appold

import gradio as gr
import numpy as np
import pandas as pd
import wfdb
import tensorflow as tf
from scipy import signal
import os

# Load the pre-trained model (adjust path as needed)
model = tf.keras.models.load_model("model.h5")

# Function to preprocess ECG data (simplified from the repo)
def preprocess_ecg(file_path):
    # Read ECG file (assuming WFDB format like .dat)
    record = wfdb.rdrecord(file_path.replace(".dat", ""))
    ecg_signal = record.p_signal[:, 0]  # Use first lead for simplicity
    
    # Resample to 360 Hz (common ECG frequency)
    target_fs = 360
    num_samples = int(len(ecg_signal) * target_fs / record.fs)
    ecg_resampled = signal.resample(ecg_signal, num_samples)
    
    # Normalize
    ecg_normalized = (ecg_resampled - np.mean(ecg_resampled)) / np.std(ecg_resampled)
    
    # Segment into fixed length (e.g., 3600 samples = 10 seconds at 360 Hz)
    if len(ecg_normalized) < 3600:
        ecg_normalized = np.pad(ecg_normalized, (0, 3600 - len(ecg_normalized)), "constant")
    else:
        ecg_normalized = ecg_normalized[:3600]
    
    # Reshape for model input (assuming model expects [1, 3600, 1])
    ecg_input = ecg_normalized.reshape(1, 3600, 1)
    return ecg_input

# Prediction function
def predict_ecg(file):
    # Extract file path from uploaded file
    file_path = file.name
    
    # Preprocess the ECG data
    ecg_data = preprocess_ecg(file_path)
    
    # Make prediction
    prediction = model.predict(ecg_data)
    
    # Assuming binary classification (e.g., Normal vs. Abnormal)
    label = "Abnormal" if prediction[0][0] > 0.5 else "Normal"
    confidence = float(prediction[0][0]) if label == "Abnormal" else float(1 - prediction[0][0])
    
    return f"Prediction: {label}\nConfidence: {confidence:.2%}"

# Gradio interface
interface = gr.Interface(
    fn=predict_ecg,
    inputs=gr.File(label="Upload ECG File (.dat format)"),
    outputs=gr.Textbox(label="ECG Interpretation"),
    title="Automated ECG Interpretation",
    description="Upload an ECG file in WFDB format (.dat) to get an automated interpretation."
)

# Launch the app
interface.launch()