app.py

import streamlit as st
import pickle
import pandas as pd
import numpy as np

# Page Title with Style
st.markdown(
    f"""
    <div style="text-align: center;">
        <h1 style="color: #800000;">🩸 Sepsis Prediction App</h1>
    </div>
    """,
    unsafe_allow_html=True
)

# Welcome Message with Style (Centered)
st.markdown(
    f"""
    <div style="text-align: center;">
        <p>👋 Welcome to the Sepsis Prediction App!</p>
    </div>
    """,
    unsafe_allow_html=True
)

# Sepsis Information
st.markdown(
    """
    **Sepsis** is a critical medical condition triggered by the body's extreme response to an infection. It can lead to organ failure and, if not detected early, poses a serious threat to life.
    """
)

# Link to WHO Fact Sheet on Sepsis
st.markdown("🔗 **Learn more about sepsis from [World Health Organization (WHO)](https://www.who.int/news-room/fact-sheets/detail/sepsis#:~:text=Overview,problems%20are%20at%20higher%20risk.)**")

st.markdown("---")

st.image("https://dinizululawgroup.com/wp-content/uploads/2020/07/news.jpg", width=700)

# Additional Information for Sample Prediction
st.write("📊 To make a sample prediction, you can refer to the training dataset information available in the sidebar.")
st.write("💉 Enter the medical data in the input fields below, then click 'Predict Sepsis', and get the patient's Sepsis status prediction.")


# About Section with Style
st.sidebar.title("ℹ️ About")
st.sidebar.info(
    "This app predicts sepsis onset using machine learning on medical data, aiding timely intervention by healthcare professionals. "
    "It utilizes a model trained on a sepsis dataset."
)

# Load The Train Dataset
train_df = pd.read_csv("Train.csv", index_col=None)

# Training Dataset Information in the sidebar
st.sidebar.markdown("📊 **Training Dataset Information:**")
st.sidebar.write("The model is trained on a sepsis dataset. Here's an overview of the dataset:")
st.sidebar.write(train_df.head())

# Auto-expand sidebar code
st.markdown(
    """
    <style>
        [data-testid="stSidebar"][aria-expanded="false"] > div:first-child {
            width: 100%;
        }
    </style>
    """,
    unsafe_allow_html=True
)

# Load the model and key components
with open('model_and_key_components.pkl', 'rb') as file:
    loaded_components = pickle.load(file)

loaded_model = loaded_components['model']
loaded_scaler = loaded_components['scaler']

# Data Fields
data_fields = {
    "**PRG**": "**Number of Pregnancies (applicable only to females)**\n   - The total number of pregnancies a female patient has experienced.",
    "**PL**": "**Plasma Glucose Concentration (mg/dL)**\n   - The concentration of glucose in the patient's blood). It provides insights into the patient's blood sugar levels.",
    "**PR**": "**Diastolic Blood Pressure (mm Hg)**\n   - The diastolic blood pressure, representing the pressure in the arteries when the heart is at rest between beats.",
    "**SK**": "**Triceps Skinfold Thickness (mm)**\n   - The thickness of the skinfold on the triceps, measured in millimeters (mm). This measurement is often used to assess body fat percentage.",
    "**TS**": "**2-hour Serum Insulin (mu U/ml)**\n   - The level of insulin in the patient's blood two hours after a meal, measured in micro international units per milliliter (mu U/ml).",
    "**M11**": "**Body Mass Index (BMI) (weight in kg / {(height in m)}^2)**\n   - BMI provides a standardized measure that helps assess the degree of body fat and categorizes individuals into different weight status categories, such as underweight, normal weight, overweight, and obesity.",
    "**BD2**": "**Diabetes pedigree function (mu U/ml)**\n   - The function provides information about the patient's family history of diabetes.",
    "**Age**": "**Age of the Patient (years)**\n   - Age is an essential factor in medical assessments and can influence various health outcomes."
}
# Organize input fields into two columns
col1, col2 = st.columns(2)

# Initialize input_data dictionary
input_data = {}

# Function to preprocess input data
def preprocess_input_data(input_data):
    numerical_cols = ['PRG', 'PL', 'PR', 'SK', 'TS', 'M11', 'BD2', 'Age']
    input_data_scaled = loaded_scaler.transform([list(input_data.values())])
    return pd.DataFrame(input_data_scaled, columns=numerical_cols)

# Function to make predictions
def make_predictions(input_data_scaled_df):
    y_pred = loaded_model.predict(input_data_scaled_df)
    sepsis_mapping = {0: 'Negative', 1: 'Positive'}
    return sepsis_mapping[y_pred[0]]

# Input Data Fields in two columns
with col1:
    input_data["PRG"] = st.slider("PRG: Number of Pregnancies", 0, 20, 0)
    input_data["PL"] = st.number_input("PL: Plasma Glucose Concentration (mg/dL)", value=0.0)
    input_data["PR"] = st.number_input("PR: Diastolic Blood Pressure (mm Hg)", value=0.0)
    input_data["SK"] = st.number_input("SK: Triceps Skinfold Thickness (mm)", value=0.0)

with col2:
    input_data["TS"] = st.number_input("TS: 2-Hour Serum Insulin (mu U/ml)", value=0.0)
    input_data["M11"] = st.number_input("M11: Body Mass Index (BMI)", value=0.0)
    input_data["BD2"] = st.number_input("BD2: Diabetes Pedigree Function (mu U/ml)", value=0.0)
    input_data["Age"] = st.slider("Age: Age of the patient (years)", 0, 100, 0)

# Predict Button with Style
if st.button("🔮 Predict Sepsis"):
    try:
        input_data_scaled_df = preprocess_input_data(input_data)
        sepsis_status = make_predictions(input_data_scaled_df)
        
        # Display the sepsis prediction and risk
        if sepsis_status == 'Negative':
            st.success(f"The patient is at **low risk** of developing sepsis. The predicted sepsis status is: **{sepsis_status}**")
        elif sepsis_status == 'Positive':
            st.error(f"**The patient is at **high risk** of developing sepsis. The predicted sepsis status is: **{sepsis_status}**")

        # Add the sepsis prediction to the input DataFrame
        input_data['Sepsis'] = sepsis_status

        # Convert the input data to a pandas DataFrame
        input_df = pd.DataFrame([input_data])

        # Display the input DataFrame
        st.markdown(
            f"""
            <div style="text-align: center; font-size: 18px; font-weight: bold;">
                <p>Input Data with Sepsis Prediction</p>
            </div>
            """,
            unsafe_allow_html=True
        )
        st.table(input_df)

    except Exception as e:
        st.error(f"An error occurred: {e}")

# Display Data Fields and Descriptions
st.sidebar.title("🔍 Data Fields")
for field, description in data_fields.items():
    st.sidebar.markdown(f"{field}: {description}")

# Copyright statement at the bottom
st.markdown(
    """
    <div style="text-align:center">
        Developed by <a href="https://www.linkedin.com/in/rasmo-/" style="font-style:italic">Rasmo Wanyama</a>.
    </div>
    """,
    unsafe_allow_html=True
)

st.stop()