app.py

import os
import gradio as gr
import pandas as pd
import numpy as np
from sklearn.ensemble import RandomForestClassifier
from sklearn.preprocessing import StandardScaler

# Create a simple model directly in the app
def create_model():
    # Sample hospital data
    data = {
        'age': [65, 72, 58, 81, 45],
        'time_in_hospital': [5, 8, 3, 12, 4],
        'num_lab_procedures': [45, 32, 28, 51, 38],
        'num_medications': [15, 22, 8, 18, 12],
        'readmitted': [1, 1, 0, 1, 0]
    }
    df = pd.DataFrame(data)
    
    X = df.drop('readmitted', axis=1)
    y = df['readmitted']
    
    # Create and train model
    model = RandomForestClassifier(n_estimators=10, random_state=42)
    model.fit(X, y)
    
    # Create and fit preprocessor
    preprocessor = StandardScaler()
    preprocessor.fit(X)
    
    return model, preprocessor

# Create model directly in memory
model, preprocessor = create_model()
print("Successfully created model and preprocessor in memory!")

# Define features
FEATURES = ['age', 'time_in_hospital', 'num_lab_procedures', 'num_medications']

def predict_readmission(age, time_in_hospital, num_lab_procedures, num_medications):
    input_data = np.array([[age, time_in_hospital, num_lab_procedures, num_medications]])
    processed = preprocessor.transform(input_data)
    prob = model.predict_proba(processed)[0][1]
    return {"Probability": float(prob), "Risk": "High" if prob > 0.5 else "Low"}

# Create interface
with gr.Blocks() as demo:
    gr.Markdown("# 🏥 Hospital Readmission Risk Predictor")
    
    with gr.Row():
        gr.Number(label="Age", value=65)
        gr.Number(label="Days in Hospital", value=5)
    with gr.Row():
        gr.Number(label="Lab Procedures", value=45)
        gr.Number(label="Medications", value=15)
    
    inputs = [
        gr.Number(label="Age", value=65),
        gr.Number(label="Days in Hospital", value=5),
        gr.Number(label="Lab Procedures", value=45),
        gr.Number(label="Medications", value=15)
    ]
    
    submit = gr.Button("Predict Readmission Risk")
    output = gr.JSON()
    
    submit.click(
        fn=predict_readmission,
        inputs=inputs,
        outputs=output
    )

demo.launch()