Update src/streamlit_app.py

src/streamlit_app.py

@@ -451,13 +451,14 @@ if submitted:
             if c not in patient_cov.columns:
                 patient_cov[c] = categorical_defaults.get(c, 0)
     
+        try:
+        # Create one-hot design row matching training design columns
         patient_design = make_patient_design_row(patient_cov, design_cols, cat_cols)
-        missing = None
         if isinstance(patient_design, tuple):
-            patient_design, missing = patient_design
-        
-        if missing:
-            st.warning(f"Design row missing columns filled with 0: {missing[:10]}" + (" ..." if len(missing) > 10 else ""))
+            patient_design = patient_design[0]  # keep only the DataFrame row
+    
+        # Predict survival curve + landmarks
+        surv_fn, landmarks = predict_patient_survival(cph, patient_design, years=[3, 5, 10])
     
         st.subheader("Individual adjusted survival curve")
         fig, ax = plt.subplots(figsize=(8, 5))
@@ -467,12 +468,15 @@ if submitted:
         ax.set_ylabel("Survival probability")
         st.pyplot(fig, bbox_inches="tight")
         plt.clf()
-
+    
         st.write("Landmark survival probabilities:")
         for y in [3, 5, 10]:
             s = landmarks.get(y, None)
             if s is not None:
                 st.write(f"{y}-year survival: {100*s:.1f}%")
+    
+    except Exception as e:
+        st.error(f"Survival prediction failed: {e}")
 
     st.caption(
         "Choose the appropriate model from the left panel based on the outcome you want to predict. "