Implement the core eligibility logic natively in Python.

api/models.py

@@ -111,7 +111,7 @@ class HBVPatientInput(BaseModel):
     hbv_dna_level: float = Field(..., description="HBV DNA level in IU/mL", ge=0)
     hbeag_status: str = Field(..., description="HBeAg status: Positive / Negative")
     alt_level: float = Field(..., description="ALT level in U/L", ge=0)
-    fibrosis_stage: str = Field(..., description="Fibrosis/Cirrhosis stage: F0-F1 / F2-F3 / F4")
+    fibrosis_stage: str = Field(..., description="Fibrosis/Cirrhosis stage: F0 / F1 / F2 / F3 / F4")
     necroinflammatory_activity: str = Field(..., description="Necroinflammatory activity: A0 / A1 / A2 / A3")
     extrahepatic_manifestations: bool = Field(..., description="Presence of extrahepatic manifestations")
     immunosuppression_status: Optional[str] = Field(None, description="Immunosuppression status: None / Chemotherapy / Other")

api/routers/hbv_assessment.py

@@ -72,7 +72,7 @@ async def assess_patient_from_text(text_input: TextAssessmentInput) -> HBVAssess
     HBV DNA: 5000 IU/mL
     HBeAg: Positive
     ALT: 80 U/L
-    Fibrosis stage: F2-F3
+    Fibrosis stage: F2
     Necroinflammatory activity: A2
     No extrahepatic manifestations
     No immunosuppression

core/hbv_assessment.py

@@ -5,7 +5,7 @@ Evaluates patient eligibility for HBV treatment according to SASLT 2021 guidelin
 import logging
 import json
 import re
-from typing import Dict, Any
+from typing import Dict, Any, Tuple
 from .config import get_llm
 
 logger = logging.getLogger(__name__)
@@ -203,6 +203,88 @@ Liver fibrosis is usually classified into five stages:
 F0—no fibrosis; F1—mild fibrosis, pericellular collagen deposits; F2—moderate fibrosis, beginning bridging fibrosis; F3—severe fibrosis, defined as presence of numerous bridges and septa; F4—cirrhosis
 """
 
+def check_eligibility_criteria(patient_data: Dict[str, Any]) -> Tuple[bool, str]:
+    """
+    Deterministically check patient eligibility based on SASLT 2021 guidelines.
+    Returns: (is_eligible, rationale_message)
+    """
+    hbsag_status = patient_data.get("hbsag_status", "N/A")
+    hbv_dna_numeric = patient_data.get("hbv_dna_level_numeric", 0.0) # Using numeric derived value
+    hbeag_status = patient_data.get("hbeag_status", "N/A")
+    alt_level = patient_data.get("alt_level", 0)
+    fibrosis_stage = patient_data.get("fibrosis_stage", "F0")
+    necroinflammatory = patient_data.get("necroinflammatory_activity", "A0")
+    age = patient_data.get("age", 0)
+    extrahepatic = patient_data.get("extrahepatic_manifestations", False)
+    immunosuppression = patient_data.get("immunosuppression_status", "None")
+    coinfections = patient_data.get("coinfections", [])
+    family_history = patient_data.get("family_history_cirrhosis_hcc", False)
+    pregnancy_status = patient_data.get("pregnancy_status", "No")
+    
+    ULN = 40 # Based on KEY DEFINITIONS
+
+    # 0. Basic Requirement Check
+    if hbsag_status != "Positive":
+        return False, "Not eligible: HBsAg negative. HBsAg positivity is required for HBV treatment consideration [SASLT 2021, Page 6]."
+    
+    # --- 1. Special Populations (CRITICAL ELIGIBILITY HIERARCHY: FIRST) ---
+    
+    # 1.1 Immunosuppression (Page 9)
+    if immunosuppression != "None":
+        return True, f"Eligible for prophylaxis: HBsAg-positive patient receiving immunosuppressive therapy/chemotherapy (Grade A) [SASLT 2021, Page 9]."
+
+    # 1.2 HBV-HIV Coinfection (Page 9)
+    if "HIV" in coinfections:
+        return True, "Eligible for ART: HBV-HIV coinfection requires TDF- or TAF-based ART regimen (Grade A) [SASLT 2021, Page 9]."
+
+    # 1.3 Pregnancy (Page 10)
+    if pregnancy_status == "Yes" and hbv_dna_numeric > 100000:
+        return True, "Eligible for prophylaxis: Pregnant woman with HBV DNA > 100,000 IU/mL started at 24-28 weeks (Grade D) [SASLT 2021, Page 10]."
+
+    # --- 2. Standard Criteria (CRITICAL ELIGIBILITY HIERARCHY: SECOND) ---
+    
+    # Helper checks
+    is_cirrhotic = fibrosis_stage == "F4"
+    has_moderate_fibrosis_or_necroinflammation = fibrosis_stage in ["F2", "F3", "F4"] or necroinflammatory in ["A2", "A3"]
+    is_alt_elevated = alt_level > ULN
+    is_alt_doubly_elevated = alt_level > (2 * ULN)
+    
+    # 2.1 Cirrhosis (F4) (Grade A)
+    if is_cirrhotic:
+        return True, "Eligible: Patient has cirrhosis (F4) regardless of ALT and HBV DNA level (Grade A) [SASLT 2021, Page 6]."
+    
+    # 2.2 Severe Viremia/ALT (HBV DNA > 20,000 & ALT > 2xULN) (Grade B)
+    if hbv_dna_numeric > 20000 and is_alt_doubly_elevated:
+        return True, "Eligible: HBV DNA > 20,000 IU/mL and ALT > 2xULN (Grade B) [SASLT 2021, Page 6]."
+
+    # 2.3 Active CHB (HBV DNA > 2,000 & ALT > ULN & F2/A2+) (Grade A) - Main Criterion
+    if hbv_dna_numeric > 2000 and is_alt_elevated and has_moderate_fibrosis_or_necroinflammation:
+        return True, "Eligible: HBV DNA > 2,000 IU/mL, ALT > ULN, and at least moderate fibrosis/necroinflammation (Grade A) [SASLT 2021, Page 6]."
+    
+    # 2.4 HBeAg+ > 30 years (HBeAg+, Normal ALT, High DNA) (Grade D)
+    if hbeag_status == "Positive" and age > 30 and not is_alt_elevated and hbv_dna_numeric > 2000:
+        return True, "Eligible: HBeAg-positive chronic infection in patient > 30 years despite normal ALT (Grade D) [SASLT 2021, Page 6]."
+
+    # 2.5 Viremia and Moderate Fibrosis (HBV DNA > 2,000 & Moderate Fibrosis) (Implicit from Page 6)
+    # "Patients with HBV DNA >2,000 IU/mL and at least moderate fibrosis may initiate treatment even if ALT levels are normal."
+    if hbv_dna_numeric > 2000 and fibrosis_stage in ["F2", "F3", "F4"]:
+         return True, "Eligible: HBV DNA > 2,000 IU/mL and at least moderate fibrosis (F2+) (Grade D/Implicit) [SASLT 2021, Page 6]."
+
+    # 2.6 Family History/Extrahepatic (HBV DNA > 2,000 & ALT > ULN & History/Extrahepatic) (Grade D)
+    if hbv_dna_numeric > 2000 and is_alt_elevated and (family_history or extrahepatic):
+        return True, "Eligible: HBV DNA > 2,000 IU/mL, ALT > ULN, and family history of HCC/cirrhosis or extrahepatic manifestations (Grade D) [SASLT 2021, Page 6]."
+
+    # --- 3. Not Eligible ---
+    
+    # Determine reason for non-eligibility for monitoring purposes
+    if hbeag_status == "Negative" and hbv_dna_numeric < 2000:
+        rationale = "Not eligible: HBV DNA < 2,000 IU/mL and ALT ≤ ULN; monitoring indicated [SASLT 2021, Page 6]."
+    elif not is_alt_elevated and not has_moderate_fibrosis_or_necroinflammation:
+        rationale = "Not eligible: ALT ≤ ULN and lack of significant fibrosis/necroinflammation; monitoring indicated [SASLT 2021, Page 6]."
+    else:
+        rationale = "Not eligible: Criteria not fully met (e.g., HBV DNA < 2,000 IU/mL) [SASLT 2021, Page 6]."
+        
+    return False, rationale
 
 
 def assess_hbv_eligibility(patient_data: Dict[str, Any]) -> Dict[str, Any]:
@@ -219,26 +301,44 @@ def assess_hbv_eligibility(patient_data: Dict[str, Any]) -> Dict[str, Any]:
         - recommendations: str (comprehensive narrative with inline citations in format [SASLT 2021, Page X])
     """
     try:
-        # # Check if HBsAg is positive (required for treatment consideration)
-        # if patient_data.get("hbsag_status") != "Positive":
-        #     return {
-        #         "eligible": False,
-        #         "recommendations": "Patient is HBsAg negative. HBV treatment is not indicated. HBsAg positivity is required for HBV treatment consideration according to SASLT 2021 guidelines."
-        #     }
-        
         # Use hardcoded SASLT 2021 guidelines instead of RAG retrieval
         logger.info("Using hardcoded SASLT 2021 guidelines (Pages 3, 4, 6, 7, 8, 9, 10)")
         
-        # Define ALT ULN for context
-        sex = patient_data.get("sex", "Male")
-        #alt_uln = 35 if sex == "Male" else 25
+        # --- Pre-processing for numeric comparison ---
+        hbv_dna = patient_data.get("hbv_dna_level", 0)
+        hbv_dna_numeric = hbv_dna
+        if isinstance(hbv_dna_numeric, str):
+            try:
+                cleaned = re.sub(r"[^\d\.]", "", hbv_dna_numeric)
+                # Attempt to parse as float, handling empty string from cleaning
+                hbv_dna_numeric = float(cleaned) if cleaned else 0.0
+            except Exception:
+                hbv_dna_numeric = 0.0
+        try:
+            hbv_dna_numeric = float(hbv_dna_numeric)
+        except (TypeError, ValueError):
+            hbv_dna_numeric = 0.0
+        
+        patient_data["hbv_dna_level_numeric"] = hbv_dna_numeric # Add numeric value to patient data
+        
+        if hbv_dna_numeric > 2000:
+            hbv_dna_2000_comparison = ">"
+        elif hbv_dna_numeric < 2000:
+            hbv_dna_2000_comparison = "<"
+        else:
+            hbv_dna_2000_comparison = "="
+        patient_data["hbv_dna_2000_comparison"] = hbv_dna_2000_comparison
+        logger.info(f"HBV DNA vs 2000 IU/mL comparison: {hbv_dna_numeric} {hbv_dna_2000_comparison} 2000")
+        
+        # --- Deterministic Eligibility Check ---
+        is_eligible, rationale_message = check_eligibility_criteria(patient_data)
         
-        # Format patient data for prompt
+        # --- Prepare LLM Prompt with Deterministic Result ---
+        sex = patient_data.get("sex", "Male")
         age = patient_data.get("age", "N/A")
         pregnancy_status = patient_data.get("pregnancy_status", "N/A")
         hbsag_status = patient_data.get("hbsag_status", "N/A")
         duration_hbsag = patient_data.get("duration_hbsag_months", "N/A")
-        hbv_dna = patient_data.get("hbv_dna_level", 0)
         hbeag_status = patient_data.get("hbeag_status", "N/A")
         alt_level = patient_data.get("alt_level", 0)
         fibrosis_stage = patient_data.get("fibrosis_stage", "N/A")
@@ -248,8 +348,11 @@ def assess_hbv_eligibility(patient_data: Dict[str, Any]) -> Dict[str, Any]:
         coinfections = patient_data.get("coinfections", [])
         family_history = patient_data.get("family_history_cirrhosis_hcc", False)
         comorbidities = patient_data.get("other_comorbidities", [])
-        
-        # Create prompt for LLM to analyze patient against guidelines
+
+        # Crucial addition: Pass the determined status and rationale to the LLM
+        DETERMINISTIC_STATUS = "Eligible: true" if is_eligible else "Eligible: false"
+        DETERMINISTIC_RATIONALE = rationale_message
+
         analysis_prompt = f"""You are an HBV treatment eligibility assessment system. Analyze the patient data against the SASLT 2021 guidelines.
 PATIENT DATA:
 - Sex: {sex}
@@ -257,9 +360,10 @@ PATIENT DATA:
 - Pregnancy Status: {pregnancy_status}
 - HBsAg Status: {hbsag_status}
 - HBsAg Duration: {duration_hbsag} months
-- HBV DNA Level: {hbv_dna:,.0f} IU/mL
+- HBV DNA Level: {hbv_dna} IU/mL
+- HBV DNA vs 2000 IU/mL: {hbv_dna_2000_comparison}
 - HBeAg Status: {hbeag_status}
-- ALT Level: {alt_level} 
+- ALT Level: {alt_level}
 - Fibrosis Stage: {fibrosis_stage}
 - Necroinflammatory Activity: {necroinflammatory}
 - Extrahepatic Manifestations: {extrahepatic}
@@ -268,14 +372,21 @@ PATIENT DATA:
 - Family History (Cirrhosis/HCC): {family_history}
 - Other Comorbidities: {', '.join(comorbidities) if comorbidities else 'None'}
 
+DETERMINISTIC ELIGIBILITY RESULT (LLM MUST USE THIS EXACT STATUS):
+{DETERMINISTIC_STATUS}
+Primary Rationale: {DETERMINISTIC_RATIONALE}
+
+NUMERIC RULES (MUST FOLLOW EXACTLY):
+- Compare integers exactly as written (e.g., 1800 < 2000 is TRUE; 1800 > 2000 is FALSE).
+- Respect inequality signs from the guidelines (>, >=, <, <=) without relaxing or tightening thresholds.
 
 SASLT 2021 GUIDELINES (Retrieved Context):
 {SASLT_GUIDELINES}
-Based STRICTLY on the SASLT 2021 guidelines and criteria provided above, assess this patient's eligibility for HBV antiviral treatment.
+Based STRICTLY on the SASLT 2021 guidelines and the DETERMINISTIC ELIGIBILITY RESULT, complete the JSON response.
 You MUST respond with a valid JSON object in this exact format:
 {{
-  "eligible": true or false,
-  "recommendations": "Comprehensive assessment with inline citations"
+  "eligible": {DETERMINISTIC_STATUS.split(': ')[1]},
+  "recommendations": "Comprehensive assessment with inline citations"
 }}
 IMPORTANT JSON FORMATTING:
 - Return ONLY valid JSON without markdown code blocks.
@@ -283,25 +394,16 @@ IMPORTANT JSON FORMATTING:
 - Do NOT include literal newline characters. Use \\n for every new bullet or line.
 - Use SINGLE \\n between lines. Do NOT use \\n\\n (double newlines) anywhere.
 
-CRITICAL ELIGIBILITY HIERARCHY:
-1.  **Check Special Populations FIRST (Pages 9-10):** "Eligible" means eligible for ANY antiviral intervention (treatment OR prophylaxis).
-    * If patient is **HBsAg-positive** AND **Immunosuppressed** (Page 9): Set **"eligible": true"** (for prophylaxis).
-    * If patient has **HBV-HIV coinfection** (Page 9): Set **"eligible": true"** (for ART).
-    * If patient is **Pregnant** with **HBV DNA > 100,000 IU/mL** (Page 10): Set **"eligible": true"** (for prophylaxis).
-    * If patient has **HBV-HCV coinfection** AND meets HBV treatment criteria (Page 9): Set **"eligible": true"**.
-2.  **Check Standard Criteria SECOND (Page 6):**
-    * If *not* eligible based on special populations, check standard criteria (HBV DNA, ALT, fibrosis, age, family history).
-    * If any Page 6 criteria are met (e.g., Cirrhosis, or HBV DNA > 2,000 + ALT > ULN + moderate fibrosis, etc.): Set **"eligible": true"**.
-3.  **If NO criteria from (1) or (2) are met:** Set **"eligible": false"**.
-4.  The "eligible" flag MUST be consistent with the "Eligibility and Rationale" bullet. Do not contradict yourself.
+CRITICAL INSTRUCTION:
+The value of the "eligible" field MUST match the DETERMINISTIC ELIGIBILITY RESULT above.
 
 STRUCTURE AND CONTENT OF "recommendations" (CONCISE & ORGANIZED):
 - Use ONLY these sections in this exact order, each as a header followed by 1-3 concise bullets:
-  1. "Eligibility and Rationale:" (1-2 bullets max, MUST state the primary reason for eligibility/ineligibility)
-  2. "Treatment Recommendations:" (1-3 bullets: first-line drugs if eligible, or "Treatment not indicated" if not eligible)
-  3. "Monitoring and Follow-up:" (1-2 bullets)
-  4. "Special Considerations:" (0-2 bullets, ONLY if applicable and *not* the primary reason for eligibility)
-  5. "References:" (1 line listing pages cited)
+  1. "Eligibility and Rationale:" (1-2 bullets max, **The first bullet MUST incorporate the Primary Rationale provided above.**)
+  2. "Treatment Recommendations:" (1-3 bullets: first-line drugs if eligible, or "Treatment not indicated" if not eligible)
+  3. "Monitoring and Follow-up:" (1-2 bullets)
+  4. "Special Considerations:" (0-2 bullets, ONLY if applicable and *not* the primary reason for eligibility)
+  5. "References:" (1 line listing pages cited)
 - OMIT "Additional Notes" and any other sections.
 - Keep each bullet to ONE sentence (max 25 words per bullet).
 - Total output: aim for 8-12 bullets maximum across all sections.
@@ -312,8 +414,7 @@ BULLETING AND CITATIONS RULES:
 - Only cite pages 6–10 that actually contain the information.
 
 STRICT ACCURACY AND CONSISTENCY RULES:
-- **NO CONTRADICTIONS:** The "eligible" flag MUST match the rationale. Follow the CRITICAL ELIGIBILITY HIERARCHY.
-- **Rationale First:** The Rationale MUST state the primary reason. If eligible due to immunosuppression, state that, even if Page 6 criteria are not met.
+- **NO CONTRADICTIONS:** The "eligible" flag MUST match the rationale derived in the code.
 - **Use ONLY the provided SASLT 2021 content;** do NOT add external knowledge.
 - **BREVITY:** Each bullet = 1 sentence, max 25 words. Total = 8-12 bullets max.
 
@@ -323,39 +424,6 @@ PAGE-TO-TOPIC MAPPING GUIDANCE (for correct citations):
 - Page 8: Treatment drugs/regimens (ETV, TDF, TAF), agents not recommended.
 - Page 9: Special populations (HBV-HCV, HBV-HDV, HBV-HIV, Immunocompromised).
 - Page 10: Pregnancy-related recommendations.
-
----
-EXAMPLE OUTPUT (ELIGIBLE - STANDARD CRITERIA) - Use SINGLE \\n only:
-{{
-  "eligible": true,
-  "recommendations": "Eligibility and Rationale:\\n- Eligible: HBV DNA > 2,000 IU/mL, ALT > ULN, moderate fibrosis (Grade A) [SASLT 2021, Page 6]\\nTreatment Recommendations:\\n- Start monotherapy with ETV, TDF, or TAF (Grade A) [SASLT 2021, Page 8]\\nMonitoring and Follow-up:\\n- Monitor treatment response per SASLT protocol [SASLT 2021, Page 7]\\nReferences:\\n- Pages 6, 7, 8: Treatment criteria, drugs, monitoring"
-}}
----
-EXAMPLE OUTPUT (NOT ELIGIBLE - STANDARD CRITERIA) - Use SINGLE \\n only:
-{{
-  "eligible": false,
-  "recommendations": "Eligibility and Rationale:\\n- Not eligible: HBV DNA < 2,000 IU/mL, ALT ≤ ULN, no significant fibrosis [SASLT 2021, Page 6]\\nTreatment Recommendations:\\n- Treatment not indicated at this time [SASLT 2021, Page 6]\\nMonitoring and Follow-up:\\n- Monitor every 6-12 months (HBeAg-negative, HBV DNA < 2,000 IU/mL) (Grade B) [SASLT 2021, Page 7]\\nReferences:\\n- Pages 6, 7: Treatment criteria, monitoring protocols"
-}}
----
-EXAMPLE OUTPUT (ELIGIBLE - IMMUNOSUPPRESSION) - Use SINGLE \\n only:
-{{
-  "eligible": true,
-  "recommendations": "Eligibility and Rationale:\\n- Eligible: HBsAg-positive patient requires prophylaxis for immunosuppressive therapy (Grade A) [SASLT 2021, Page 9]\\nTreatment Recommendations:\\n- Start antiviral prophylaxis (e.g., ETV, TDF, TAF) [SASLT 2021, Page 8, 9]\\nMonitoring and Follow-up:\\n- Continue prophylaxis for ≥6 months after immunosuppression (12 for anti-CD20) [SASLT 2021, Page 9]\\nReferences:\\n- Pages 8, 9: Prophylaxis criteria, drug options, monitoring"
-}}
----
-EXAMPLE OUTPUT (ELIGIBLE - HIV COINFECTION) - Use SINGLE \\n only:
-{{
-  "eligible": true,
-  "recommendations": "Eligibility and Rationale:\\n- Eligible: Patient has HBV-HIV coinfection and should start ART (Grade A) [SASLT 2021, Page 9]\\nTreatment Recommendations:\\n- ART regimen must include TDF- or TAF-based therapy (Grade A) [SASLT 2021, Page 9]\\nMonitoring and Follow-up:\\n- Monitor patient closely after ART initiation for immune reconstitution [SASLT 2021, Page 9]\\nReferences:\\n- Page 9: HBV-HIV coinfection management, ART regimens"
-}}
----
-CRITICAL REQUIREMENTS:
-1. Base assessment ONLY on SASLT 2021 guidelines provided.
-2. Follow the CRITICAL ELIGIBILITY HIERARCHY to avoid contradictions.
-3. Keep output SHORT: 8-12 bullets total, 1 sentence per bullet (max 25 words).
-4. Use ONLY the 5 sections listed above.
-5. Cite exact page at end of each bullet: [SASLT 2021, Page X].
-6. Return ONLY valid JSON, no markdown, no extra text.
 """
         
         
@@ -374,6 +442,9 @@ CRITICAL REQUIREMENTS:
         
         # Extract JSON from response
         response_text = response.content if hasattr(response, 'content') else str(response)
+        # Normalize and trim response text to avoid parsing issues due to leading/trailing whitespace
+        if isinstance(response_text, str):
+            response_text = response_text.strip()
         
         # Log LLM response
         logger.info(f"\n{'='*80}")
@@ -399,6 +470,14 @@ CRITICAL REQUIREMENTS:
             json_end = response_text.rfind('}') + 1
             if json_start >= 0 and json_end > json_start:
                 json_str = response_text[json_start:json_end]
+                # Remove invisible Unicode separators that break json.loads
+                # Includes: ZERO WIDTH SPACE (\u200B), ZERO WIDTH NON-JOINER (\u200C),
+                # ZERO WIDTH JOINER (\u200D), BYTE ORDER MARK (\uFEFF), and NO-BREAK SPACE (\u00A0)
+                invisible_chars = ["\u200b", "\u200c", "\u200d", "\ufeff", "\xa0"]
+                for ch in invisible_chars:
+                    json_str = json_str.replace(ch, "")
+                # Also remove their literal forms if present
+                json_str = json_str.replace("\u200b", "").replace("\u200c", "").replace("\u200d", "").replace("\ufeff", "").replace("\u00a0", "")
                 
                 # Clean the JSON string to escape control characters within string values
                 cleaned_json_str = clean_json_string(json_str)
@@ -411,9 +490,14 @@ CRITICAL REQUIREMENTS:
                 raise ValueError("No JSON found in response")
             
             # Validate and return result
+            # Ensure the LLM didn't override the deterministic eligibility flag
+            if result.get("eligible") != is_eligible:
+                 logger.warning(f"LLM contradicted deterministic eligibility. Expected: {is_eligible}, Got: {result.get('eligible')}. Overriding.")
+                 result["eligible"] = is_eligible # Force correct eligibility status
+
             normalized_recs = normalize_recommendations(result.get("recommendations", ""))
             assessment_result = {
-                "eligible": result.get("eligible", False),
+                "eligible": result.get("eligible", is_eligible),
                 "recommendations": normalized_recs
             }
             

core/text_parser.py

@@ -39,7 +39,7 @@ Required fields:
 - hbv_dna_level: float (IU/mL)
 - hbeag_status: "Positive" or "Negative"
 - alt_level: float (U/L)
-- fibrosis_stage: "F0-F1", "F2-F3", or "F4"
+- fibrosis_stage: "F0", "F1", "F2", "F3", or "F4"
 - necroinflammatory_activity: "A0", "A1", "A2", or "A3"
 - extrahepatic_manifestations: true or false
 - immunosuppression_status: "None", "Chemotherapy", or "Other"
@@ -56,7 +56,7 @@ IMPORTANT EXTRACTION RULES:
 6. For HBV DNA: Look for numbers followed by "IU/mL" or "IU/ml"
 7. For HBeAg: Look for "HBeAg positive" or "HBeAg negative"
 8. For ALT: Look for "ALT" followed by number and "U/L"
-9. For fibrosis: Look for "F0", "F1", "F2", "F3", "F4" or descriptions like "significant fibrosis", "cirrhosis"
+9. For fibrosis: Look for "F0", "F1", "F2", "F3", or "F4"
 10. For necroinflammatory: Look for "A0", "A1", "A2", "A3"
 11. For extrahepatic manifestations: Look for mentions of extrahepatic conditions
 12. For immunosuppression: Look for "immunosuppression", "chemotherapy", etc.
@@ -83,7 +83,7 @@ Example format:
   "hbv_dna_level": 5000.0,
   "hbeag_status": "Positive",
   "alt_level": 80.0,
-  "fibrosis_stage": "F2-F3",
+  "fibrosis_stage": "F2",
   "necroinflammatory_activity": "A2",
   "extrahepatic_manifestations": false,
   "immunosuppression_status": "None",