Update app.py

app.py

@@ -171,6 +171,28 @@ def build_survival_targets(df: pd.DataFrame):
 
 
 import re
+def align_columns_to_schema(df: pd.DataFrame, required_cols: list[str]) -> pd.DataFrame:
+    """
+    Rename df columns to match required_cols using normalized matching.
+    This fixes newline/space/underscore/case differences.
+    """
+    # normalize inference columns
+    df_cols = list(df.columns)
+    norm_to_actual = {norm_col(c): c for c in df_cols}
+
+    rename_map = {}
+    for req in required_cols:
+        k = norm_col(req)
+        if k in norm_to_actual:
+            src = norm_to_actual[k]
+            if src != req:
+                rename_map[src] = req
+
+    if rename_map:
+        df = df.rename(columns=rename_map)
+
+    return df
+
 
 def norm_col(s: str) -> str:
     """
@@ -2326,7 +2348,12 @@ with tab_predict:
             " ".join(str(c).replace("\u00A0"," ").split())
             for c in st.session_state.df_inf.columns
         ]
-    
+
+        # ✅ ALIGN TO TRAINED MODEL SCHEMA (THIS FIXES YOUR ERROR)
+        st.session_state.df_inf = align_columns_to_schema(
+            st.session_state.df_inf,
+            meta["schema"]["features"]
+            
         st.session_state.df_inf = add_ethnicity_region(
             st.session_state.df_inf,
             "Ethnicity",