Update app.py

app.py

@@ -2,164 +2,98 @@ import streamlit as st
 import pandas as pd
 import plotly.graph_objs as go
 
-# Step 1: Load the cleaned sheet based on the selected site
-def load_clean_bmd_data(file_path, site):
-    sheet_name = ''
-    if site == 'Total Hip':
-        sheet_name = 'Clean TH avg rise BMD'
-    elif site == 'Femoral Neck':
-        sheet_name = 'clean FN avg rise BMD'
-    elif site == 'Lumbar Spine (L1-L4)':
-        sheet_name = 'clean LS avg rise BMD'
-
-    df_clean_bmd_data = pd.read_excel(file_path, sheet_name=sheet_name)
-
-    # Select relevant columns: Drug names and BMD percentage increases
-    df_cleaned = df_clean_bmd_data[['Unnamed: 0', '1st', '2nd', '3rd', '4th', '5th', '6th', '8th', '10th']]
-
-    # Rename the first column to 'Drug'
-    df_cleaned.columns = ['Drug', '1st', '2nd', '3rd', '4th', '5th', '6th', '8th', '10th']
-
-    # Remove any rows with missing drug names
-    df_cleaned = df_cleaned.dropna(subset=['Drug'])
-
-    return df_cleaned
-
-# Step 2: Adjust constants based on patient's BMD and T-score
-def adjust_constants(bmd_patient, tscore_patient, c_avg, c_sd):
-    # Adjust C_avg and C_sd based on patient's BMD and T-score
-    error = tscore_patient - (bmd_patient - c_avg) / c_sd
-    c_avg_new = c_avg + error * c_sd
-    c_sd_new = (bmd_patient - c_avg) / tscore_patient
-    return c_avg_new, c_sd_new
-
-# Step 3: Calculate BMD increase after medication (corrected version)
-def calculate_bmd_increase(baseline_bmd, percentage_increase):
-    return baseline_bmd * (1 + percentage_increase)
-
-# Step 4: Create a table showing BMD prediction and T-score conversion for each year for each drug
-def create_bmd_and_tscore_prediction_tables(df_bmd_data, selected_drugs, bmd_patient, tscore_patient, c_avg_new, c_sd_new):
-    years = ['1st', '2nd', '3rd', '4th', '5th', '6th', '8th', '10th']
-    drug_tables = {}
-
-    # Loop through each selected drug and generate the prediction table
-    for drug in selected_drugs:
-        predictions = [('0', bmd_patient, tscore_patient)]  # Add initial value (Year 0)
+# Constants from linear regression
+REGRESSION_CONSTANTS = {
+    'FN': {'mu': 0.916852, 'sigma': 0.120754},
+    'TH': {'mu': 0.955439, 'sigma': 0.125406},
+    'LS': {'mu': 1.131649, 'sigma': 0.139618},
+}
+
+# Load medication data
+@st.cache_data
+def load_medication_data():
+    # Use the combined medication data we cleaned earlier
+    file_path = "/mnt/data/BMD medication prediction.xlsx"
+    return pd.read_excel(file_path)
+
+# Calculate predicted BMD after medication
+def calculate_bmd(bmd, percentage_increase):
+    return bmd * (1 + percentage_increase)
+
+# Convert BMD to T-score
+def calculate_tscore(bmd, mu, sigma):
+    return (bmd - mu) / sigma
+
+# Generate prediction table for all drugs
+def generate_predictions(medication_data, site, bmd, mu, sigma):
+    site_data = medication_data[medication_data['Site'] == site]
+    years = ['1st Year', '2nd Year', '3rd Year', '4th Year', '5th Year', '6th Year', '8th Year', '10th Year']
+    
+    prediction_results = []
+    for _, row in site_data.iterrows():
+        drug = row['Medication']
+        bmd_predictions = [bmd]
+        tscore_predictions = [calculate_tscore(bmd, mu, sigma)]
+        
         for year in years:
-            if not pd.isna(df_bmd_data.loc[df_bmd_data['Drug'] == drug, year].values[0]):
-                percent_increase = df_bmd_data.loc[df_bmd_data['Drug'] == drug, year].values[0]
-                bmd_new = calculate_bmd_increase(bmd_patient, percent_increase)  # Use baseline BMD
-                tscore_new = calculate_tscore_from_bmd(bmd_new, c_avg_new, c_sd_new)  # Calculate predicted T-score
-                predictions.append((year, bmd_new, tscore_new))
+            if not pd.isna(row[year]):
+                bmd_new = calculate_bmd(bmd_predictions[-1], row[year])
+                tscore_new = calculate_tscore(bmd_new, mu, sigma)
+                bmd_predictions.append(bmd_new)
+                tscore_predictions.append(tscore_new)
         
-        # Create DataFrame for each drug
-        drug_table = pd.DataFrame(predictions, columns=['Year', 'Predicted BMD', 'Predicted T-score'])
-        drug_tables[drug] = drug_table
-
-    return drug_tables
-
-# Step 5: Plot BMD and T-score as separate graphs side by side for each drug
-def display_prediction_tables_and_plots(prediction_tables, baseline_bmd, baseline_tscore):
-    # Loop through each drug's prediction table
-    for drug, table in prediction_tables.items():
-        st.write(f"### {drug} Results")
+        prediction_results.append({
+            'Drug': drug,
+            'BMD Predictions': bmd_predictions,
+            'T-score Predictions': tscore_predictions,
+        })
+    return prediction_results
+
+# Display results as table and plots
+def display_results(predictions, site):
+    st.subheader(f"Predictions for {site}")
+    
+    for result in predictions:
+        drug = result['Drug']
+        bmd_predictions = result['BMD Predictions']
+        tscore_predictions = result['T-score Predictions']
+        years = ['0', '1st', '2nd', '3rd', '4th', '5th', '6th', '8th', '10th']
         
-        # Display the prediction table for the current drug
-        st.dataframe(table)
-
-        # Create and display separate plots for each drug
-        years = list(table['Year'])
-        bmd_values = list(table['Predicted BMD'])
-        tscore_values = list(table['Predicted T-score'])
-
-        # Create BMD plot
-        trace_bmd = go.Scatter(x=years, y=bmd_values, mode='lines+markers', name=f'{drug} (BMD)', line=dict(color='blue'))
-        fig_bmd = go.Figure(data=[trace_bmd], layout=go.Layout(title=f'{drug} - Predicted BMD over Time', xaxis=dict(title='Years'), yaxis=dict(title='BMD (g/cm²)')))
+        # Create table
+        data = {
+            'Year': years[:len(bmd_predictions)],
+            'Predicted BMD': bmd_predictions,
+            'Predicted T-score': tscore_predictions,
+        }
+        st.write(f"### {drug}")
+        st.dataframe(pd.DataFrame(data))
         
-        # Create T-score plot
-        trace_tscore = go.Scatter(x=years, y=tscore_values, mode='lines+markers', name=f'{drug} (T-score)', line=dict(color='green'))
-        fig_tscore = go.Figure(data=[trace_tscore], layout=go.Layout(title=f'{drug} - Predicted T-score over Time', xaxis=dict(title='Years'), yaxis=dict(title='T-score')))
-
-        # Display the plots
-        col1, col2 = st.columns(2)
-        with col1:
-            st.plotly_chart(fig_bmd)
-        with col2:
-            st.plotly_chart(fig_tscore)
-
-        # Step 6: Check if goal is achieved and show the predicted BMD and T-score for each year
-        goal_achieved = False
-        for i, row in table.iterrows():
-            if row['Predicted T-score'] >= -2.49:
-                st.success(f"Goal achieved for {drug} at year {row['Year']} with T-score = {row['Predicted T-score']:.2f}")
-                goal_achieved = True
-                break
-        if not goal_achieved:
-            st.warning(f"Goal not achieved for {drug}")
-
-# Step 6: Calculate T-score from adjusted BMD
-def calculate_tscore_from_bmd(bmd_patient, c_avg, c_sd):
-    return (bmd_patient - c_avg) / c_sd
-
-# Main function to load data, run the application, and plot results with T-score labels
-def main_with_separate_tables(file_path, bmd_patient, tscore_patient, C_avg_lunar, C_sd_lunar, selected_drugs, site):
-    # Step 1: Load and clean BMD data from the selected site (Total Hip, Femoral Neck, Lumbar Spine)
-    df_bmd_data = load_clean_bmd_data(file_path, site)
-
-    # Step 2: Adjust constants based on the patient's data
-    c_avg_new, c_sd_new = adjust_constants(bmd_patient, tscore_patient, C_avg_lunar, C_sd_lunar)
-
-    # Step 4: Create separate prediction tables for each selected drug
-    prediction_tables = create_bmd_and_tscore_prediction_tables(df_bmd_data, selected_drugs, bmd_patient, tscore_patient, c_avg_new, c_sd_new)
-
-    # Display baseline BMD and T-score
-    st.write(f"Baseline: BMD = {bmd_patient:.3f}, T-score = {tscore_patient:.2f}")
-
-    # Step 5: Display prediction tables and plots for each drug
-    display_prediction_tables_and_plots(prediction_tables, bmd_patient, tscore_patient)
+        # Create plots
+        fig_bmd = go.Figure(data=[go.Scatter(x=data['Year'], y=data['Predicted BMD'], mode='lines+markers', name='BMD')])
+        fig_bmd.update_layout(title=f"{drug} - Predicted BMD over Time", xaxis_title="Year", yaxis_title="BMD (g/cm²)")
+        st.plotly_chart(fig_bmd)
+        
+        fig_tscore = go.Figure(data=[go.Scatter(x=data['Year'], y=data['Predicted T-score'], mode='lines+markers', name='T-score')])
+        fig_tscore.update_layout(title=f"{drug} - Predicted T-score over Time", xaxis_title="Year", yaxis_title="T-score")
+        st.plotly_chart(fig_tscore)
 
 # Streamlit UI
 def main():
     st.title("BMD and T-score Prediction Tool")
-
+    
     # Input patient data
-    bmd_patient = st.number_input("Initial BMD", min_value=0.0, max_value=2.0, value=0.635, step=0.001, format="%.3f")
-    tscore_patient = st.number_input("Initial T-score", min_value=-5.0, max_value=2.0, value=-2.5, step=0.01, format="%.2f")
-
-    # Site options (Total Hip, Femoral Neck, Lumbar Spine)
-    site_options = ['Total Hip', 'Femoral Neck', 'Lumbar Spine (L1-L4)']
-    site = st.selectbox("Select site", site_options)
-
-    # Drug options
-    drug_options = ['Teriparatide', 'Teriparatide + Denosumab', 'Denosumab', 'Denosumab + Teriparatide',
-                    'Romosozumab', 'Romosozumab + Denosumab', 'Romosozumab + Alendronate',
-                    'Romosozumab + Ibandronate', 'Romosozumab + Zoledronate', 'Alendronate',
-                    'Risedronate', 'Ibandronate oral', 'Ibandronate IV (3mg)', 'Zoledronate']
-
-    # Add option to select multiple drugs
-    selected_drugs = st.multiselect("Select drugs to compare", drug_options)
-
-    # Set constants for each site
-    if site == 'Total Hip':
-        C_avg_lunar = 0.95  # Example: Average BMD for Total Hip from Excel (Lunar)
-        C_sd_lunar = 0.12
-    elif site == 'Femoral Neck':
-        C_avg_lunar = 0.905  # Example: Average BMD for Femoral Neck from Excel (Lunar)
-        C_sd_lunar = 0.116   # Example: SD for Femoral Neck (Lunar)
-    elif site == 'Lumbar Spine (L1-L4)':
-        C_avg_lunar = 1.097  # Example: Average BMD for Lumbar Spine (L1-L4) from Excel (Lunar)
-        C_sd_lunar = 0.128   # Example: SD for Lumbar Spine (L1-L4) (Lunar)
-
-    # Example file path
-    file_path = "BMD constant calculator.xlsx"
-
-    # ตรวจสอบว่ามีการเลือกยาแล้วหรือไม่
-    if len(selected_drugs) == 0:
-        st.warning("Please select at least one drug to compare.")
-    else:
-        # Run prediction and plot results
-        if st.button("Predict"):
-            main_with_separate_tables(file_path, bmd_patient, tscore_patient, C_avg_lunar, C_sd_lunar, selected_drugs, site)
+    bmd_patient = st.number_input("Initial BMD", min_value=0.0, max_value=2.0, value=0.8, step=0.01)
+    site_options = ['FN', 'TH', 'LS']
+    site = st.selectbox("Select Region (Site)", site_options)
+    
+    # Load constants and medication data
+    constants = REGRESSION_CONSTANTS[site]
+    medication_data = load_medication_data()
+    
+    # Generate and display predictions
+    if st.button("Predict"):
+        predictions = generate_predictions(medication_data, site, bmd_patient, constants['mu'], constants['sigma'])
+        display_results(predictions, site)
 
 if __name__ == "__main__":
     main()