import gradio as gr
import xgboost
import pandas as pd
import numpy as np


def predicter(SpO2, Age, Weight, Height, Temperature, Gender, Race):
    xgb_reg = xgboost.XGBClassifier(tree_method = 'approx',
                                enable_categorical = True,
                                learning_rate=.1,
                                   max_depth=2,
                                   n_estimators=70,
                              early_stopping_rounds = 0,
                               scale_pos_weight=1)

    xgb_reg.load_model('classifier_fewer_features_HH.json')

    if Gender == "Male":
        gen = "M"
    elif Gender == "Female":
        gen = "F"

    cont_features = ['SpO2','anchor_age','weight','height','temperature']
    cat_features = ['gender','race_group']
    
    user_input = pd.DataFrame([[SpO2/100,Age/91,Weight/309,Height/213,Temperature/42.06,gen,Race]],columns = cont_features+cat_features)
    user_input[cat_features] = user_input[cat_features].copy().astype('category')
    pred = xgb_reg.predict_proba(user_input)

    #return str(pred)
    return {"No Hidden Hypoxemia": float(pred[0][0]), "Hidden Hypoxemia": float(pred[0][1])}
    

demo = gr.Interface(
    fn=predicter,
    inputs=[gr.Slider(88, 100),"number",gr.inputs.Number(label = "Weight in kg"),gr.inputs.Number(label = "Height in cm"),gr.inputs.Number(label = "Temperature in Celcius"),gr.Radio(["Male", "Female"]),gr.Radio(["White", "Black", "Asian", "Hispanic", "Other"])],
    outputs=[gr.Label(label = "Probabilities")],
    title = "Model Predictions"
)
demo.launch()