Update app.py

app.py

@@ -2,69 +2,39 @@ import gradio as gr
 import numpy as np
 import matplotlib.pyplot as plt
 
-# Function to simulate blood glucose changes using our exact manual method
-def predict_glucose(current_glucose, meal_type, meal_time_hr, meal_time_min, galvus_dose, galvus_time_hr, galvus_time_min, exercise_duration, fast_carbs_ml, chocolate_time_hr, chocolate_time_min):
-    # Convert time to total minutes
-    meal_time = meal_time_hr * 60 + meal_time_min
-    galvus_time = galvus_time_hr * 60 + galvus_time_min
-    chocolate_time = chocolate_time_hr * 60 + chocolate_time_min
-
-    # Constants based on manual logic
-    galvus_total_effect = 60  # mg/dL drop over 2 hours
-    galvus_rate = galvus_total_effect / 2  # 30 mg/dL per hour
-    fasting_drop_first_3h = 0.48  # mg/dL per hour for first 3 hours
-    fasting_drop_after_3h = 2  # mg/dL per hour after 3 hours
-    carb_effect_per_ml = 1.5  # mg/dL per mL of fast carbs
-    exercise_reduction_rate = 2  # mg/dL per minute of exercise
-    chocolate_spike = 40  # mg/dL rise from chocolate bar
-
-    # Step 1: Calculate Meal Impact
+# Function to simulate blood glucose changes over time
+def predict_glucose(current_glucose, meal_type, meal_time, galvus_dose, galvus_time, exercise_duration, fast_carbs_ml, prediction_time=3):
+    # Constants for glucose reduction effects
+    post_meal_reduction = 63.6  # mg/dL (avg reduction for Vildagliptin in 2 hours)
+    fasting_reduction = 27.7  # mg/dL (avg reduction over 6-12 hours)
+    
+    # Adjust for fast carbs (milk, juice, etc.)
+    carb_effect = fast_carbs_ml * 1.5  # Approximate glucose rise per mL of fast carbs
+
+    # Calculate blood glucose over time considering meal type, Galvus dose, and exercise
     if meal_type == 'High-carb':
-        glucose_after_meal = current_glucose + 50 + (fast_carbs_ml * carb_effect_per_ml)
+        glucose_after_meal = current_glucose + carb_effect
     elif meal_type == 'Protein-heavy':
-        glucose_after_meal = current_glucose + 20
+        glucose_after_meal = current_glucose + 20  # Small glucose increase due to protein
     elif meal_type == 'Low-carb':
-        glucose_after_meal = current_glucose - 10
-    elif meal_type == 'Fast carbs':
-        glucose_after_meal = current_glucose + (fast_carbs_ml * carb_effect_per_ml)
+        glucose_after_meal = current_glucose - 10  # Small reduction due to low-carb meal
     else:
-        glucose_after_meal = current_glucose  # No big change for normal meals
+        glucose_after_meal = current_glucose  # Normal meal with moderate carbs
 
-    # Step 2: Apply Fasting Effect
-    time_since_last_meal = (meal_time_hr * 60 + meal_time_min) / 60  # Convert to hours
-    if time_since_last_meal <= 3:
-        fasting_effect = time_since_last_meal * fasting_drop_first_3h
-    else:
-        fasting_effect = (3 * fasting_drop_first_3h) + ((time_since_last_meal - 3) * fasting_drop_after_3h)
-
-    glucose_after_fasting = glucose_after_meal - fasting_effect
-
-    # Step 3: Apply Galvus Effect
-    time_since_galvus = (meal_time - galvus_time) / 60  # Convert back to hours
-    if time_since_galvus >= 1:
-        glucose_after_galvus = glucose_after_fasting - galvus_rate
-    else:
-        glucose_after_galvus = glucose_after_fasting  # No effect yet
-
-    # Step 4: Apply Chocolate Effect
-    time_since_chocolate = (meal_time - chocolate_time) / 60  # Convert back to hours
-    if time_since_chocolate >= 0:  # Chocolate was eaten after meal
-        glucose_after_chocolate = glucose_after_galvus + chocolate_spike
-    else:
-        glucose_after_chocolate = glucose_after_galvus
-
-    # Step 5: Apply Second Hour Galvus Effect
-    glucose_after_2hr = glucose_after_chocolate - galvus_rate  # Second hour of Galvus effect
+    # Simulate blood glucose levels over 1 hour and 3 hours
+    glucose_1hr = glucose_after_meal - post_meal_reduction + carb_effect * 0.5  # Adjust for carb effect
+    glucose_3hr = glucose_1hr - fasting_reduction
 
-    # Step 6: Apply Exercise Effect
-    glucose_after_exercise = glucose_after_2hr - (exercise_duration * exercise_reduction_rate)
+    # Adjust the effects of Galvus based on its administration time (medication effect starts at galvus_time + 1-2 hours)
+    time_since_galvus = meal_time - galvus_time
+    if time_since_galvus >= 1:  # After 1 hour, the effects of Galvus start kicking in
+        glucose_3hr -= fasting_reduction  # Galvus effect after 3 hours
 
-    # Step 7: Predict 1-Hour and 3-Hour Levels
-    glucose_1hr = glucose_after_chocolate  # Since we assume peak at 1-hour
-    glucose_3hr = glucose_after_exercise  # After all effects take place
+    # Exercise effect on glucose (hypothetical value, may vary based on intensity)
+    glucose_3hr -= exercise_duration * 2  # Exercise reduces glucose by 2 mg/dL per minute
 
-    # Graphing the glucose change
-    time_points = [0, 1, 3]
+    # Plotting the graph of glucose prediction over time
+    time_points = [0, 1, 3]  # Time: 0 hours, 1 hour, 3 hours
     glucose_values = [current_glucose, glucose_1hr, glucose_3hr]
 
     plt.plot(time_points, glucose_values, marker='o', color='b')
@@ -75,47 +45,47 @@ def predict_glucose(current_glucose, meal_type, meal_time_hr, meal_time_min, gal
     plt.grid(True)
     plt.tight_layout()
 
-    # Save the graph
+    # Save the graph as a file to show it in Gradio
     plt.savefig('/tmp/blood_glucose_prediction.png')
     plt.close()
 
+    # Return glucose predictions and the image file path
     return glucose_1hr, glucose_3hr, '/tmp/blood_glucose_prediction.png'
 
+
 # Gradio Interface
 def build_interface():
     with gr.Blocks() as iface:
-        gr.Markdown("# Blood Glucose Prediction Model (Step-by-Step Thinking)")
+        gr.Markdown("# Blood Glucose Prediction Model (With Vildagliptin Effects)")
 
+        # Inputs for current glucose, meal info, medication dose, exercise, fast carbs, and Galvus time
         with gr.Row():
-            current_glucose = gr.Number(label="Current Blood Glucose (mg/dL)", value=150)
-            meal_type = gr.Radio(choices=["Normal", "High-carb", "Protein-heavy", "Low-carb", "Fast carbs"], label="Meal Type", value="Normal")
-            meal_time_hr = gr.Number(label="Last Meal Time (Hours)", value=2)
-            meal_time_min = gr.Number(label="Last Meal Time (Minutes)", value=0)
+            current_glucose = gr.Number(label="Current Blood Glucose (mg/dL)", value=105)
+            meal_type = gr.Radio(choices=["Normal", "High-carb", "Protein-heavy", "Low-carb"], label="Meal Type", value="Low-carb")
+            meal_time = gr.Number(label="Last Meal Time (in hours)", value=6)
             galvus_dose = gr.Number(label="Galvus Dose (mg)", value=50)
-            galvus_time_hr = gr.Number(label="Galvus Time (Hours)", value=0)
-            galvus_time_min = gr.Number(label="Galvus Time (Minutes)", value=0)
-            exercise_duration = gr.Number(label="Exercise Duration (Minutes)", value=0)
+            galvus_time = gr.Number(label="Galvus Time of Administration (hours)", value=2)  # Input for when Galvus was taken
+            exercise_duration = gr.Number(label="Exercise Duration (min)", value=60)
             fast_carbs_ml = gr.Number(label="Fast Carbs (mL)", value=0)
-            chocolate_time_hr = gr.Number(label="Chocolate Bar Time (Hours)", value=0)
-            chocolate_time_min = gr.Number(label="Chocolate Bar Time (Minutes)", value=0)
 
+        # Output predictions and graph
         glucose_1hr_output = gr.Textbox(label="Predicted Glucose Level in 1 Hour (mg/dL)")
         glucose_3hr_output = gr.Textbox(label="Predicted Glucose Level in 3 Hours (mg/dL)")
         glucose_graph = gr.Image(label="Blood Glucose Prediction Graph")
 
+        # Button to trigger prediction
         predict_button = gr.Button("Predict Blood Glucose")
 
-        def on_predict(*inputs):
-            return predict_glucose(*inputs)
-
+        # Set button action
         predict_button.click(
-            on_predict,
-            inputs=[current_glucose, meal_type, meal_time_hr, meal_time_min, galvus_dose, galvus_time_hr, galvus_time_min, exercise_duration, fast_carbs_ml, chocolate_time_hr, chocolate_time_min],
+            predict_glucose,
+            inputs=[current_glucose, meal_type, meal_time, galvus_dose, galvus_time, exercise_duration, fast_carbs_ml],
             outputs=[glucose_1hr_output, glucose_3hr_output, glucose_graph]
         )
 
     return iface
 
-# Launch the interface
+
+# Build and launch the Gradio interface
 iface = build_interface()
 iface.launch()