File size: 13,071 Bytes
683d635
 
4b79b85
683d635
ce08beb
 
 
 
683d635
ce08beb
 
 
 
 
 
683d635
ce08beb
683d635
 
ce08beb
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
683d635
ce08beb
 
 
 
 
 
683d635
ce08beb
 
 
 
 
 
 
 
 
 
 
 
 
 
 
683d635
ce08beb
683d635
ce08beb
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
683d635
ce08beb
683d635
ce08beb
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
4b79b85
ce08beb
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
import streamlit as st
import pandas as pd
import numpy as np
import joblib
import json
import plotly.express as px
import plotly.graph_objects as go
from datetime import datetime

# Page config
st.set_page_config(
    page_title="TMJ Injection Success Predictor",
    page_icon="๐Ÿ’‰",
    layout="wide"
)

# Load model and materials
@st.cache_resource
def load_artifacts():
    """Load the trained model and materials list"""
    try:
        # Load model
        model = joblib.load('src/best_tmj_success_classifier_without_fe.pkl')
        
        # Load materials list
        try:
            with open('src/material_list.json', 'r') as f:
                materials_data = json.load(f)
                materials = materials_data.get('materials', [])
        except FileNotFoundError:
            # Fallback to default materials
            materials = ['Local Anaesthesia', 'Dry Needle', 'Botox', 
                        'Saline', 'Magnesium', 'PRF']
            st.warning("Using default materials list. Train the model to generate actual materials from your data.")
        
        # Load metadata if available
        metadata = {}
        try:
            with open('model_metadata.json', 'r') as f:
                metadata = json.load(f)
        except FileNotFoundError:
            pass
        
        return model, materials, metadata
    except Exception as e:
        st.error(f"Error loading model: {str(e)}")
        st.stop()

# Initialize
model, materials, metadata = load_artifacts()

# Title and description
st.title("๐Ÿฆท TMJ Injection Success Predictor")
st.markdown("""
This tool predicts the 3-month treatment success probability for TMJ injections based on patient baseline characteristics.
Enter the patient information below to see predictions for different injection materials.
""")

# Display model info if available
if metadata:
    with st.expander("โ„น๏ธ Model Information"):
        col1, col2, col3 = st.columns(3)
        with col1:
            st.metric("Model Type", metadata.get('model_type', 'Unknown'))
        with col2:
            st.metric("Test ROC-AUC", f"{metadata.get('test_roc_auc', 0):.3f}")
        with col3:
            st.metric("Training Date", metadata.get('training_date', 'Unknown')[:10])
        
        st.write(f"**Success Definition:** {metadata.get('success_definition', 'Unknown')}")
        
        if metadata.get('simplified_version', False):
            st.info("This model uses the simplified feature set without text analysis.")

st.divider()

# Create form
with st.form("patient_form"):
    st.subheader("Patient Information")
    
    # Required fields
    col1, col2 = st.columns(2)
    
    with col1:
        st.markdown("**Required Fields**")
        sex = st.selectbox("Sex", options=['Male', 'Female'], help="Patient's biological sex")
        age = st.number_input("Age", min_value=10, max_value=100, value=45, help="Patient age in years")
        pain_m0 = st.slider("Baseline Pain (M0)", min_value=0, max_value=10, value=7, 
                           help="Pain score at baseline (0-10 scale)")
    
    with col2:
        st.markdown("** **")  # Empty space to align with "Required Fields"
        mmo_m0 = st.slider("Baseline MMO (M0)", min_value=0, max_value=80, value=35,
                          help="Maximum mouth opening at baseline (mm)")
        ohip_14_m0 = st.slider("Baseline OHIP-14 (M0)", min_value=0, max_value=56, value=28,
                              help="Oral Health Impact Profile score at baseline (0-56)")
    
    st.divider()
    
    # Optional fields
    st.markdown("**Optional Fields**")
    col3, col4 = st.columns(2)
    
    with col3:
        location = st.text_input("Location", placeholder="e.g., Right TMJ",
                                help="Injection location")
        muscle_injected = st.text_input("Muscle Injected", placeholder="e.g., Masseter",
                                      help="Specific muscle targeted")
        adjunctive_treatment = st.text_input("Adjunctive Treatment", placeholder="e.g., Physical therapy",
                                           help="Additional treatments")
    
    with col4:
        previous_injection = st.selectbox("Previous Injection", options=['No', 'Yes'],
                                        help="Has the patient had previous TMJ injections?")
        if previous_injection == 'Yes':
            material_in_previous_injection = st.selectbox("Previous Material", 
                                                         options=[''] + materials,
                                                         help="Material used in previous injection")
        else:
            material_in_previous_injection = ''
    
    st.divider()
    
    # Material selection for primary prediction
    st.markdown("**Primary Prediction**")
    selected_material = st.selectbox("Select Material for Prediction", 
                                   options=materials,
                                   help="Choose the material you're considering for this patient")
    
    # Compare all materials option
    compare_all = st.checkbox("Compare all available materials", value=True,
                            help="Show predictions for all materials to help with decision making")
    
    # Submit button
    submitted = st.form_submit_button("๐Ÿ”ฎ Predict Success", use_container_width=True, type="primary")

# Process form submission
if submitted:
    # Create input dataframe
    input_data = pd.DataFrame({
        'sex': [sex],
        'age': [age],
        'pain_m0': [pain_m0],
        'mmo_m0': [mmo_m0],
        'ohip_14_m0': [ohip_14_m0],
        'location': [location if location else np.nan],
        'muscle_injected': [muscle_injected if muscle_injected else np.nan],
        'adjunctive_treatment': [adjunctive_treatment if adjunctive_treatment else np.nan],
        'previous_injection': [1 if previous_injection == 'Yes' else 0],
        'material_in_previous_injection': [material_in_previous_injection if material_in_previous_injection else np.nan],
        'material_injected': [selected_material]
    })
    
    # Make prediction for selected material
    try:
        prediction_proba = model.predict_proba(input_data)[0, 1]
        
        # Display primary prediction
        st.divider()
        st.subheader("Prediction Results")
        
        # Create a visual indicator
        col1, col2, col3 = st.columns([1, 2, 1])
        with col2:
            # Success probability gauge
            fig = go.Figure(go.Indicator(
                mode = "gauge+number+delta",
                value = prediction_proba * 100,
                domain = {'x': [0, 1], 'y': [0, 1]},
                title = {'text': f"Success Probability with {selected_material}"},
                number = {'suffix': "%", 'font': {'size': 40}},
                gauge = {
                    'axis': {'range': [None, 100]},
                    'bar': {'color': "darkblue"},
                    'steps': [
                        {'range': [0, 30], 'color': "lightgray"},
                        {'range': [30, 70], 'color': "gray"},
                        {'range': [70, 100], 'color': "lightgreen"}
                    ],
                    'threshold': {
                        'line': {'color': "red", 'width': 4},
                        'thickness': 0.75,
                        'value': 50
                    }
                }
            ))
            fig.update_layout(height=400)
            st.plotly_chart(fig, use_container_width=True)
        
        # Interpretation
        if prediction_proba >= 0.7:
            st.success(f"โœ… High likelihood of success ({prediction_proba:.1%}) with {selected_material}")
        elif prediction_proba >= 0.5:
            st.warning(f"โš ๏ธ Moderate likelihood of success ({prediction_proba:.1%}) with {selected_material}")
        else:
            st.error(f"โŒ Low likelihood of success ({prediction_proba:.1%}) with {selected_material}")
        
        # Compare all materials if requested
        if compare_all:
            st.divider()
            st.subheader("๐Ÿ“Š Material Comparison")
            
            # Predict for all materials
            material_results = []
            for material in materials:
                temp_data = input_data.copy()
                temp_data['material_injected'] = material
                prob = model.predict_proba(temp_data)[0, 1]
                material_results.append({
                    'Material': material,
                    'Success Probability': prob,
                    'Success %': f"{prob:.1%}"
                })
            
            # Sort by probability
            material_df = pd.DataFrame(material_results)
            material_df = material_df.sort_values('Success Probability', ascending=False)
            
            # Display results
            col1, col2 = st.columns([1, 1])
            
            with col1:
                # Table view
                st.markdown("**Ranked Materials**")
                display_df = material_df[['Material', 'Success %']].reset_index(drop=True)
                display_df.index += 1  # Start index at 1
                st.dataframe(display_df, use_container_width=True)
                
                # Highlight best option
                best_material = material_df.iloc[0]['Material']
                best_prob = material_df.iloc[0]['Success Probability']
                if best_material != selected_material:
                    st.info(f"๐Ÿ’ก Consider using **{best_material}** for potentially better outcomes ({best_prob:.1%} vs {prediction_proba:.1%})")
            
            with col2:
                # Bar chart
                st.markdown("**Visual Comparison**")
                fig = px.bar(material_df, 
                           x='Success Probability', 
                           y='Material',
                           orientation='h',
                           color='Success Probability',
                           color_continuous_scale='RdYlGn',
                           range_color=[0, 1],
                           text='Success %')
                
                fig.update_traces(textposition='outside')
                fig.update_layout(
                    xaxis_title="Success Probability",
                    yaxis_title="",
                    showlegend=False,
                    xaxis=dict(range=[0, 1.1]),
                    height=400
                )
                
                # Add vertical line at 50%
                fig.add_vline(x=0.5, line_dash="dash", line_color="gray", 
                            annotation_text="50% threshold")
                
                st.plotly_chart(fig, use_container_width=True)
        
        # Additional insights
        st.divider()
        with st.expander("๐Ÿ“‹ Patient Summary"):
            st.write("**Baseline Characteristics:**")
            summary_cols = st.columns(3)
            with summary_cols[0]:
                st.write(f"- Age: {age} years")
                st.write(f"- Sex: {sex}")
                st.write(f"- Previous injection: {previous_injection}")
            with summary_cols[1]:
                st.write(f"- Pain score: {pain_m0}/10")
                st.write(f"- MMO: {mmo_m0} mm")
                st.write(f"- OHIP-14: {ohip_14_m0}/56")
            with summary_cols[2]:
                if location:
                    st.write(f"- Location: {location}")
                if muscle_injected:
                    st.write(f"- Muscle: {muscle_injected}")
                if adjunctive_treatment:
                    st.write(f"- Adjunctive: {adjunctive_treatment}")
        
    except Exception as e:
        st.error(f"Error making prediction: {str(e)}")
        st.info("Please ensure the model was trained with all the necessary features.")

# Footer
st.divider()
st.markdown("""
<div style='text-align: center; color: gray;'>
    <small>
    TMJ Injection Success Predictor | 
    Model trained on historical patient data | 
    Predictions are probabilistic and should be used alongside clinical judgment
    </small>
</div>
""", unsafe_allow_html=True)

# Sidebar with instructions
with st.sidebar:
    st.header("๐Ÿ“– Instructions")
    st.markdown("""
    1. **Enter patient baseline data** in the form
    2. **Select the material** you're considering
    3. **Click Predict** to see the success probability
    4. **Compare materials** to find the optimal choice
    
    ---
    
    ### ๐ŸŽฏ Success Definition
    Treatment success is typically defined as:
    - Pain reduction > 2 points
    - MMO increase > 5 mm  
    - OHIP-14 reduction > 5 points
    
    ---
    
    ### ๐Ÿ“Š Interpretation Guide
    - **70%+**: High success likelihood โœ…
    - **50-70%**: Moderate success โš ๏ธ
    - **<50%**: Low success likelihood โŒ
    
    ---
    
    ### โš•๏ธ Clinical Note
    These predictions are based on statistical models and should complement, not replace, clinical expertise and patient-specific considerations.
    """)