app.py

import streamlit as st 
import torch
import torch.nn.functional as f
import torch.nn as nn
import joblib 

html_temp = """
    <div style="background-color:black;padding:10px">
    <h2 style="color:white;text-align:center;"> Fetal Health App </h2>
    </div>
    """
st.markdown(html_temp, unsafe_allow_html=True)
st.markdown(f"""
    <style>
        /* Set the background image for the entire app */
        .stApp {{
            background-color:#a0aecd;
            background-size: 100px;
            background-repeat:no;
            background-attachment: auto;
            background-position:full;
        }}
    </style>
    """, unsafe_allow_html=True)
## model development 
class ClassificationModel(nn.Module):
    def __init__(self, input_dim):
        super(ClassificationModel, self).__init__()
        self.fc1 = nn.Linear(input_dim, 16)
        self.relu = nn.ReLU()
        self.dropout = nn.Dropout(0.2)
        self.fc2 = nn.Linear(16,8)
        self.fc3 = nn.Linear(8, 4)
        self.fc4 = nn.Linear(4, 2) # 2 outputs for binary classification

    def forward(self, x):
        x = f.relu(self.fc1(x))
        x = f.relu(self.fc2(x))
        x = f.relu(self.fc3(x))
        x = self.fc4(x)  # No softmax, since CrossEntropyLoss applies it
        return x
    
torch.manual_seed(42)    
model=ClassificationModel(input_dim=11)
##  loading the models

scaler_x=joblib.load('x_sacaler_model.joblib')

# Load the entire model

model.load_state_dict(torch.load( "fetal_health_model.pth", map_location=torch.device("cpu")))
model.eval()  

## inputs 
col1,col2 =st.columns(2)

with col1:
    #  Age
    age = st.slider("Slide the age of Patient in date wise",min_value=10798,max_value=23713,value=1500)

with col2:
    # gender
    gender_option =["MALE","Female"] 
    gender=st.selectbox("Select the Gender of the Patient ",gender_option)
    gender_value = gender_option.index(gender)
with col1:
    # height
    height = st.slider("Slide Height of the Patient in cm's ",min_value=55,max_value=250,value=75)
with col2:
    #  Weight
    weight = st.slider("Slide the Weight of the Patient in Kg's ",min_value=10.0,max_value=200.0,value=50.0)
col1,col2=st.columns(2)
with col1:
    #  ap_hi
    ap_hi =st.slider("Slide the ap high of the Patient ",min_value=-150,max_value=16020,value=15)
with col2:
    # ap_lo
    ap_lo=st.slider("Slide the ap Low of the Patient ",min_value=-70,max_value=11000,value=200)
col1,col2=st.columns(2)
with col1:   
    # cholesterol
     chol_option =["Normal","Above Normal","Well"] 
     chol=st.selectbox("Select the condition of the Patient",chol_option)
     chol_value=chol_option.index(chol)
    
with col2:
     glue_option =["Normal","Above Normal","Well"] 
     glue=st.selectbox("Select the Glue of the Patient",glue_option)
     glue_value=glue_option.index(glue)
col1,col2=st.columns(2)
with col1:  
     smoke_option =["NO","Yes"] 
     smoke=st.selectbox("Are YOU smoker ",smoke_option)
     smoke_value=smoke_option.index(smoke)
with col2:
     aclo_option =["NO","Yes"] 
     aclo=st.selectbox(" Are you Alcholic ",aclo_option)
     aclo_value=aclo_option.index(aclo)
    
active_option =["NO","Yes"] 
active=st.selectbox("Are you active person ",active_option)
active_value=active_option.index(active)

if st.button("Enter"):
    # st.write(age,gender_value,height,weight,ap_hi,ap_lo,chol_value,
                                # glue_value,smoke_value,aclo_value,active_value)
    
    values=scaler_x.transform([[age,gender_value,height,weight,ap_hi,ap_lo,chol_value,
                                glue_value,smoke_value,aclo_value,active_value]])
    
    # st.write(values)
    values_tensor=torch.tensor(values,dtype=torch.float32)

    with torch.no_grad():  # Disable gradient calculation (faster inference)
        output = model(values_tensor)
    # st.write(output.argmax(dim=1))
    out_value=output.argmax(dim=1)
    if out_value ==0:
        st.write("The Patient is Healthy")
    else :
        st.write("The Patient is UN Healthy")