Upload engine.py

engine.py

@@ -1,6 +1,7 @@
 import os
 import sqlite3
 from openai import OpenAI
+from difflib import get_close_matches
 
 
 # =========================
@@ -11,6 +12,34 @@ client = OpenAI(api_key=os.getenv("OPENAI_API_KEY"))
 conn = sqlite3.connect("hospital.db", check_same_thread=False)
 
 
+# =========================
+# Known Terms for Spell Correction
+# =========================
+
+KNOWN_TERMS = [
+    "patient", "patients", "condition", "conditions", "diagnosis", "encounter", "encounters",
+    "visit", "visits", "observation", "observations", "lab", "labs", "test", "tests",
+    "medication", "medications", "drug", "drugs", "prescription", "prescriptions",
+    "diabetes", "hypertension", "asthma", "cancer"
+]
+
+
+def correct_spelling(question: str) -> str:
+    words = question.split()
+    corrected_words = []
+
+    for word in words:
+        clean_word = word.lower().strip(",.?")
+        matches = get_close_matches(clean_word, KNOWN_TERMS, n=1, cutoff=0.8)
+
+        if matches:
+            corrected_words.append(matches[0])
+        else:
+            corrected_words.append(word)
+
+    return " ".join(corrected_words)
+
+
 # =========================
 # Metadata Loader
 # =========================
@@ -152,13 +181,10 @@ def run_query(sql: str):
 
 def is_question_answerable(question):
     schema = load_ai_schema()
-    schema_text = " ".join(schema.keys()).lower()
-
     keywords = ["patient", "encounter", "condition", "observation", "medication", "visit", "diagnosis", "lab", "vital"]
 
     q = question.lower()
 
-    # If none of the core domain keywords are present, likely out of scope
     if not any(k in q for k in keywords):
         return False
 
@@ -197,6 +223,9 @@ def interpret_empty_result(question: str):
 # =========================
 
 def process_question(question: str):
+    # 0. Spell correction
+    question = correct_spelling(question)
+
     # 1. Guardrail
     if not is_question_answerable(question):
         return {