg, #667eea 0%, #764ba2 100%); padding: 20px; border-radius: 10px; margin-bottom: 20px;">
<h2 style="color: white; margin: 0;">π Search Results: "{query}"</h2>
<p style="color: white; margin: 5px 0 0 0;">Found {len(results)} results in {dataset}</p>
</div>
"""
for ds, idx, preview in results[:20]: # Show top 20
dataset_name = ds.replace('_', ' ').title()
html += f"""
<div style="background: #fff; padding: 15px; margin: 10px 0; border-radius: 8px; border-left: 4px solid #667eea;">
<p style="margin: 0; color: #666; font-size: 12px;"><strong>{dataset_name}</strong> - Question #{idx + 1}</p>
<p style="margin: 5px 0 0 0;">{preview}</p>
</div>
"""
if len(results) > 20:
html += f"<p>... and {len(results) - 20} more results</p>"
return html
# ============================================================================
# GRADIO APP
# ============================================================================
with gr.Blocks(theme=gr.themes.Soft(), title="MedQA Database Explorer") as app:
gr.Markdown("""
# π₯ MedQA Database Explorer
Explore medical question-answering databases including **Med-Gemini** and **MedQA USMLE**.
""")
# Statistics
with gr.Accordion("π Database Statistics", open=False):
gr.Markdown(db.get_stats())
# Main interface
with gr.Tabs():
# Browse Tab
with gr.Tab("π Browse Questions"):
with gr.Row():
with gr.Column(scale=1):
dataset_dropdown = gr.Dropdown(
choices=['medgemini', 'medqa_train', 'medqa_dev', 'medqa_test'],
value='medgemini',
label="Select Database"
)
question_slider = gr.Slider(
minimum=0,
maximum=len(db.data['medgemini']) - 1,
value=0,
step=1,
label="Question Number"
)
with gr.Row():
prev_btn = gr.Button("β¬
οΈ Previous", size="sm")
random_btn = gr.Button("π² Random", size="sm", variant="primary")
next_btn = gr.Button("Next β‘οΈ", size="sm")
info_text = gr.Textbox(label="Info", interactive=False)
with gr.Column(scale=2):
question_display = gr.HTML()
# Update slider max when dataset changes
def update_slider(dataset):
max_val = len(db.data.get(dataset, [])) - 1
return gr.Slider(maximum=max_val, value=0)
dataset_dropdown.change(
fn=update_slider,
inputs=[dataset_dropdown],
outputs=[question_slider]
)
# Browse functions
def show_question(dataset, index):
return browse_questions(dataset, int(index))
question_slider.change(
fn=show_question,
inputs=[dataset_dropdown, question_slider],
outputs=[question_display, info_text]
)
dataset_dropdown.change(
fn=show_question,
inputs=[dataset_dropdown, question_slider],
outputs=[question_display, info_text]
)
# Navigation buttons
def prev_question(dataset, index):
new_index = max(0, int(index) - 1)
html, info = browse_questions(dataset, new_index)
return html, info, new_index
def next_question(dataset, index):
max_idx = len(db.data.get(dataset, [])) - 1
new_index = min(max_idx, int(index) + 1)
html, info = browse_questions(dataset, new_index)
return html, info, new_index
prev_btn.click(
fn=prev_question,
inputs=[dataset_dropdown, question_slider],
outputs=[question_display, info_text, question_slider]
)
next_btn.click(
fn=next_question,
inputs=[dataset_dropdown, question_slider],
outputs=[question_display, info_text, question_slider]
)
random_btn.click(
fn=random_question,
inputs=[dataset_dropdown],
outputs=[question_display, info_text, question_slider]
)
# Load first question on start
app.load(
fn=show_question,
inputs=[dataset_dropdown, question_slider],
outputs=[question_display, info_text]
)
# Search Tab
with gr.Tab("π Search"):
with gr.Row():
search_query = gr.Textbox(
label="Search Query",
placeholder="Enter keywords (e.g., 'diabetes', 'heart failure', 'treatment')...",
scale=3
)
search_dataset = gr.Dropdown(
choices=['all', 'medgemini', 'medqa_train', 'medqa_dev', 'medqa_test'],
value='all',
label="Search In",
scale=1
)
search_btn = gr.Button("π Search", variant="primary")
search_results = gr.HTML()
search_btn.click(
fn=search_interface,
inputs=[search_query, search_dataset],
outputs=[search_results]
)
# Also search on Enter key
search_query.submit(
fn=search_interface,
inputs=[search_query, search_dataset],
outputs=[search_results]
)
gr.Markdown("""
---
### π About the Databases
**Med-Gemini**: Expert-relabeled medical questions with detailed explanations from Google's Med-Gemini project.
**MedQA**: Original USMLE-style medical questions from the MedQA dataset.
### π Sources
- [Med-Gemini Paper](https://arxiv.org/abs/2404.18416)
- [MedQA Dataset](https://github.com/jind11/MedQA)
""")
if __name__ == "__main__":
app.launch() |