app.py

# app.py (for Hugging Face Spaces)
import gradio as gr
import numpy as np
import random
from datetime import datetime
import pandas as pd

# Simulated glucose log
glucose_log = []

# Mock CGM data (simulates Dexcom-like readings)
def fetch_mock_cgm_reading():
    base = random.uniform(150, 250)  # Type 1 diabetic range (e.g., your 201 mg/dL)
    variability = random.uniform(-30, 30)
    return base + variability

# Prediction and analysis function
def glucothumb_predict(calibrate_value=None):
    global glucose_log
    
    try:
        # Fetch mock CGM reading
        current_glucose = fetch_mock_cgm_reading()
        
        # Apply calibration if provided
        if calibrate_value is not None:
            calibration_offset = calibrate_value - current_glucose
            current_glucose = calibrate_value
        else:
            calibration_offset = 0
        
        # Clamp to realistic range
        current_glucose = max(50, min(400, current_glucose))
        
        # Log the reading
        timestamp = datetime.now().strftime("%Y-%m-%d %H:%M:%S")
        glucose_log.append({"Time": timestamp, "Glucose (mg/dL)": current_glucose})
        
        # Trend analysis (last 5 readings)
        if len(glucose_log) > 1:
            df = pd.DataFrame(glucose_log[-5:])
            avg_glucose = df["Glucose (mg/dL)"].mean()
            trend = "Rising" if df["Glucose (mg/dL)"].iloc[-1] > df["Glucose (mg/dL)"].iloc[-2] else "Falling"
            high_alert = "High Alert!" if current_glucose > 250 else ""
            low_alert = "Low Alert!" if current_glucose < 70 else ""
            summary = f"Average (last 5): {avg_glucose:.2f} mg/dL\nTrend: {trend}\n{high_alert}{low_alert}"
        else:
            summary = "Initial reading logged. Add more for trends."
        
        # Shareable log
        log_text = "\n".join([f"{entry['Time']}: {entry['Glucose (mg/dL)']:.2f} mg/dL" for entry in glucose_log])
        
        return (
            f"Current Blood Sugar Level: {current_glucose:.2f} mg/dL (Simulated CGM Data)",
            summary,
            log_text
        )
    except Exception as e:
        return f"Error: {str(e)}", "", ""

# Gradio interface
interface = gr.Interface(
    fn=glucothumb_predict,
    inputs=[
        gr.Number(label="Calibrate with Known Glucose (Optional, e.g., 201 mg/dL)", value=None)
    ],
    outputs=[
        gr.Textbox(label="Current Glucose"),
        gr.Textbox(label="Trend Analysis & Alerts"),
        gr.Textbox(label="Glucose Log (Share with Doctor)")
    ],
    title="GlucoThumb",
    description="Simulates real-time blood sugar monitoring for Type 1 diabetes using mock CGM data. Calibrate with a known value (e.g., 201 mg/dL) if desired. Click 'Submit' to refresh. Note: Demo only—replace with Dexcom API for real data.",
    live=False
)

# Launch the app
interface.launch()