|
|
|
|
|
""" |
|
|
Direct Ollama CodeLlama Test - bypassing Docker network limitations |
|
|
""" |
|
|
|
|
|
import asyncio |
|
|
import httpx |
|
|
import json |
|
|
import time |
|
|
|
|
|
async def test_direct_codellama(): |
|
|
"""Test CodeLlama directly for medical entity extraction""" |
|
|
|
|
|
print("🚀 Direct CodeLlama Medical AI Test") |
|
|
print("=" * 40) |
|
|
|
|
|
medical_text = """ |
|
|
MEDICAL RECORD |
|
|
Patient: Sarah Johnson |
|
|
DOB: 1985-09-12 |
|
|
Chief Complaint: Type 2 diabetes follow-up |
|
|
|
|
|
Current Medications: |
|
|
- Metformin 1000mg twice daily |
|
|
- Insulin glargine 15 units at bedtime |
|
|
- Lisinopril 10mg daily for hypertension |
|
|
|
|
|
Vital Signs: |
|
|
- Blood Pressure: 142/88 mmHg |
|
|
- HbA1c: 7.2% |
|
|
- Fasting glucose: 145 mg/dL |
|
|
|
|
|
Assessment: Diabetes with suboptimal control, hypertension |
|
|
""" |
|
|
|
|
|
prompt = f"""You are a medical AI assistant. Extract medical information from this clinical note and return ONLY a JSON response: |
|
|
|
|
|
{medical_text} |
|
|
|
|
|
Return this exact JSON structure: |
|
|
{{ |
|
|
"patient_info": "patient name if found", |
|
|
"conditions": ["list", "of", "conditions"], |
|
|
"medications": ["list", "of", "medications"], |
|
|
"vitals": ["list", "of", "vital", "measurements"], |
|
|
"confidence_score": 0.85 |
|
|
}}""" |
|
|
|
|
|
print("📋 Processing medical text with CodeLlama 13B...") |
|
|
print(f"📄 Input length: {len(medical_text)} characters") |
|
|
|
|
|
start_time = time.time() |
|
|
|
|
|
try: |
|
|
|
|
|
import os |
|
|
ollama_url = os.getenv("OLLAMA_BASE_URL", "http://host.docker.internal:11434") |
|
|
|
|
|
async with httpx.AsyncClient(timeout=30.0) as client: |
|
|
response = await client.post( |
|
|
f"{ollama_url}/api/generate", |
|
|
json={ |
|
|
"model": "codellama:13b-instruct", |
|
|
"prompt": prompt, |
|
|
"stream": False, |
|
|
"options": { |
|
|
"temperature": 0.1, |
|
|
"top_p": 0.9, |
|
|
"num_predict": 1024 |
|
|
} |
|
|
} |
|
|
) |
|
|
|
|
|
if response.status_code == 200: |
|
|
result = response.json() |
|
|
processing_time = time.time() - start_time |
|
|
|
|
|
print(f"✅ CodeLlama processing completed!") |
|
|
print(f"⏱️ Processing time: {processing_time:.2f}s") |
|
|
print(f"🧠 Model: {result.get('model', 'Unknown')}") |
|
|
|
|
|
generated_text = result.get("response", "") |
|
|
print(f"📝 Raw response length: {len(generated_text)} characters") |
|
|
|
|
|
|
|
|
try: |
|
|
json_start = generated_text.find('{') |
|
|
json_end = generated_text.rfind('}') + 1 |
|
|
|
|
|
if json_start >= 0 and json_end > json_start: |
|
|
json_str = generated_text[json_start:json_end] |
|
|
extracted_data = json.loads(json_str) |
|
|
|
|
|
print("\n🏥 EXTRACTED MEDICAL DATA:") |
|
|
print(f" Patient: {extracted_data.get('patient_info', 'N/A')}") |
|
|
print(f" Conditions: {', '.join(extracted_data.get('conditions', []))}") |
|
|
print(f" Medications: {', '.join(extracted_data.get('medications', []))}") |
|
|
print(f" Vitals: {', '.join(extracted_data.get('vitals', []))}") |
|
|
print(f" AI Confidence: {extracted_data.get('confidence_score', 0):.1%}") |
|
|
|
|
|
return True |
|
|
else: |
|
|
print("⚠️ No valid JSON found in response") |
|
|
print(f"Raw response preview: {generated_text[:200]}...") |
|
|
return False |
|
|
|
|
|
except json.JSONDecodeError as e: |
|
|
print(f"❌ JSON parsing failed: {e}") |
|
|
print(f"Raw response preview: {generated_text[:200]}...") |
|
|
return False |
|
|
else: |
|
|
print(f"❌ Ollama API error: {response.status_code}") |
|
|
return False |
|
|
|
|
|
except Exception as e: |
|
|
print(f"💥 Connection failed: {e}") |
|
|
print("💡 Make sure 'ollama serve' is running") |
|
|
return False |
|
|
|
|
|
async def main(): |
|
|
success = await test_direct_codellama() |
|
|
return 0 if success else 1 |
|
|
|
|
|
if __name__ == "__main__": |
|
|
exit_code = asyncio.run(main()) |
|
|
exit(exit_code) |
