amonfortc commited on
Commit
15e8b42
·
0 Parent(s):

Add all files

Browse files
README.md.txt ADDED
@@ -0,0 +1,154 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ title: FibroPred Predictive System
3
+ emoji: 🩺
4
+ colorFrom: indigo
5
+ colorTo: blue
6
+ sdk: gradio
7
+ sdk_version: "3.35.0"
8
+ app_file: app.py
9
+ pinned: false
10
+ ---
11
+
12
+ ### README: **FibroPred Predictive System**
13
+
14
+ ---
15
+
16
+ #### **Description**
17
+ FibroPred is a predictive system designed to analyze clinical data and provide predictions for specific medical outcomes related to fibrosis. The tool employs machine learning models, primarily Random Forest classifiers, to forecast the likelihood of various events or conditions, such as mortality, necessity for transplantation, and progressive disease.
18
+
19
+ The system provides an easy-to-use interface built with Gradio, allowing users to input feature values and obtain predictions. Models and their configurations are stored and loaded dynamically, ensuring modularity and adaptability.
20
+
21
+ ---
22
+
23
+ #### **Features**
24
+ 1. **Prediction Targets**:
25
+ - **Death**: Likelihood of patient mortality based on clinical and diagnostic features.
26
+ - **Binary Diagnosis**: Classification of patients into specific diagnostic categories.
27
+ - **Necessity of Transplantation**: Assessment of whether a patient is likely to need a transplant.
28
+ - **Progressive Disease**: Prediction of disease progression based on longitudinal data.
29
+
30
+ 2. **Dynamic Model Loading**: Models and feature sets are loaded dynamically from a pre-configured directory (`models`).
31
+
32
+ 3. **Gradio Interface**: A tabbed interface for each prediction target, where users can input values interactively and receive predictions in real-time.
33
+
34
+ ---
35
+
36
+ #### **Predictive Features**
37
+ Below are the features used for prediction across all targets:
38
+
39
+ 1. **Pedigree** (0 - 67):
40
+ Represents the familial history related to fibrotic conditions. Higher values indicate a greater genetic predisposition.
41
+
42
+ 2. **Age at diagnosis** (36.0 - 92.0):
43
+ Age of the patient at the time of diagnosis. A critical factor as progression and treatment response vary with age.
44
+
45
+ 3. **FVC (L) at diagnosis** (0.0 - 5.0):
46
+ Forced vital capacity in liters at the time of diagnosis, reflecting lung function.
47
+
48
+ 4. **FVC (%) at diagnosis** (0.0 - 200.0):
49
+ Forced vital capacity as a percentage of the expected value for the patient’s age and sex.
50
+
51
+ 5. **DLCO (%) at diagnosis** (0.0 - 200.0):
52
+ Diffusion capacity for carbon monoxide as a percentage, measuring gas exchange efficiency in the lungs.
53
+
54
+ 6. **RadioWorsening2y** (0 - 3):
55
+ Radiological assessment of lung deterioration over two years. Higher values indicate significant progression.
56
+
57
+ 7. **Severity of telomere shortening - Transform 4** (1 - 6):
58
+ Indicates the degree of telomere shortening. Higher values reflect greater telomeric damage.
59
+
60
+ 8. **Progressive disease** (0 - 1):
61
+ Binary variable indicating whether the disease is progressive (1) or stable (0).
62
+
63
+ 9. **Antifibrotic Drug** (0 - 1):
64
+ Binary variable representing the use of antifibrotic drugs. 1 indicates use; 0 indicates none.
65
+
66
+ 10. **Prednisone** (0 - 1):
67
+ Binary variable reflecting prednisone usage. 1 indicates use; 0 indicates none.
68
+
69
+ 11. **Mycophenolate** (0 - 1):
70
+ Binary variable indicating mycophenolate usage. 1 indicates use; 0 indicates none.
71
+
72
+ 12. **FVC (L) 1 year after diagnosis** (0.0 - 5.0):
73
+ Forced vital capacity in liters one year after diagnosis, used to evaluate changes in lung function.
74
+
75
+ 13. **FVC (%) 1 year after diagnosis** (0.0 - 200.0):
76
+ Forced vital capacity as a percentage one year after diagnosis.
77
+
78
+ 14. **DLCO (%) 1 year after diagnosis** (0.0 - 200.0):
79
+ Diffusion capacity for carbon monoxide as a percentage one year after diagnosis.
80
+
81
+ 15. **Genetic mutation studied in patient** (0 - 1):
82
+ Binary variable indicating the presence of specific genetic mutations. 1 indicates mutation found; 0 indicates none.
83
+
84
+ 16. **Comorbidities** (0 - 1):
85
+ Binary variable representing the presence of relevant comorbidities. 1 indicates presence; 0 indicates absence.
86
+
87
+ ---
88
+
89
+ #### **Setup Instructions**
90
+ 1. Clone or download the repository.
91
+ 2. Ensure Python 3.8+ is installed.
92
+ 3. Install dependencies using the command:
93
+ ```bash
94
+ pip install -r requirements.txt
95
+ ```
96
+ 4. Place trained models in the `models` directory. Models should be `.pkl` files containing:
97
+ - `model`: Trained Random Forest model.
98
+ - `features`: Feature list used during model training.
99
+
100
+ ---
101
+
102
+ #### **Usage**
103
+ 1. Run the application:
104
+ ```bash
105
+ python app.py
106
+ ```
107
+ 2. Access the Gradio interface through the displayed local or public URL.
108
+ 3. Select a prediction tab, input feature values, and click "Submit" to get predictions.
109
+
110
+ ---
111
+
112
+ #### **Key Scripts**
113
+ 1. **app.py**:
114
+ - Implements the Gradio interface.
115
+ - Maps user-friendly model names to actual models and their features.
116
+
117
+ 2. **fibropred_model.py**:
118
+ - Contains the preprocessing pipeline, including imputation and feature selection.
119
+ - Includes functions for training, evaluation, and visualization.
120
+
121
+ 3. **model_utils.py**:
122
+ - Functions to load models and their features dynamically.
123
+ - Handles predictions using preloaded models.
124
+
125
+ 4. **requirements.txt**:
126
+ - Lists the Python dependencies required for the system.
127
+
128
+ ---
129
+
130
+ #### **Technical Highlights**
131
+ - **Machine Learning Framework**: Models are built using `scikit-learn`, leveraging Random Forest classifiers for robust predictions.
132
+ - **Visualization**: The script includes utilities for plotting feature importance, ROC-AUC curves, and overfitting diagnostics.
133
+ - **Dynamic Handling**: Feature lists and models are dynamically linked, ensuring flexibility when adding new prediction targets.
134
+
135
+ ---
136
+
137
+ #### **Future Improvements**
138
+ - Expand model capabilities with additional machine learning algorithms (e.g., Gradient Boosting).
139
+ - Integrate longitudinal data analysis for enhanced prediction accuracy.
140
+ - Add options for model retraining with new data directly through the interface.
141
+
142
+ ---
143
+
144
+ This README provides a comprehensive guide to understanding and using the **FibroPred** predictive system effectively.
145
+
146
+ ---
147
+
148
+ #### **Setup Instructions**
149
+ 1. Clone or download the repository.
150
+ 2. Ensure Python 3.8+ is installed.
151
+ 3. Install dependencies using the command:
152
+ ```bash
153
+ pip install -r requirements.txt
154
+
app.py ADDED
@@ -0,0 +1,115 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ import gradio as gr
2
+ from model_utils import load_all_models, predict_with_model
3
+
4
+ # Load all models
5
+ models, model_features = load_all_models()
6
+
7
+ # Mapeo de nombres amigables a nombres reales
8
+ MODEL_MAPPING = {
9
+ "Death": "Death_random_forest_model",
10
+ "Binary diagnosis": "Binary diagnosis_random_forest_model",
11
+ "Necessity of transplantation": "Necessity of transplantation_random_forest_model",
12
+ "Progressive disease": "Progressive disease_random_forest_model"
13
+ }
14
+
15
+ # Invertir el mapeo (opcional para facilidad)
16
+ INVERSE_MODEL_MAPPING = {v: k for k, v in MODEL_MAPPING.items()}
17
+
18
+ # Feature sets for each target variable
19
+ FEATURES = {
20
+ "Death": [
21
+ 'Pedigree', 'Age at diagnosis', 'FVC (L) at diagnosis',
22
+ 'FVC (%) at diagnosis', 'DLCO (%) at diagnosis', 'RadioWorsening2y',
23
+ 'Severity of telomere shortening - Transform 4', 'Progressive disease'
24
+ ],
25
+ "Binary diagnosis": [
26
+ 'Pedigree', 'Age at diagnosis', 'Antifibrotic Drug',
27
+ 'Prednisone', 'Mycophenolate', 'FVC (L) at diagnosis',
28
+ 'FVC (%) at diagnosis', 'DLCO (%) at diagnosis'
29
+ ],
30
+ "Necessity of transplantation": [
31
+ 'Pedigree','Age at diagnosis','FVC (L) at diagnosis', 'FVC (%) at diagnosis', 'DLCO (%) at diagnosis',
32
+ 'FVC (L) 1 year after diagnosis','FVC (%) 1 year after diagnosis','DLCO (%) 1 year after diagnosis',
33
+ 'RadioWorsening2y'
34
+ ],
35
+ "Progressive disease": [
36
+ 'Pedigree', 'Age at diagnosis', 'FVC (L) at diagnosis','FVC (%) at diagnosis', 'DLCO (%) at diagnosis','FVC (L) 1 year after diagnosis',
37
+ 'FVC (%) 1 year after diagnosis', 'DLCO (%) 1 year after diagnosis',
38
+ 'RadioWorsening2y', 'Genetic mutation studied in patient'
39
+ ]
40
+
41
+ }
42
+
43
+ FEATURE_RANGES = {
44
+ 'Pedigree': (0, 67),
45
+ 'Age at diagnosis': (36.0, 92.0),
46
+ 'FVC (L) at diagnosis': (0.0, 5.0),
47
+ 'FVC (%) at diagnosis': (0.0, 200.0),
48
+ 'DLCO (%) at diagnosis': (0.0, 200.0),
49
+ 'RadioWorsening2y': (0, 3),
50
+ 'Severity of telomere shortening - Transform 4': (1, 6),
51
+ 'Progressive disease': (0, 1),
52
+ 'Antifibrotic Drug': (0, 1),
53
+ 'Prednisone': (0, 1),
54
+ 'Mycophenolate': (0, 1),
55
+ 'FVC (L) 1 year after diagnosis': (0.0, 5.0),
56
+ 'FVC (%) 1 year after diagnosis': (0.0, 200.0),
57
+ 'DLCO (%) 1 year after diagnosis': (0.0, 200.0),
58
+ 'Genetic mutation studied in patient': (0, 1),
59
+ 'Comorbidities': (0, 1)
60
+ }
61
+
62
+
63
+ # Define prediction function
64
+ def make_prediction(input_features, friendly_model_name):
65
+ """
66
+ Predict using the selected model and input features.
67
+ """
68
+ # Map the friendly model name to the real model name
69
+ target_model = MODEL_MAPPING.get(friendly_model_name)
70
+ if target_model not in models:
71
+ return f"Model '{friendly_model_name}' not found. Please select a valid model."
72
+
73
+ model = models[target_model]
74
+ features = model_features[target_model]
75
+
76
+ if len(input_features) != len(features):
77
+ return f"Invalid input. Expected features: {features}"
78
+
79
+ input_array = [float(x) for x in input_features]
80
+ prediction = predict_with_model(model, input_array)
81
+ return f"Prediction for {friendly_model_name}: {prediction}"
82
+
83
+ # Define Gradio interface
84
+ def gradio_interface():
85
+ def create_inputs_for_features(features):
86
+ inputs = []
87
+ for feature in features:
88
+ min_val, max_val = FEATURE_RANGES.get(feature, (None, None))
89
+ inputs.append(gr.Number(label=f"{feature} (Range: {min_val} - {max_val})", minimum=min_val, maximum=max_val))
90
+ return inputs
91
+
92
+ # Create a separate interface for each target variable
93
+ interfaces = []
94
+ for target, features in FEATURES.items():
95
+ inputs = create_inputs_for_features(features)
96
+ interface = gr.Interface(
97
+ fn=lambda *args, target=target: make_prediction(args, target),
98
+ inputs=inputs,
99
+ outputs=gr.Text(label="Prediction Result"),
100
+ title=f"Prediction for {target}",
101
+ description=f"Provide values for features relevant to {target}"
102
+ )
103
+ interfaces.append(interface)
104
+
105
+ # Combine all interfaces into a tabbed layout
106
+ tabbed_interface = gr.TabbedInterface(
107
+ interface_list=interfaces,
108
+ tab_names=list(FEATURES.keys())
109
+ )
110
+ return tabbed_interface
111
+
112
+ # Launch Gradio app
113
+ if __name__ == "__main__":
114
+ interface = gradio_interface()
115
+ interface.launch()
fibropred_model.py ADDED
@@ -0,0 +1,195 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ import pandas as pd
2
+ import numpy as np
3
+ import os
4
+ import joblib
5
+ from sklearn.model_selection import train_test_split, cross_val_score, StratifiedKFold
6
+ from sklearn.ensemble import RandomForestClassifier, GradientBoostingClassifier
7
+ from sklearn.impute import SimpleImputer
8
+ from sklearn.metrics import classification_report, accuracy_score, roc_curve, auc
9
+ from sklearn.feature_selection import SelectFromModel
10
+ import matplotlib.pyplot as plt
11
+ import seaborn as sns
12
+
13
+ # Load dataset
14
+ def load_data(file_path):
15
+ df = pd.read_excel(file_path, header=1)
16
+ return df
17
+
18
+ # Preprocess data including categorical variables
19
+ def preprocess_data_with_categoricals(df):
20
+ # Replace -9 with NaN for missing values
21
+ df.replace(-9, np.nan, inplace=True)
22
+
23
+ # Drop columns with >50% missing values
24
+ missing_percentage = df.isnull().sum() / len(df) * 100
25
+ df = df.drop(columns=missing_percentage[missing_percentage > 50].index)
26
+
27
+ # Drop specific columns
28
+ drop_columns = ['ProgressiveDisease', 'Final diagnosis', 'Transplantation date', 'Cause of death', 'Date of death', 'COD NUMBER']
29
+ df = df.drop(columns=[col for col in drop_columns if col in df.columns])
30
+
31
+ # Impute missing values
32
+ imputer = SimpleImputer(strategy='median')
33
+ numeric_cols = df.select_dtypes(include=['number']).columns
34
+ df[numeric_cols] = imputer.fit_transform(df[numeric_cols])
35
+
36
+ # Handle binary variables specifically
37
+ if 'Binary diagnosis' in df.columns:
38
+ df['Binary diagnosis'] = df['Binary diagnosis'].apply(
39
+ lambda x: 1 if str(x).strip().lower() == "ipf" else 0
40
+ )
41
+
42
+ if 'Death' in df.columns:
43
+ df['Death'] = df['Death'].apply(
44
+ lambda x: 1 if str(x).strip().lower() == "yes" else 0
45
+ )
46
+
47
+ # Apply one-hot encoding to categorical variables
48
+ df = apply_one_hot_encoding(df)
49
+
50
+ # Separate categorical and numerical variables
51
+ categorical_cols = df.select_dtypes(include=['object']).columns
52
+ numeric_cols = df.select_dtypes(include=['number']).columns
53
+ print("Categorical Variables:", categorical_cols.tolist())
54
+ print("Numerical Variables:", numeric_cols.tolist())
55
+ return df, numeric_cols, categorical_cols
56
+
57
+ # Apply one-hot encoding to categorical variables
58
+ def apply_one_hot_encoding(df):
59
+ categorical_cols = df.select_dtypes(include=['object']).columns
60
+ df = pd.get_dummies(df, columns=categorical_cols, drop_first=True)
61
+ return df
62
+
63
+ # Select predictors using feature importance
64
+ def select_important_features(X, y, threshold=0.03):
65
+ model = RandomForestClassifier(random_state=42)
66
+ model.fit(X, y)
67
+ selector = SelectFromModel(model, threshold=threshold, prefit=True)
68
+ X_reduced = selector.transform(X)
69
+ selected_features = X.columns[selector.get_support()]
70
+ return pd.DataFrame(X_reduced, columns=selected_features), selected_features
71
+
72
+ # Visualize feature importance
73
+ def plot_feature_importance(model, features, target):
74
+ importance = model.feature_importances_
75
+ sorted_idx = np.argsort(importance)[::-1]
76
+ plt.figure(figsize=(10, 6))
77
+ sns.barplot(x=importance[sorted_idx], y=np.array(features)[sorted_idx])
78
+ plt.title(f'Feature Importance for {target}')
79
+ plt.xlabel('Importance')
80
+ plt.ylabel('Feature')
81
+ plt.tight_layout()
82
+ plt.show()
83
+
84
+ # Visualize overfitting and optimization results
85
+ def plot_model_performance(cv_scores, train_scores, test_scores, target ,metric_name="Accuracy"):
86
+ plt.figure(figsize=(12, 6))
87
+
88
+ # Cross-validation scores
89
+ plt.subplot(1, 2, 1)
90
+ plt.plot(cv_scores, label='Cross-validation scores', marker='o')
91
+ plt.title(f'Cross-validation {metric_name} for {target}')
92
+ plt.xlabel('Fold')
93
+ plt.ylabel(metric_name)
94
+ plt.grid(True)
95
+ plt.legend()
96
+
97
+ # Train vs Test comparison
98
+ plt.subplot(1, 2, 2)
99
+ plt.bar(['Train', 'Test'], [train_scores.mean(), test_scores], color=['blue', 'orange'])
100
+ plt.title(f'{metric_name}: Train vs Test')
101
+ plt.ylabel(metric_name)
102
+ plt.grid(True)
103
+
104
+ plt.tight_layout()
105
+ plt.show()
106
+
107
+ # Plot ROC-AUC curve
108
+ def plot_roc_auc(model, X_test, y_test, target):
109
+ y_prob = model.predict_proba(X_test)[:, 1] # Probabilidades para la clase positiva
110
+ fpr, tpr, thresholds = roc_curve(y_test, y_prob)
111
+ roc_auc = auc(fpr, tpr)
112
+
113
+ plt.figure(figsize=(8, 6))
114
+ plt.plot(fpr, tpr, color='blue', lw=2, label=f'ROC curve (area = {roc_auc:.2f})')
115
+ plt.plot([0, 1], [0, 1], color='gray', linestyle='--')
116
+ plt.xlabel('False Positive Rate')
117
+ plt.ylabel('True Positive Rate')
118
+ plt.title(f'ROC-AUC Curve for {target}')
119
+ plt.legend(loc="lower right")
120
+ plt.grid(True)
121
+ plt.show()
122
+
123
+ # Save trained model
124
+ def save_model(model, target, selected_features):
125
+
126
+ if not os.path.exists("models"):
127
+ os.makedirs("models")
128
+ file_name = f"models/{target}_random_forest_model.pkl"
129
+ joblib.dump({'model': model, 'features': selected_features}, file_name)
130
+ print(f"Model and features saved to {file_name}")
131
+
132
+
133
+ # Main pipeline
134
+ def main():
135
+ file_path = 'FibroPredCODIFICADA.xlsx'
136
+ df = load_data(file_path)
137
+
138
+ # Target columns
139
+ target_columns = ['Death', 'Progressive disease', 'Necessity of transplantation']
140
+
141
+ # Preprocess data
142
+ df, numeric_cols, categorical_cols = preprocess_data_with_categoricals(df)
143
+
144
+ for target in target_columns:
145
+ print(f"Processing target: {target}")
146
+ X = df[numeric_cols].drop(columns=target_columns, errors='ignore') # Ensure target variables are excluded
147
+ y = df[target]
148
+
149
+ # Split data
150
+ X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)
151
+
152
+ # Select important features
153
+ X_train_selected, selected_features = select_important_features(X_train, y_train)
154
+ X_test_selected = X_test[selected_features]
155
+
156
+ print(f"Selected predictors for training {target} ({len(selected_features)} predictors): {selected_features.tolist()}")
157
+
158
+ # Train RandomForest model
159
+ model = RandomForestClassifier(n_estimators=300,
160
+ max_depth=4,
161
+ min_samples_split=10,
162
+ min_samples_leaf=10,
163
+ class_weight='balanced',
164
+ max_features='sqrt',
165
+ random_state=42)
166
+ model.fit(X_train_selected, y_train)
167
+
168
+ # Cross-validation to check overfitting
169
+ cv = StratifiedKFold(n_splits=15, shuffle=True, random_state=42)
170
+ cv_scores = cross_val_score(model, X_train_selected, y_train, cv=cv, scoring='accuracy')
171
+ train_scores = cross_val_score(model, X_train_selected, y_train, cv=15, scoring='accuracy')
172
+ y_pred_test = model.predict(X_test_selected)
173
+ test_score = accuracy_score(y_test, y_pred_test)
174
+
175
+ print(f"Cross-validation accuracy for {target}: {cv_scores.mean():.4f} (+/- {cv_scores.std():.4f})")
176
+ print(f"Test accuracy for {target}: {test_score:.4f}")
177
+ print(classification_report(y_test, y_pred_test))
178
+
179
+ # Plot model performance
180
+ plot_model_performance(cv_scores, train_scores, test_score, target, metric_name="Accuracy")
181
+
182
+ # Plot feature importance
183
+ print(f"Feature importance for {target}:")
184
+ plot_feature_importance(model, selected_features, target)
185
+
186
+ # Plot ROC-AUC Curve
187
+ plot_roc_auc(model, X_test_selected, y_test, target)
188
+
189
+ # Save trained model
190
+ save_model(model, target, selected_features.tolist())
191
+
192
+ print("Pipeline completed.")
193
+
194
+ if __name__ == "__main__":
195
+ main()
model_utils.py ADDED
@@ -0,0 +1,34 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ import os
2
+ import joblib
3
+
4
+ def load_all_models(models_dir="models"):
5
+ """
6
+ Load all models and their features from the given directory.
7
+ """
8
+ models = {}
9
+ features = {}
10
+ if not os.path.exists(models_dir):
11
+ raise FileNotFoundError(f"Models directory '{models_dir}' not found.")
12
+
13
+ for model_file in os.listdir(models_dir):
14
+ if model_file.endswith(".pkl"):
15
+ model_name = os.path.splitext(model_file)[0]
16
+ data = joblib.load(os.path.join(models_dir, model_file))
17
+ models[model_name] = data['model']
18
+ features[model_name] = data['features']
19
+ print(f"Model '{model_name}' loaded successfully with features: {features[model_name]}")
20
+ return models, features
21
+
22
+ def predict_with_model(model, input_data):
23
+ """
24
+ Predict using a loaded model.
25
+
26
+ Parameters:
27
+ - model: The loaded model.
28
+ - input_data: A dictionary or Pandas DataFrame row containing input features.
29
+
30
+ Returns:
31
+ - prediction: Model prediction.
32
+ """
33
+ prediction = model.predict([input_data])
34
+ return int(prediction[0])
models/Binary diagnosis_random_forest_model.pkl ADDED
Binary file (452 kB). View file
 
models/Death_random_forest_model.pkl ADDED
Binary file (370 kB). View file
 
models/Necessity of transplantation_random_forest_model.pkl ADDED
Binary file (379 kB). View file
 
models/Progressive disease_random_forest_model.pkl ADDED
Binary file (450 kB). View file
 
requirements.txt.txt ADDED
@@ -0,0 +1,8 @@
 
 
 
 
 
 
 
 
 
1
+ gradio
2
+ joblib
3
+ scikit-learn
4
+ pandas
5
+ numpy
6
+ matplotlib
7
+ seaborn
8
+ openpyxl