Update app.py

app.py

@@ -1,32 +1,42 @@
 import gradio as gr
 import joblib
 import pandas as pd
-
-
+import numpy as np
 
 # --- 1. Define all expected features for each model ---
+# These lists are based on the output you provided, but with the one-hot encoded target features removed.
 SAFETY_FEATURES = [
-    'Impact Force Level', 
     'Post-Fall Inactivity Duration (Seconds)', 
     'hour_of_day', 
-    'day_of_week', 
-    'Location_Living Room', 
-    'Location_Kitchen', 
-    'Location_Bedroom'
+    'day_of_week',
+    'rolling_Post-Fall Inactivity Duration (Seconds)_mean',
+    'rolling_hour_of_day_mean',
+    'rolling_day_of_week_mean',
+    'safety_Bathroom',
+    'safety_Bedroom',
+    'safety_Kitchen',
+    'safety_Living Room'
 ]
 
 HEALTH_FEATURES = [
     'Heart Rate', 
     'Glucose Levels', 
-    'Oxygen Saturation (SpO2%)', 
-    'Systolic BP', 
+    'Oxygen Saturation (SpO₂%)', 
+    'Systolic BP',
     'Diastolic BP', 
     'hour_of_day', 
-    'day_of_week'
+    'day_of_week', 
+    'rolling_Heart Rate_mean',
+    'rolling_Glucose Levels_mean',
+    'rolling_Oxygen Saturation (SpO₂%)_mean',
+    'rolling_Systolic BP_mean',
+    'rolling_Diastolic BP_mean',
+    'rolling_hour_of_day_mean',
+    'rolling_day_of_week_mean'
 ]
 
 # All possible locations for one-hot encoding
-all_locations = ['Living Room', 'Kitchen', 'Bedroom']
+all_locations = ['Living Room', 'Kitchen', 'Bedroom', 'Bathroom']
 
 # --- 2. Load the Models and Label Encoders with Error Handling ---
 try:
@@ -35,28 +45,27 @@ try:
     safety_label_encoder = joblib.load('safety_label_encoder.joblib')
     health_label_encoder = joblib.load('health_label_encoder.joblib')
     print("Models and Label Encoders loaded successfully.")
-    # Insert the print statements here to see the expected feature names
-    print("Safety Model expects these features:", rf_safety_model.feature_names_in_)
-    print("Health Model expects these features:", rf_health_model.feature_names_in_)
-    
 except FileNotFoundError as e:
     print(f"Error: One of the required files was not found. Please ensure all files are in the same directory. {e}")
     exit()
 
 # --- 3. Prediction Function for the Safety Model ---
-def predict_safety(impact_force_level, post_fall_inactivity_duration, location, hour_of_day, day_of_week):
+def predict_safety(post_fall_inactivity_duration, hour_of_day, day_of_week, location):
     """
     Prepares and predicts a single row of safety data.
     """
     data = {
-        'Impact Force Level': impact_force_level,
         'Post-Fall Inactivity Duration (Seconds)': post_fall_inactivity_duration,
         'hour_of_day': hour_of_day,
-        'day_of_week': day_of_week
+        'day_of_week': day_of_week,
+        'rolling_Post-Fall Inactivity Duration (Seconds)_mean': 0, # Placeholder for single prediction
+        'rolling_hour_of_day_mean': 0, # Placeholder
+        'rolling_day_of_week_mean': 0, # Placeholder
     }
     
+    # Dynamically create one-hot encoded columns for all possible locations
     for loc in all_locations:
-        data[f'Location_{loc}'] = 1 if location == loc else 0
+        data[f'safety_{loc}'] = 1 if location == loc else 0
     
     input_df = pd.DataFrame([data])
     input_df = input_df[SAFETY_FEATURES]
@@ -64,10 +73,8 @@ def predict_safety(impact_force_level, post_fall_inactivity_duration, location,
     prediction_encoded = rf_safety_model.predict(input_df)
     prediction_label = safety_label_encoder.inverse_transform(prediction_encoded)
     
-    # This line returns the formatted string with "Yes" or "No"
     return f"Fall Detected: {prediction_label[0]}"
 
-
 # --- 4. Prediction Function for the Health Model ---
 def predict_health(heart_rate, glucose_levels, oxygen_saturation, systolic_bp, diastolic_bp, hour_of_day, day_of_week):
     """
@@ -76,11 +83,18 @@ def predict_health(heart_rate, glucose_levels, oxygen_saturation, systolic_bp, d
     data = {
         'Heart Rate': heart_rate,
         'Glucose Levels': glucose_levels,
-        'Oxygen Saturation (SpO2%)': oxygen_saturation,
+        'Oxygen Saturation (SpO₂%)': oxygen_saturation,
         'Systolic BP': systolic_bp,
         'Diastolic BP': diastolic_bp,
         'hour_of_day': hour_of_day,
-        'day_of_week': day_of_week
+        'day_of_week': day_of_week,
+        'rolling_Heart Rate_mean': 0, # Placeholder
+        'rolling_Glucose Levels_mean': 0, # Placeholder
+        'rolling_Oxygen Saturation (SpO₂%)_mean': 0, # Placeholder
+        'rolling_Systolic BP_mean': 0, # Placeholder
+        'rolling_Diastolic BP_mean': 0, # Placeholder
+        'rolling_hour_of_day_mean': 0, # Placeholder
+        'rolling_day_of_week_mean': 0 # Placeholder
     }
     
     input_df = pd.DataFrame([data])
@@ -101,7 +115,6 @@ with gr.Blocks(theme=gr.themes.Soft()) as demo:
     with gr.Tab("Fall Detection (Safety Model)"):
         gr.Markdown("### Predict if a fall has occurred based on sensor data.")
         with gr.Row():
-            impact_force = gr.Slider(0.0, 15.0, label="Impact Force Level (g)", info="Impact force detected by the accelerometer.", step=0.1)
             duration = gr.Slider(0.0, 30.0, label="Post-Fall Inactivity Duration (Seconds)", info="Duration of no movement after impact.", step=0.1)
         
         location_dropdown = gr.Dropdown(all_locations, label="Location", info="Location where the event occurred.")
@@ -114,7 +127,7 @@ with gr.Blocks(theme=gr.themes.Soft()) as demo:
         safety_button = gr.Button("Run Safety Prediction")
         safety_button.click(
             fn=predict_safety,
-            inputs=[impact_force, duration, location_dropdown, hour_slider, day_dropdown],
+            inputs=[duration, hour_slider, day_dropdown, location_dropdown],
             outputs=safety_output
         )
 
@@ -125,22 +138,4 @@ with gr.Blocks(theme=gr.themes.Soft()) as demo:
             glucose_slider = gr.Slider(60, 200, label="Glucose Levels (mg/dL)")
         
         with gr.Row():
-            oxygen_saturation_slider = gr.Slider(85, 100, label="Oxygen Saturation (SpO2%)")
-            systolic_bp_slider = gr.Slider(80, 180, label="Systolic Blood Pressure (mmHg)")
-        
-        with gr.Row():
-            diastolic_bp_slider = gr.Slider(50, 120, label="Diastolic Blood Pressure (mmHg)")
-            hour_slider_health = gr.Slider(0, 23, label="Hour of Day", step=1)
-            day_dropdown_health = gr.Dropdown(choices=[0, 1, 2, 3, 4, 5, 6], label="Day of Week (0=Monday, 6=Sunday)")
-        
-        health_output = gr.Textbox(label="Prediction Result")
-        health_button = gr.Button("Run Health Prediction")
-        health_button.click(
-            fn=predict_health,
-            inputs=[heart_rate_slider, glucose_slider, oxygen_saturation_slider, systolic_bp_slider, diastolic_bp_slider, hour_slider_health, day_dropdown_health],
-            outputs=health_output
-        )
-
-# --- 6. Launch the Gradio App ---
-if __name__ == "__main__":
-    demo.launch()
+            oxygen_saturation_slider = gr.Slider(85, 100, label="Oxygen Saturation (SpO₂%)")