Create app.py

app.py

@@ -0,0 +1,218 @@
+# app.py
+
+import streamlit as st
+import pandas as pd
+import numpy as np
+from sklearn.ensemble import RandomForestClassifier
+from sklearn.model_selection import train_test_split
+from sklearn.metrics import accuracy_score, roc_curve, auc
+from sklearn.decomposition import PCA
+import matplotlib.pyplot as plt
+import seaborn as sns
+
+# Set random seed for reproducibility
+np.random.seed(42)
+
+# Title of the app
+st.title('Breast Cancer Treatment Outcome Prediction')
+
+st.markdown("""
+This Streamlit app simulates a precision medicine system that predicts the effectiveness of treatment for breast cancer patients based on clinical and genomic data.
+
+**Disclaimer:** This is a simulation using synthetic data and is intended for educational purposes only. It should not be used for clinical decision-making.
+""")
+
+# --- Data Simulation ---
+
+st.header('1. Data Simulation')
+
+# Simulate clinical data
+num_samples = 500
+age = np.random.randint(30, 80, num_samples)
+tumor_size = np.random.uniform(0.5, 5.0, num_samples)
+lymph_nodes = np.random.randint(0, 10, num_samples)
+tumor_grade = np.random.randint(1, 4, num_samples)  # Grades 1 to 3
+er_status = np.random.choice([0, 1], num_samples)  # Estrogen Receptor status
+pr_status = np.random.choice([0, 1], num_samples)  # Progesterone Receptor status
+her2_status = np.random.choice([0, 1], num_samples)  # HER2 status
+
+# Simulate genomic data (expression levels of 50 genes)
+gene_expression = np.random.normal(0, 1, (num_samples, 50))
+
+# Simulate treatment outcomes (0 = non-responsive, 1 = responsive)
+outcome = (
+    (er_status == 1) &
+    (pr_status == 1) &
+    (her2_status == 0) &
+    (tumor_grade < 3) &
+    (tumor_size < 2.5)
+).astype(int)
+
+# Create a DataFrame for clinical data
+clinical_data = pd.DataFrame({
+    'Age': age,
+    'Tumor_Size': tumor_size,
+    'Lymph_Nodes': lymph_nodes,
+    'Tumor_Grade': tumor_grade,
+    'ER_Status': er_status,
+    'PR_Status': pr_status,
+    'HER2_Status': her2_status,
+    'Outcome': outcome
+})
+
+# Create a DataFrame for genomic data
+gene_columns = [f'Gene_{i}' for i in range(1, 51)]
+genomic_data = pd.DataFrame(gene_expression, columns=gene_columns)
+
+# Combine clinical and genomic data
+data = pd.concat([clinical_data, genomic_data], axis=1)
+
+st.write('Simulated Data Preview:')
+st.dataframe(data.head())
+
+# --- Exploratory Data Analysis (EDA) ---
+
+st.header('2. Exploratory Data Analysis')
+
+# Correlation Matrix
+st.subheader('Correlation Matrix')
+corr = data.corr()
+fig_corr, ax_corr = plt.subplots(figsize=(10, 8))
+sns.heatmap(corr.iloc[:8, :8], annot=True, fmt=".2f", cmap='coolwarm', ax=ax_corr)
+st.pyplot(fig_corr)
+
+# Outcome Distribution
+st.subheader('Outcome Distribution')
+st.bar_chart(data['Outcome'].value_counts())
+
+# --- Feature Selection ---
+
+st.header('3. Feature Selection and Preprocessing')
+
+# Features and Labels
+features = data.drop('Outcome', axis=1)
+labels = data['Outcome']
+
+# Optional: Dimensionality Reduction on Genomic Data
+pca_genes = PCA(n_components=5)
+genes_pca = pca_genes.fit_transform(genomic_data)
+
+# Create a new DataFrame with PCA components
+genes_pca_df = pd.DataFrame(genes_pca, columns=[f'PC{i}' for i in range(1, 6)])
+
+# Combine clinical data with PCA components
+features_pca = pd.concat([clinical_data.drop('Outcome', axis=1), genes_pca_df], axis=1)
+
+st.write('Features after PCA on Genomic Data:')
+st.dataframe(features_pca.head())
+
+# --- Model Development ---
+
+st.header('4. Model Development and Evaluation')
+
+# Split the data
+X_train, X_test, y_train, y_test = train_test_split(features_pca, labels, test_size=0.2, random_state=42)
+
+# Model Training
+model = RandomForestClassifier(n_estimators=100, random_state=42)
+model.fit(X_train, y_train)
+
+# Model Prediction
+y_pred = model.predict(X_test)
+accuracy = accuracy_score(y_test, y_pred)
+
+st.subheader('Model Accuracy')
+st.write(f'Accuracy on test set: **{accuracy * 100:.2f}%**')
+
+# ROC Curve
+st.subheader('ROC Curve')
+y_score = model.predict_proba(X_test)[:, 1]
+fpr, tpr, thresholds = roc_curve(y_test, y_score)
+roc_auc = auc(fpr, tpr)
+
+fig_roc, ax_roc = plt.subplots()
+ax_roc.plot(fpr, tpr, color='darkorange', lw=2, label=f'ROC curve (AUC = {roc_auc:.2f})')
+ax_roc.plot([0, 1], [0, 1], color='navy', lw=2, linestyle='--')
+ax_roc.set_xlim([0.0, 1.0])
+ax_roc.set_ylim([0.0, 1.05])
+ax_roc.set_xlabel('False Positive Rate')
+ax_roc.set_ylabel('True Positive Rate')
+ax_roc.set_title('Receiver Operating Characteristic')
+ax_roc.legend(loc='lower right')
+st.pyplot(fig_roc)
+
+# --- User Input Interface ---
+
+st.header('5. Predictive Simulation')
+
+st.sidebar.header('Input Patient Parameters')
+
+def user_input_features():
+    age = st.sidebar.slider('Age', 30, 80, 50)
+    tumor_size = st.sidebar.slider('Tumor Size (cm)', 0.5, 5.0, 2.0)
+    lymph_nodes = st.sidebar.slider('Number of Lymph Nodes', 0, 10, 2)
+    tumor_grade = st.sidebar.selectbox('Tumor Grade', [1, 2, 3])
+    er_status = st.sidebar.selectbox('Estrogen Receptor Status (ER)', [0, 1])
+    pr_status = st.sidebar.selectbox('Progesterone Receptor Status (PR)', [0, 1])
+    her2_status = st.sidebar.selectbox('HER2 Status', [0, 1])
+
+    # Simulate gene expression levels for PCA components
+    gene_inputs = np.random.normal(0, 1, (1, 50))
+    genes_pca_input = pca_genes.transform(gene_inputs)
+    genes_pca_input_df = pd.DataFrame(genes_pca_input, columns=[f'PC{i}' for i in range(1, 6)])
+    
+    data_input = pd.DataFrame({
+        'Age': [age],
+        'Tumor_Size': [tumor_size],
+        'Lymph_Nodes': [lymph_nodes],
+        'Tumor_Grade': [tumor_grade],
+        'ER_Status': [er_status],
+        'PR_Status': [pr_status],
+        'HER2_Status': [her2_status]
+    })
+
+    features_input = pd.concat([data_input, genes_pca_input_df], axis=1)
+    return features_input
+
+input_df = user_input_features()
+
+st.subheader('Input Parameters')
+st.write(input_df)
+
+# Prediction
+prediction = model.predict(input_df)
+prediction_proba = model.predict_proba(input_df)
+
+st.subheader('Prediction Outcome')
+outcome_label = np.array(['Non-Responsive', 'Responsive'])
+st.write(f'The model predicts that the patient is: **{outcome_label[prediction][0]}**')
+
+st.subheader('Prediction Probability')
+st.write(f'Probability of being Responsive: **{prediction_proba[0][1] * 100:.2f}%**')
+
+# --- Feature Importance ---
+
+st.header('6. Model Interpretation')
+
+st.subheader('Feature Importances')
+importances = model.feature_importances_
+indices = np.argsort(importances)[::-1]
+feature_names = features_pca.columns
+
+fig_fi, ax_fi = plt.subplots()
+ax_fi.bar(range(len(importances)), importances[indices], color='r', align='center')
+ax_fi.set_xticks(range(len(importances)))
+ax_fi.set_xticklabels(feature_names[indices], rotation=90)
+ax_fi.set_title('Feature Importances')
+ax_fi.set_ylabel('Importance Score')
+st.pyplot(fig_fi)
+
+# --- Conclusion ---
+
+st.header('7. Conclusion')
+
+st.markdown("""
+This simulation demonstrates how integrating clinical and genomic data can aid in predicting breast cancer treatment outcomes. By adjusting the input parameters, you can see how different factors influence the prediction.
+
+**Note:** This app uses synthetic data and a basic machine learning model. For real-world applications, a more sophisticated approach with validated data and models is required.
+""")