def predict_and_explain(income, credit_score, debt_ratio, employment):
    try:
        # Prepare input
        input_data = pd.DataFrame([[income, credit_score, debt_ratio, employment]], 
                                  columns=feature_names)
        
        # Get Prediction
        prob = model.predict_proba(input_data)[0][1]
        status = "APPROVED" if prob > 0.5 else "REJECTED"
        color = "#28a745" if status == "APPROVED" else "#dc3545"
        
        # Generate SHAP Explanation (Stable version)
        shap_values = explainer(input_data)
        
        # Create a clean Matplotlib plot
        plt.clf() # Clear previous plots
        fig = plt.figure(figsize=(8, 4))
        
        # We plot the explanation for the 'Positive' class (Approval)
        # For tree models, we take the index [0] for the first row, and [1] for the class
        shap.plots.bar(shap_values[0][..., 1], show=False)
        plt.title("Decision Transparency Score")
        plt.tight_layout()
        
        result_html = f"<div style='text-align: center; padding: 20px; border-radius: 10px; background-color: {color}; color: white;'>" \
                      f"<h1 style='margin:0;'>LOAN STATUS: {status}</h1>" \
                      f"<h3>Confidence Score: {prob*100:.1f}%</h3></div>"
                      
        return result_html, fig
    
    except Exception as e:
        # This will show you exactly what is wrong in the UI if it fails again
        return f"<div style='color:red;'>Error: {str(e)}</div>", None