Spaces:
Sleeping
Sleeping
File size: 12,191 Bytes
acaf471 | 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 | from typing import Dict, List, Any
import os
from openai import OpenAI
try:
from dotenv import load_dotenv
load_dotenv()
except ImportError:
pass # dotenv not available, use system env vars
client = OpenAI(api_key=os.getenv("OPENAI_API_KEY"))
def generate_comprehensive_report(
original_text: str,
structured_data: Dict[str, Any],
ontology_mappings: List[Dict],
clinical_recommendations: List[Dict],
detected_language: str,
semantic_analysis: Dict = None
) -> str:
"""
Generate comprehensive clinical report using Medical Anthropologist framework.
Uses detailed 4-layer analysis (linguistic, cultural, clinical, psychosocial).
Outputs as Markdown text for frontend display.
Args:
original_text: Patient's original pain description
structured_data: Structured PainOntology data
ontology_mappings: Ontology mapping trace
clinical_recommendations: List of triggered recommendations
detected_language: Detected language name
semantic_analysis: Optional semantic distance analysis for unmapped terms
Returns:
Markdown-formatted clinical report for frontend rendering
"""
# Prepare ontology mappings summary (MAPPED TERMS ONLY - exact dictionary matches)
mapped_terms_summary = []
for mapping in ontology_mappings[:10]: # Limit to first 10
# Skip suggestions - only show exact/direct mappings
if mapping.get('is_suggestion') or mapping.get('confidence') == 'suggestion_only':
continue
original = mapping.get('original_term', '')
english = mapping.get('mapped_english', '')
pain_type = mapping.get('pain_type', '')
if original and english:
mapped_terms_summary.append(f" - '{original}' โ {english} ({pain_type})")
mappings_text = '\n'.join(mapped_terms_summary) if mapped_terms_summary else " (No direct dictionary mappings found)"
# Prepare recommendations summary
rec_summary = []
for rec in clinical_recommendations:
rule = rec.get('triggered_by_rule', 'Clinical Recommendation')
text = rec.get('recommendation', '')
rec_summary.append(f" - {rule}: {text}")
recs_text = '\n'.join(rec_summary) if rec_summary else " (Standard pain assessment recommended)"
# Prepare UNMAPPED terms semantic analysis summary
unmapped_text = ""
if semantic_analysis and semantic_analysis.get('unmapped_analysis'):
semantic_items = []
for item in semantic_analysis['unmapped_analysis']:
original = item['original_term']
matches = item['closest_matches']
confidence = item['confidence']
if matches:
# Show ALL top matches (usually top 3)
match_list = []
for i, match in enumerate(matches[:3], 1):
# V2: Use new field names (native_term + english)
native = match.get('native_term', match.get('chinese_term', match.get('term', 'Unknown')))
english = match.get('english', '')
match_list.append(f" {i}. {native} ({english}) - similarity: {match['score']:.3f}")
semantic_items.append(
f" - Original: '{original}'\n"
f" Confidence: {confidence}\n"
f" Top matches:\n" + '\n'.join(match_list)
)
if semantic_items:
unmapped_text = "\n\n===== UNMAPPED TERMS - SEMANTIC DISTANCE ANALYSIS (AI-Assisted Interpretation) =====\n"
unmapped_text += "These terms were NOT found in the standard medical dictionary. AI semantic analysis suggests possible matches:\n\n"
unmapped_text += '\n'.join(semantic_items)
unmapped_text += "\n\n โ ๏ธ Important: These are AI-generated suggestions based on semantic similarity, NOT exact dictionary matches.\n Scores closer to 1.0 indicate stronger semantic relationship. Always verify with clinical context."
prompt = f"""You are an expert Medical Anthropologist specializing in cross-cultural pain expression.
Your goal is to translate cultural pain metaphors into structured medical ontologies.
โ ๏ธ DO NOT act as a doctor making a final diagnosis.
โ ๏ธ DO NOT infer beyond the given information.
===== PATIENT INPUT =====
Language: {detected_language}
Original Words: "{original_text}"
===== STRUCTURED CLINICAL DATA (from neuro-symbolic pipeline) =====
Pain Type: {structured_data.get('pain_type', 'Not specified')}
Location: {structured_data.get('location', 'Not specified')}
Temporal Pattern: {structured_data.get('temporal_pattern', 'Not specified')}
Intensity: {structured_data.get('intensity', 'Not stated')}
Emotional Impact: {structured_data.get('emotion', 'None noted')}
Functional Impact: {structured_data.get('functional_impact', 'None noted')}
===== MAPPED TERMS (Direct Matches from Medical Dictionary) =====
{mappings_text}{unmapped_text}
===== CLINICAL RECOMMENDATIONS (from rule engine) =====
{recs_text}
===== YOUR TASK =====
Generate a comprehensive clinical report using this MANDATORY four-layer analytical framework:
**Layer 1: Linguistic Layer (Patient's Voice)**
- Provide literal translation preserving the patient's EXACT wording
- Keep cultural expressions intact (e.g., "ๆญป็ผๆญป็ผ็", "๋ถ๊ฐ์ด ์ํ์")
- Do NOT simplify or standardize the patient's words
**Layer 2: Cultural-Semantic Layer**
- Identify any culturally specific metaphors or expressions
- Explain their clinical meaning
- If no cultural metaphors exist, clearly state that
**Layer 3: Clinical Abstraction Layer (McGill Pain Questionnaire)**
- Sensory qualities (e.g., sharp, burning, aching)
- Affective qualities (e.g., tiring, distressing)
- Temporal pattern and intensity
- Body location
**Layer 4: Psychosocial Layer**
- Emotional distress indicators
- Under-reporting risk (stoicism patterns)
- Communication considerations
**Layer 5: Semantic Distance Analysis (CRITICAL - if applicable)**
- IF semantic analysis data is provided in the input:
- Display EACH unmapped term's semantic similarity scores
- Show the top 3 closest medical terms with similarity scores
- Include confidence levels (high/medium/low)
- Explain in plain language what the similarity scores suggest
- IF no semantic analysis data: skip this layer entirely
===== OUTPUT FORMAT =====
Generate in clear Markdown format with these sections:
**๐ Patient's Description (Literal Translation)**
[First quote patient's exact words in original language, then provide word-for-word English translation preserving sentence structure and cultural expressions]
Example format:
> Original: "ๆญป็ผๆญป็ผ็๏ผ็็ๅไธไบไบ"
> English: "Deadly painful, deadly painful, really can't bear it anymore"
**๐ Cultural Expression Analysis**
[Analyze any cultural metaphors. If none: "No specific cultural metaphors identified."]
**๐ฅ McGill Pain Assessment**
- **Sensory Qualities:** [list descriptors]
- **Affective Qualities:** [list descriptors]
- **Temporal Pattern:** [pattern]
- **Location:** [body location]
- **Intensity:** [severity estimate]
**๐ง Psychosocial Considerations**
- **Emotional Distress:** [Yes/No with brief evidence]
- **Under-reporting Risk:** [Low/Medium/High with reasoning]
- **Communication Notes:** [any relevant observations]
**๐ฌ Semantic Distance Analysis (AI-Based Interpretation)**
[ONLY include this section IF semantic analysis data exists in the input]
For each unmapped creative expression/metaphor:
- **Original Term:** [patient's exact words]
- **Top 3 Similar Medical Terms:**
1. [term] (similarity: X.XX, confidence: high/medium/low)
2. [term] (similarity: X.XX)
3. [term] (similarity: X.XX)
- **Clinical Interpretation:** [1-sentence plain-language explanation of what these similarities suggest about the pain quality]
[If NO semantic analysis data provided, completely omit this section]
**โ๏ธ Clinical Action Plan**
[Synthesize the clinical recommendations above into 2-3 actionable sentences]
===== CRITICAL RULES =====
1. Preserve patient's exact words and emotional tone
2. Base all assessments on actual evidence - don't speculate
3. If no cultural metaphors, state clearly
4. Integrate the provided clinical recommendations
5. **MANDATORY: If semantic analysis data is provided above, you MUST include the "๐ฌ Semantic Distance Analysis" section with all similarity scores displayed clearly**
6. Keep report professional but comprehensive
Generate the report now:"""
try:
response = client.chat.completions.create(
model='gpt-5.2',
messages=[
{"role": "system", "content": "You are a Medical Anthropologist generating clinical reports. Output clear Markdown text. ALWAYS include the Clinical Action Plan section at the end."},
{"role": "user", "content": prompt}
],
temperature=0.2,
max_completion_tokens=2500 # Increased for longer reports
)
return response.choices[0].message.content
except Exception as e:
print(f"[Warning] Report generation error: {e}")
# Fallback to template matching new format
return f"""**๐ Patient's Description**
"{original_text[:300]}..."
**๐ Cultural Expression Analysis**
Unable to analyze cultural expressions at this time.
**๐ฅ McGill Pain Assessment**
- **Sensory Qualities:** Based on structured data
- **Pain Type:** {structured_data.get('pain_type', 'Not specified')}
- **Location:** {structured_data.get('location', 'Not specified')}
- **Temporal Pattern:** {structured_data.get('temporal_pattern', 'Not specified')}
- **Intensity:** {structured_data.get('intensity', 'Not stated')}
**๐ง Psychosocial Considerations**
- **Emotional Distress:** {'Yes' if structured_data.get('emotion') else 'Unknown'}
- **Functional Impact:** {structured_data.get('functional_impact', 'Not noted')}
**โ๏ธ Clinical Action Plan**
{chr(10).join([f"- {rec.get('recommendation', '')}" for rec in clinical_recommendations[:3]]) if clinical_recommendations else 'Standard pain assessment and management recommended.'}
(Note: Full anthropological analysis unavailable. Using template fallback.)
"""
def translate_to_english_simple(text: str) -> str:
"""
Simple translation utility to convert short phrases to English.
Used for translating intensity levels, functional impacts, etc.
If text is already in English, returns it unchanged.
Args:
text: Short text to translate (e.g., "ๅพ็", "difficulty walking")
Returns:
English translation or original text if already English
"""
if not text or text.strip() == "":
return text
# Quick check: if text is already mostly English (ASCII), return as-is
try:
text.encode('ascii')
return text # Already English
except UnicodeEncodeError:
pass # Contains non-ASCII, needs translation
system_prompt = """You are a medical translator. Translate the given text into concise medical English.
Rules:
- Keep it brief and clinical
- Preserve medical meaning
- If already English, return unchanged
- Output ONLY the translation, no explanations"""
try:
response = client.chat.completions.create(
model="gpt-5.2",
messages=[
{"role": "system", "content": system_prompt},
{"role": "user", "content": text}
],
temperature=0.1,
max_tokens=50
)
translation = response.choices[0].message.content.strip()
# Remove quotes if present
if translation.startswith('"') and translation.endswith('"'):
translation = translation[1:-1]
if translation.startswith("'") and translation.endswith("'"):
translation = translation[1:-1]
return translation
except Exception as e:
# Fallback: return original
return text |