fix(mild bug): enhance user query prompts (more robust dealing process with .txt or .json) and add postpartum hemorrhage condition mapping

evaluation/user_query.txt

@@ -17,18 +17,18 @@ Suspected acute ischemic stroke. Tell me the next steps to take
 
 ### 一、Diagnosis-Focused（診斷為主）
 
-1. I have a 68-year-old man with atrial fibrillation presenting with sudden slurred speech and right-sided weakness—what are the possible diagnoses, and how would you evaluate them?
-2. A 40-year-old woman reports fever, urinary frequency, and dysuria—what differential diagnoses should I consider, and which tests would you order?
-3. A 50-year-old patient has progressive dyspnea on exertion and orthopnea over two weeks—what are the likely causes, and what diagnostic steps should I take?
+1. I have a 68-year-old man with atrial fibrillation presenting with sudden slurred speech and right-sided weakness. what are the possible diagnoses, and how would you evaluate them?
+2. A 40-year-old woman reports fever, urinary frequency, and dysuria. what differential diagnoses should I consider, and which tests would you order?
+3. A 50-year-old patient has progressive dyspnea on exertion and orthopnea over two weeks. what are the likely causes, and what diagnostic steps should I take?
 
 ### 二、Treatment-Focused（治療為主）
 
-4. ECG shows a suspected acute STEMI—what immediate interventions should I initiate in the next five minutes?
-5. I have a patient diagnosed with bacterial meningitis—what empiric antibiotic regimen and supportive measures should I implement?
+4. ECG shows a suspected acute STEMI. what immediate interventions should I initiate in the next five minutes?
+5. I have a patient diagnosed with bacterial meningitis. What empiric antibiotic regimen and supportive measures should I implement?
 6. A patient is in septic shock with BP 80/50 mmHg and HR 120 bpm—what fluid resuscitation and vasopressor strategy would you recommend?
 
 ### 三、Mixed（診斷＋治療綜合）
 
 7. A 75-year-old diabetic presents with a non-healing foot ulcer and fever—what differential for osteomyelitis, diagnostic workup, and management plan do you suggest?
-8. A 60-year-old COPD patient has worsening dyspnea and hypercapnia on ABG—how would you confirm the diagnosis, and what is your stepwise treatment approach?
-9. A 28-year-old woman is experiencing postpartum hemorrhage—what are the possible causes, what immediate resuscitation steps should I take, and how would you proceed with definitive management?
+8. A 60-year-old COPD patient has worsening dyspnea and hypercapnia on ABG. How would you confirm the diagnosis, and what is your stepwise treatment approach?
+9. A 28-year-old woman is experiencing postpartum hemorrhage. what are the possible causes, what immediate resuscitation steps should I take, and how would you proceed with definitive management?

src/llm_clients.py

@@ -9,6 +9,8 @@ Date: 2025-07-29
 
 import logging
 import os
+import json
+import re
 from typing import Dict, Optional, Union
 from huggingface_hub import InferenceClient
 from dotenv import load_dotenv
@@ -68,6 +70,91 @@ class llm_Med42_70BClient:
             self.logger.error(f"Detailed Error: {repr(e)}")
             raise ValueError(f"Failed to initialize Medical LLM client: {str(e)}") from e
 
+    def fix_json_formatting(self, response_text: str) -> str:
+        """
+        Fix common JSON formatting errors
+        
+        Args:
+            response_text: Raw response text that may contain JSON errors
+            
+        Returns:
+            Fixed JSON string
+        """
+        # 1. Fix missing commas between key-value pairs
+        # Look for "value" "key" pattern and add comma
+        fixed = re.sub(r'"\s*\n\s*"', '",\n  "', response_text)
+        
+        # 2. Fix missing commas between values and keys
+        fixed = re.sub(r'"\s*(["\[])', '",\1', fixed)
+        
+        # 3. Remove trailing commas
+        fixed = re.sub(r',\s*}', '}', fixed)
+        fixed = re.sub(r',\s*]', ']', fixed)
+        
+        # 4. Ensure string values are properly quoted
+        fixed = re.sub(r':\s*([^",{}\[\]]+)\s*([,}])', r': "\1"\2', fixed)
+        
+        return fixed
+
+    def parse_medical_response(self, response_text: str) -> Dict:
+        """
+        Enhanced JSON parsing logic with error recovery
+        
+        Args:
+            response_text: Raw response text from Med42-70B
+            
+        Returns:
+            Parsed response dictionary
+        """
+        try:
+            return json.loads(response_text)
+        except json.JSONDecodeError as e:
+            self.logger.warning(f"Initial JSON parsing failed: {e}")
+            
+            # Attempt to fix common JSON errors
+            try:
+                fixed_response = self.fix_json_formatting(response_text)
+                self.logger.info("Attempting to parse fixed JSON")
+                return json.loads(fixed_response)
+            except json.JSONDecodeError as e2:
+                self.logger.error(f"Fixed JSON parsing also failed: {e2}")
+                
+                # Try to extract partial information
+                try:
+                    return self.extract_partial_medical_info(response_text)
+                except:
+                    # Final fallback format
+                    return {
+                        "extracted_condition": "parsing_error",
+                        "confidence": "0.0",
+                        "is_medical": True,
+                        "raw_response": response_text,
+                        "error": str(e)
+                    }
+
+    def extract_partial_medical_info(self, response_text: str) -> Dict:
+        """
+        Extract partial medical information from malformed response
+        
+        Args:
+            response_text: Malformed response text
+            
+        Returns:
+            Dictionary with extracted information
+        """
+        # Try to extract condition
+        condition_match = re.search(r'"extracted_condition":\s*"([^"]*)"', response_text)
+        confidence_match = re.search(r'"confidence":\s*"([^"]*)"', response_text)
+        medical_match = re.search(r'"is_medical":\s*(true|false)', response_text)
+        
+        return {
+            "extracted_condition": condition_match.group(1) if condition_match else "unknown",
+            "confidence": confidence_match.group(1) if confidence_match else "0.0",
+            "is_medical": medical_match.group(1) == "true" if medical_match else True,
+            "raw_response": response_text,
+            "parsing_method": "partial_extraction"
+        }
+
     def analyze_medical_query(
         self, 
         query: str, 
@@ -138,6 +225,27 @@ class llm_Med42_70BClient:
             self.logger.info(f"Raw LLM Response: {response_text}")
             self.logger.info(f"Query Latency: {latency:.4f} seconds")
             
+            # Enhanced response parsing - handle both JSON and text formats
+            try:
+                # Try to parse as JSON first (in case API returns JSON)
+                parsed_response = self.parse_medical_response(response_text)
+                
+                # If it's a valid JSON response, extract condition from it
+                if isinstance(parsed_response, dict) and 'extracted_condition' in parsed_response:
+                    extracted_condition = parsed_response.get('extracted_condition', '')
+                    confidence = parsed_response.get('confidence', '0.8')
+                    self.logger.info(f"Parsed JSON response - Condition: {extracted_condition}")
+                else:
+                    # Fallback to text extraction
+                    extracted_condition = self._extract_condition(response_text)
+                    confidence = '0.8'
+                    
+            except Exception as parse_error:
+                self.logger.warning(f"Response parsing failed: {parse_error}")
+                # Fallback to text extraction
+                extracted_condition = self._extract_condition(response_text)
+                confidence = '0.8'
+            
             # Detect abnormal response
             if self._is_abnormal_response(response_text):
                 self.logger.error(f"❌ Abnormal LLM response detected: {response_text[:50]}...")
@@ -149,15 +257,12 @@ class llm_Med42_70BClient:
                     'latency': latency
                 }
             
-            # Extract condition from response
-            extracted_condition = self._extract_condition(response_text)
-            
             # Log the extracted condition
             self.logger.info(f"Extracted Condition: {extracted_condition}")
             
             return {
                 'extracted_condition': extracted_condition,
-                'confidence': '0.8',
+                'confidence': confidence,
                 'raw_response': response_text,
                 'latency': latency  # Add latency to the return dictionary
             }

src/medical_conditions.py

@@ -63,6 +63,10 @@ CONDITION_KEYWORD_MAPPING: Dict[str, Dict[str, str]] = {
     "seizure disorder": {
         "emergency": "seizure|status epilepticus|postictal state",
         "treatment": "antiepileptic drugs|EEG monitoring|neurology consult"
+    },
+    "postpartum_hemorrhage": {
+    "emergency": "postpartum hemorrhage|uterine atony|placental retention|vaginal laceration",
+    "treatment": "uterine massage|IV oxytocin infusion|blood transfusion|surgical intervention"
     }
 }