Merge pull request #175 from manikumarreddyu/mani

Health Track

Health care/Apps/Diabetes.py

@@ -0,0 +1,322 @@
+import time
+import streamlit as st
+import pandas as pd
+import Classifier_Models.Classifier_model_builder_diabetes as cmb
+import pickle
+import numpy as np
+from streamlit_toggle import st_toggle_switch
+import json
+from streamlit_lottie import st_lottie
+
+def app():
+    def load_lottiefile(filepath: str):
+        with open(filepath, "r") as f:
+            return json.load(f)
+
+    lottie_coding = load_lottiefile("res/Yoga_Feet.json")
+
+    st.title("Diabetes Detector")
+    st.info("This app predicts whether a person have any diabetes or not")
+    st.markdown("""
+    **Note** - :red[This Prediction Model is only applicable for Females.]
+    """)
+
+    st.sidebar.header('User Input Features')
+    # st.sidebar.markdown("""
+    # [Import input CSV file](https://github.com/ChakraDeep8/Heart-Disease-Detector/tree/master/res)""")
+
+    uploaded_file = st.sidebar.file_uploader("Upload your input CSV file", type=["csv"])
+
+    if uploaded_file is not None:
+        input_df = pd.read_csv(uploaded_file)
+    else:
+        def patient_details():
+            Pregnancies = st.sidebar.slider('Pregnancies', 0, 17)
+            Glucose = st.sidebar.slider('Glucose', 0, 199)
+            BloodPressure = st.sidebar.slider('BloodPressure', 0, 122)
+            SkinThickness = st.sidebar.slider('SkinThickness', 0, 99)
+            Insulin = st.sidebar.slider('Insulin', 0, 846)
+            BMI = st.sidebar.slider('BMI', 0.0, 67.1, step=0.1)
+            DiabetesPedigreeFunction = st.sidebar.slider('DiabetesPedigreeFunction', 0.08, 2.42, step=0.01)
+            Age = st.sidebar.slider('Age', 21, 81)
+
+            data = {'Pregnancies': Pregnancies,
+                    'Glucose': Glucose,
+                    'BloodPressure': BloodPressure,
+                    'SkinThickness': SkinThickness,
+                    'Insulin': Insulin,
+                    'BMI': BMI,
+                    'DiabetesPedigreeFunction': DiabetesPedigreeFunction,
+                    'Age': Age, }
+
+            features = pd.DataFrame(data, index=[0])
+            return features
+
+        input_df = patient_details()
+    heart = cmb.X
+    df = pd.concat([input_df, heart], axis=0)
+    df = df[:1]  # Selects only the first row (the user input data)
+    df.loc[:, ~df.columns.duplicated()]
+
+    if uploaded_file is not None:
+        st.write(df)
+
+    else:
+        st.write('Awaiting CSV file to be uploaded. Currently using example input parameters (shown below).')
+        df = df.loc[:, ~df.columns.duplicated()]
+        st.write(df)
+
+    # Load the classification models
+    load_clf_NB = pickle.load(open('res/pickle/diabetes_disease_classifier_NB.pkl', 'rb'))
+    load_clf_KNN = pickle.load(open('res/pickle/diabetes_disease_classifier_KNN.pkl', 'rb'))
+    load_clf_DT = pickle.load(open('res/pickle/diabetes_disease_classifier_DT.pkl', 'rb'))
+    load_clf_LR = pickle.load(open('res/pickle/diabetes_disease_classifier_LR.pkl', 'rb'))
+    load_clf_RF = pickle.load(open('res/pickle/diabetes_disease_classifier_RF.pkl', 'rb'))
+    # Apply models to make predictions
+    prediction_NB = load_clf_NB.predict(df)
+    prediction_proba_NB = load_clf_NB.predict_proba(df)
+    prediction_KNN = load_clf_KNN.predict(df)
+    prediction_proba_KNN = load_clf_KNN.predict_proba(df)
+    prediction_DT = load_clf_DT.predict(df)
+    prediction_proba_DT = load_clf_DT.predict_proba(df)
+    prediction_LR = load_clf_LR.predict(df)
+    prediction_proba_LR = load_clf_LR.predict_proba(df)
+    prediction_RF = load_clf_RF.predict(df)
+    prediction_proba_RF = load_clf_RF.predict_proba(df)
+
+    def NB():
+        st.subheader('Naive Bayes Prediction')
+        NB_prediction = np.array([0, 1])
+        if NB_prediction[prediction_NB] == 1:
+            st.write("<p style='font-size:20px;color: orange'><b>You have diabetes.</b></p>",
+                     unsafe_allow_html=True)
+        else:
+            st.write("<p style='font-size:20px;color: green'><b>You are fine 👍</b></p>", unsafe_allow_html=True)
+        enabled = st_toggle_switch("See detailed prediction")
+        if enabled:
+            st.subheader('Naive Bayes Prediction Probability')
+            st.write(prediction_proba_NB)
+            col1, col2 = st.columns(2)
+            with col1:
+                st.text('Why Classifier Report',
+                        help="It helps assess the model's ability to correctly identify classes and its overall performance in classifying data.")
+            with col2:
+                st.text('How to read',
+                        help="By looking at the cells where the true and predicted labels intersect, you can see the counts of correct and incorrect predictions. This helps evaluate the model's performance in distinguishing between 'No Disease' and 'Disease' categories.")
+
+            cmb.plt_NB()
+
+    def KNN():
+        st.subheader('K-Nearest Neighbour Prediction')
+        knn_prediction = np.array([0, 1])
+        if knn_prediction[prediction_KNN] == 1:
+            st.write("<p style='font-size:20px;color: orange'><b>You have diabetes.</b></p>",
+                     unsafe_allow_html=True)
+        else:
+            st.write("<p style='font-size:20px;color: green'><b>You are fine 👍</b></p>", unsafe_allow_html=True)
+        enabled = st_toggle_switch("See detailed prediction")
+        if enabled:
+            st.subheader('KNN Prediction Probability')
+            st.write(prediction_proba_KNN)
+            col1, col2 = st.columns(2)
+            with col1:
+                st.text('Why Classifier Report',
+                        help="It helps assess the model's ability to correctly identify classes and its overall performance in classifying data.")
+            with col2:
+                st.text('How to read',
+                        help="By looking at the cells where the true and predicted labels intersect, you can see the counts of correct and incorrect predictions. This helps evaluate the model's performance in distinguishing between 'No Disease' and 'Disease' categories.")
+
+            cmb.plt_KNN()
+
+    def DT():
+        st.subheader('Decision Tree Prediction')
+        DT_prediction = np.array([0, 1])
+        if DT_prediction[prediction_DT] == 1:
+            st.write("<p style='font-size:20px; color: orange'><b>You have diabetes.</b></p>",
+                     unsafe_allow_html=True)
+        else:
+            st.write("<p style='font-size:20px;color: green'><b>You are fine 👍</b></p>", unsafe_allow_html=True)
+        enabled = st_toggle_switch("See detailed prediction")
+        if enabled:
+            st.subheader('Decision Tree Prediction Probability')
+            st.write(prediction_proba_DT)
+            col1, col2 = st.columns(2)
+            with col1:
+                st.text('Why Classifier Report',
+                        help="It helps assess the model's ability to correctly identify classes and its overall performance in classifying data.")
+            with col2:
+                st.text('How to read',
+                        help="By looking at the cells where the true and predicted labels intersect, you can see the counts of correct and incorrect predictions. This helps evaluate the model's performance in distinguishing between 'No Disease' and 'Disease' categories.")
+
+            cmb.plt_DT()
+
+    def LR():
+        st.subheader('Logistic Regression Prediction')
+        LR_prediction = np.array([0, 1])
+        if LR_prediction[prediction_LR] == 1:
+            st.write("<p style='font-size:20px; color: orange'><b>You have diabetes.<b></p>",
+                     unsafe_allow_html=True)
+        else:
+            st.write("<p style='font-size:20px;color: green'><b>You are fine 👍</b></p>", unsafe_allow_html=True)
+        enabled = st_toggle_switch("See detailed prediction")
+        if enabled:
+            st.subheader('Logistic Regression Probability')
+            st.write(prediction_proba_LR)
+            col1, col2 = st.columns(2)
+            with col1:
+                st.text('Why Classifier Report',
+                        help="It helps assess the model's ability to correctly identify classes and its overall performance in classifying data.")
+            with col2:
+                st.text('How to read',
+                        help="By looking at the cells where the true and predicted labels intersect, you can see the counts of correct and incorrect predictions. This helps evaluate the model's performance in distinguishing between 'No Disease' and 'Disease' categories.")
+
+            cmb.plt_LR()
+
+    def RF():
+        st.subheader('Random Forest Prediction')
+        RF_prediction = np.array([0, 1])
+        if RF_prediction[prediction_RF] == 1:
+            st.write("<p style='font-size:20px; color: orange'><b>You have diabetes.</b></p>",
+                     unsafe_allow_html=True)
+        else:
+            st.write("<p style='font-size:20px;color: green'><b>You are fine 👍</b></p>", unsafe_allow_html=True)
+        enabled = st_toggle_switch("See detailed prediction")
+        if enabled:
+            st.subheader('Random Forest Probability')
+            st.write(prediction_proba_RF)
+            col1, col2 = st.columns(2)
+            with col1:
+                st.text('Why Classifier Report',
+                        help="It helps assess the model's ability to correctly identify classes and its overall performance in classifying data.")
+            with col2:
+                st.text('How to read',
+                        help="By looking at the cells where the true and predicted labels intersect, you can see the counts of correct and incorrect predictions. This helps evaluate the model's performance in distinguishing between 'No Disease' and 'Disease' categories.")
+            cmb.plt_RF()
+
+    def predict_best_algorithm():
+        if cmb.best_model == 'Naive Bayes':
+            NB()
+
+        elif cmb.best_model == 'K-Nearest Neighbors (KNN)':
+            KNN()
+
+        elif cmb.best_model == 'Decision Tree':
+            DT()
+
+        elif cmb.best_model == 'Logistic Regression':
+            LR()
+
+        elif cmb.best_model == 'Random Forest':
+            RF()
+        else:
+            st.write("<p style='font-size:20px;color: green'><b>You are fine 👍</b></p>", unsafe_allow_html=True)
+
+    st.markdown("👈 Provide your input data in the sidebar")    # Displays the user input features
+    with st.expander("Prediction Results",expanded=False):
+        # Display the input dataframe
+        st.write("Your input values are shown below:")
+        st.dataframe(input_df)
+        # Call the predict_best_algorithm() function
+        st.text('Here, The best algorithm is selected among all algorithm', help='It is based on classifier report')
+        predict_best_algorithm()
+        # Tips, Diagnosis, Treatment, and Recommendations.
+        st.subheader("👨‍⚕️ Expert Insights on Disease")
+        tab1, tab2, tab3 = st.tabs(["Tips", "Exercises", "Diet"])
+        with tab1:
+            st.subheader("Tips for Managing Diabetes:")
+            management_tips = [
+                "Cut sugar and refined carbohydrates from your diet.",
+                "Quit smoking.",
+                "Watch your portions.",
+                "Drink water.",
+                "Eat fiber."
+            ]
+            for tip in management_tips:
+                st.write(f"- {tip}")
+        with tab2:
+            st.subheader("Exercise Recommendations:")
+            c1, c2, c3 = st.columns([1, 1, 1], gap="small")
+            with c1:
+                exercise_recommendations = [
+                    "Running",
+                    "Cycling",
+                    "Yoga",
+                    "Water aerobics"
+                ]
+                for exercise in exercise_recommendations:
+                    st.write(f"- {exercise}")
+            with c3:
+                st_lottie(
+                    lottie_coding,
+                    speed=1,
+                    reverse=False,
+                    loop=True,
+                    quality="medium",
+                    height=None,
+                    width=None,
+                    key=None,
+                )
+        with tab3:
+            st.subheader("Dietary Recommendations:")
+            dietary_recommendations = [
+                "Beans",
+                "Dark leafy vegetables",
+                "Citrus fruit",
+                "Berries",
+                "Tomatoes",
+                "Fish high in omega-3 fatty acids",
+                "Nuts",
+                "Whole grains",
+                "Milk and yogurt"
+            ]
+            for food in dietary_recommendations:
+                st.write(f"- {food}")
+
+    # Create a multiselect for all the plot options
+    selected_plots = st.multiselect("You can see all plots here👇",
+                                    ["Naive Bayes", "K-Nearest Neighbors", "Decision Tree", "Logistic Regression",
+                                     "Random Forest"], default=[],key="ms_D")
+    if "ms_D" not in st.session_state:
+        st.session_state.selected_plots = []
+
+    # Check the selected plots and call the corresponding plot functions
+    if selected_plots:
+        col1, col2 = st.columns(2)
+        with col1:
+            st.text('Why Classifier Report',
+                    help="It helps assess the model's ability to correctly identify classes and its overall performance in classifying data.")
+        with col2:
+            st.text('How to read',
+                    help="By looking at the cells where the true and predicted labels intersect, you can see the counts of correct and incorrect predictions. This helps evaluate the model's performance in distinguishing between 'No Disease' and 'Disease' categories.")
+
+
+    placeholder = st.empty()
+
+    # Check the selected plots and call the corresponding plot functions
+    if "Naive Bayes" in selected_plots:
+        with st.spinner("Generating Naive Bayes...."):
+            cmb.plt_NB()
+            time.sleep(1)
+
+    if "K-Nearest Neighbors" in selected_plots:
+        with st.spinner("Generating KNN...."):
+            cmb.plt_KNN()
+            time.sleep(1)
+
+    if "Decision Tree" in selected_plots:
+        with st.spinner("Generating Decision Tree...."):
+            cmb.plt_DT()
+            time.sleep(1)
+
+    if "Logistic Regression" in selected_plots:
+        with st.spinner("Generating Logistic Regression...."):
+            cmb.plt_LR()
+            time.sleep(1)
+
+    if "Random Forest" in selected_plots:
+        with st.spinner("Generating Random Forest...."):
+            cmb.plt_RF()
+            time.sleep(1)
+
+    # Remove the placeholder to display the list options
+    placeholder.empty()