app.py

import json, joblib, pandas as pd, numpy as np
import gradio as gr
import seaborn as sns
import matplotlib.pyplot as plt

# Load model + metadata + dataset
model = joblib.load("maternal_rf_model.joblib")
with open("maternal_metadata.json","r",encoding="utf-8") as f:
    meta = json.load(f)

try:
    df_clean = pd.read_csv("maternal_cleaned.csv")
except FileNotFoundError:
    df_clean = None

numeric_features = meta["numeric_features"]
categorical_features = meta["categorical_features"]

# ---------- Prediction history ----------
prediction_history = []

# ---------- Prediction function ----------
def predict_risk(age, gravida, gest_weeks, weight, height_cm,
                 bp_sys, bp_dias, fetal_hr,
                 anaemia, jaundice, fetal_position, fetal_movement,
                 urine_albumin, urine_sugar):
    row = {
        "Age": age, "Gravida": gravida, "GestationWeeks": gest_weeks,
        "WeightKg": weight, "HeightCm": height_cm,
        "BP_Systolic": bp_sys, "BP_Diastolic": bp_dias,
        "FetalHR": fetal_hr,
        "Anaemia": anaemia, "Jaundice": jaundice,
        "FetalPosition": fetal_position, "FetalMovement": fetal_movement,
        "UrineAlbumin": urine_albumin, "UrineSugar": urine_sugar
    }
    X = pd.DataFrame([row], columns=numeric_features+categorical_features)
    prob = model.predict_proba(X)[:,1][0]
    pred = int(model.predict(X)[0])
    label = "High Risk" if pred==1 else "Not High Risk"

    # Save to history
    history_row = row.copy()
    history_row["Prediction"] = label
    history_row["Probability"] = round(prob, 4)
    prediction_history.append(history_row)

    return {"Prediction": label, "Probability_high_risk": round(prob,4)}

# ---------- Plot functions ----------
def plot_age_distribution():
    if df_clean is None: return plt.figure()
    fig, ax = plt.subplots(figsize=(6,4))
    sns.histplot(df_clean["Age"], bins=10, kde=True, ax=ax, color="skyblue")
    ax.set_title("Age Distribution")
    return fig

def plot_risk_counts():
    if df_clean is None: return plt.figure()
    fig, ax = plt.subplots(figsize=(6,4))
    sns.countplot(x="HighRisk", data=df_clean, ax=ax, palette="Set2")
    ax.set_title("High Risk vs Non-Risk Counts")
    return fig

def plot_gestation_box():
    if df_clean is None: return plt.figure()
    fig, ax = plt.subplots(figsize=(6,4))
    sns.boxplot(x="HighRisk", y="GestationWeeks", data=df_clean, ax=ax, palette="Set2")
    ax.set_title("Gestation Weeks vs Risk")
    return fig

def plot_feature_importance():
    ohe = model.named_steps["preprocessor"].named_transformers_["cat"].named_steps["onehot"]
    cat_names = ohe.get_feature_names_out(categorical_features)
    feature_names = numeric_features + list(cat_names)
    importances = model.named_steps["clf"].feature_importances_
    feat_imp = pd.DataFrame({"Feature":feature_names,"Importance":importances})
    feat_imp = feat_imp.sort_values("Importance",ascending=False).head(10)
    fig, ax = plt.subplots(figsize=(8,5))
    sns.barplot(x="Importance", y="Feature", data=feat_imp, ax=ax, palette="viridis")
    ax.set_title("Top 10 Feature Importances")
    return fig

def plot_corr_heatmap():
    if df_clean is None: return plt.figure()
    fig, ax = plt.subplots(figsize=(8,6))
    corr = df_clean[numeric_features+["HighRisk"]].corr()
    sns.heatmap(corr, annot=True, cmap="coolwarm", fmt=".2f", ax=ax)
    ax.set_title("Correlation Heatmap")
    return fig

# ---------- History update ----------
def update_history():
    return pd.DataFrame(prediction_history)

# ---------- Gradio UI ----------
with gr.Blocks(title="Maternal Risk Prediction Dashboard") as demo:
    gr.Markdown("## Maternal Risk Prediction Dashboard")

    with gr.Tab("Prediction"):
        gr.Markdown("Enter maternal health parameters to predict risk.")
        with gr.Row():
            age = gr.Number(label="Age")
            gravida = gr.Number(label="Gravida")
            gest = gr.Number(label="Gestation Weeks")
            weight = gr.Number(label="Weight (kg)")
            height = gr.Number(label="Height (cm)")
        with gr.Row():
            bp_sys = gr.Number(label="BP Systolic")
            bp_dias = gr.Number(label="BP Diastolic")
            fetal_hr = gr.Number(label="Fetal Heart Rate")
            anaemia = gr.Dropdown(["None","Minimal","Medium","Higher"], label="Anaemia")
            jaundice = gr.Dropdown(["None","Minimal","Medium"], label="Jaundice")
        with gr.Row():
            fetal_pos = gr.Dropdown(["Normal","Abnormal"], label="Fetal Position")
            fetal_mov = gr.Dropdown(["Yes","No"], label="Fetal Movement")
            urine_alb = gr.Dropdown(["Negative","Positive"], label="Urine Albumin")
            urine_sug = gr.Dropdown(["Negative","Positive"], label="Urine Sugar")
        out = gr.JSON(label="Result")
        btn = gr.Button("Predict Risk")
        btn.click(predict_risk,
                  inputs=[age,gravida,gest,weight,height,
                          bp_sys,bp_dias,fetal_hr,
                          anaemia,jaundice,fetal_pos,fetal_mov,urine_alb,urine_sug],
                  outputs=out)

    with gr.Tab("Data Insights"):
        gr.Plot(plot_age_distribution)
        gr.Plot(plot_risk_counts)
        gr.Plot(plot_gestation_box)

    with gr.Tab("Model Insights"):
        gr.Plot(plot_feature_importance)
        gr.Plot(plot_corr_heatmap)

    with gr.Tab("Prediction History"):
        history_table = gr.DataFrame(label="Prediction History", interactive=False)
        refresh_btn = gr.Button("Refresh History")
        refresh_btn.click(fn=update_history, outputs=history_table)

    with gr.Tab("About"):
        gr.Markdown("""
        ### About this App
        This dashboard predicts maternal high-risk pregnancy using a RandomForest model.
        - **Dataset:** Cleaned maternal health records
        - **Features:** Age, Gravida, Gestation Weeks, Weight, Height, BP, Fetal HR, Anaemia, Jaundice, Fetal Position, Fetal Movement, Urine Albumin, Urine Sugar
        - **Output:** High Risk vs Not High Risk with probability
        """)

demo.launch()