import streamlit as st

def input_form() -> dict:
    """Collect numeric-encoded patient features via sidebar widgets."""
    st.sidebar.header("Patient Information")

    return {
        "gender": st.sidebar.selectbox("Gender", [(0.0, "Male"), (1.0, "Female")])[0],
        "age": st.sidebar.slider("Age", 0.0, 100.0, 50.0),
        "hypertension": st.sidebar.selectbox("Hypertension", [(0, "No"), (1, "Yes")])[0],
        "heart_disease": st.sidebar.selectbox("Heart Disease", [(0, "No"), (1, "Yes")])[0],
        "ever_married": st.sidebar.selectbox("Ever Married", [(0.0, "No"), (1.0, "Yes")])[0],
        "work_type": st.sidebar.selectbox(
            "Work Type",
            [(0.0, "Private"), (1.0, "Self-employed"), (2.0, "Govt_job"), (3.0, "children"), (4.0, "Never_worked")]
        )[0],
        "Residence_type": st.sidebar.selectbox(
            "Residence Type", [(0.0, "Urban"), (1.0, "Rural")]
        )[0],
        "avg_glucose_level": st.sidebar.number_input("Avg Glucose Level", 40.0, 300.0, 100.0),
        "bmi": st.sidebar.number_input("BMI", 10.0, 60.0, 25.0),
        "smoking_status": st.sidebar.selectbox(
            "Smoking Status",
            [(0.0, "formerly smoked"), (1.0, "never smoked"), (2.0, "smokes"), (3.0, "Unknown")]
        )[0]
    }

def display_result(label: str, proba: float):
    """Render prediction and confidence."""
    st.header("Prediction Result")
    st.markdown(f"**Stroke Type:** {label}")
    st.markdown(f"**Confidence:** {proba:.1%}")
    if proba < 0.5:
        st.info("Model confidence is low — consider additional evaluation.")