Python FastAPI&；Streamlight-错误422，无响应返回

我正在尝试学习如何使用FastAPI创建API。为此，我决定使用streamlightpython包创建一个应用程序，并调用API来根据我使用TensorFlow准备的机器学习模型提供响应

问题是，当我通过“.post”调用API时，它会返回错误422，我已经测试了多种方法来解决这个问题，但最终导致API崩溃或无法工作

该API当前部署在Heroku（）上

以下是代码：

Streamlight应用程序（前端）：

# Loading packages ##########################################################################
import streamlit as st
import streamlit.components.v1 as components
import numpy as np
import pandas as pd
import requests,json



###############################################################################################
st.set_page_config(layout="wide")

# Loading of the data
@st.cache
def load_data():
    data=pd.read_csv('heart_failure_clinical_records_dataset.csv')
    return data

df = load_data()

# Sidebar creation ####################################################################
st.sidebar.title('Patient Parameters')

# Creating parameters
params = {
    'Age':st.sidebar.slider('Age', min(df['age']), max(df['age']),1.0),
    'Anemia':st.sidebar.selectbox('Anemia', ('Yes','No')),
    'CPK':st.sidebar.slider('CPK [mcg/L]', min(df['creatinine_phosphokinase']), max(df['creatinine_phosphokinase']), 1),
    'Diabetes':st.sidebar.selectbox('Diabetes', ('Yes','No')),
    'Ejection fraction':st.sidebar.slider('Ejection fraction [%]', min(df['ejection_fraction']), max(df['ejection_fraction']), 1),
    'High Blood Pressure':st.sidebar.selectbox('High Blood Pressure', ('Yes','No')),
    'Platelets in the blood':st.sidebar.slider('Platelets in the blood [kiloplatelets/mL]', min(df['platelets']), max(df['platelets']),1.0),
    'Serum Creatinine':st.sidebar.slider('Serum Creatinine [mg/dL]', min(df['serum_creatinine']), max(df['serum_creatinine']),1.0),
    'Serum Sodium':st.sidebar.slider('Serum Sodium [mEq/L]', min(df['serum_sodium']), max(df['serum_sodium']),1),
    'Woman or Man':st.sidebar.selectbox('Woman or Man', ('Woman','Men')),
    'Smoking':st.sidebar.selectbox('Smoking',('Yes','No')),
    'Time':st.sidebar.slider('Time [days]', min(df['time']), max(df['time']),1)

}

# Method for changing attribute
def change_type(X):
    if X == 'Yes':
        X = 1
    else:
        X = 0    
    return X

def change_sex(X):
    if X == 'Woman':
        X = 1
    else:
        X = 0
    return X            

# Defining user input
def user_input():
    age = params['Age']
    anemia = params['Anemia']
    cpk = params['CPK']
    diabetes = params['Diabetes']
    ejection_fraction = params['Ejection fraction']
    hbp = params['High Blood Pressure']
    platelets = params['Platelets in the blood']
    serum_creatinine = params['Serum Creatinine']
    serum_sodium = params['Serum Sodium']
    woman_man = params['Woman or Man']
    smoking = params['Smoking']
    time = params['Time']

    data = {
        'Age': age,
        'Anemia': change_type(anemia),
        'CPK': cpk,
        'Diabetes':change_type(diabetes),
        'Ejection_fraction':ejection_fraction,
        'HBP':change_type(hbp),
        'Platelets':platelets,
        'Serum_Creatinine':serum_creatinine,
        'Serum_Sodium':serum_sodium,
        'Woman_Man':change_sex(woman_man),
        'Smoking':change_type(smoking),
        'Time':time
    }

    # features = pd.DataFrame(data, index=[0])

    return data

# API information
url = 'https://heart-api-first.herokuapp.com'
endpoint = '/prediction'


# defining process for calling API
def process(data, server_url: str):
    heart_data = data

    result = requests.post(server_url,data = heart_data)

    return result


# Expander ##################################################################################
with st.beta_expander('Information', expanded=True):
    st.header('Goal of the Web Application')
    st.write('The main goal of this project to learn on how to use the fastAPI package for creating background services, in this case, a machine learning model.')
    st.write('**Source of data**: https://www.kaggle.com/andrewmvd/heart-failure-clinical-data')
    st.write("""
    > **Attention**: 
    * This app is **NOT** meant as a medical device/applicaiton for medical diagnosis; 
    * It is just for educational purpose;
    * If you think you have an health problem, please reach out to an authorized medical professional;
    * The creator of this applicaiton will **NOT** assume any responsability for any misuse of the application.
    """)
    st.header('Parameters Information')
    st.write("""
    * **Age**: Age of the patient;
    * **Anemia**: Decrease of red blood cells or hemoglobin;
    * **Creatinine Phosphokinase Enzyme (CPK)**: Level of CPK enzyme in the blood (mcg/L);
    * **Diabetes**: Patient is diabetic or not;
    * **Ejection Fraction**: Percentage of blood that leaves the heart at each contraction;
    * **High Blood Pressure**: Patient has high blood pressure;
    * **Platelets in the blood**: Number of platelets in the blood [kiloplatelets/mL];
    * **Serum Creatinine**: Level of serum creatinine in the blood [mg/dL];
    * **Serum Sodium**: Level of serum sodium in the blood [mEq/L];
    * **Woman or Man**: If patient is a woman or man;
    * **Smoking**: If patient smokes or not;
    * **Time**: Follow-up periods, number of days after the first check.
    """)

    

# Columns creation #####################################################################
left_column1, right_column1 = st.beta_columns(2)

with left_column1:
    st.header('Heart Animation')
    st.write('In this GIF, you can see how the heart works, with the name of its major components.')
    components.iframe("https://healthblog.uofmhealth.org/sites/consumer/files/2020-01/heart_pumping.gif",height=500)
with right_column1:
    st.header('Heart Attack Video')
    st.write('In this video, created by the YouTube channell _Institute of Human Anatomy_, explain what happens during a heart attack.')
    st.write('**Attention**: Video contains content (real human heart from a dead body) that may cause some problem for some people.')
    st.video("https://www.youtube.com/watch?v=m0fiHGlqivo&t=1s")


# Creating the data ####################################################################

user_input_original_df = user_input()
user_input_original_df

# Creating button



test_df = pd.DataFrame(user_input_original_df).dtypes
test_df




# if st.sidebar.button("Predict"):
#     result = process(user_input_original_df, url+endpoint)
#     result
# else:
#   pass

API:

---

如果您对如何解决此问题有任何建议，我将不胜感激。

错误消息

422

应该（在其正文中）指出您预期输入（心脏

模型）的哪些字段或功能缺失或未正确验证。
import uvicorn
from fastapi import FastAPI


# Download basic ML model for creating class
from pydantic import BaseModel

# Code for creating the model of the API
import numpy as np
import pandas as pd

from sklearn.preprocessing import StandardScaler
from sklearn.preprocessing import OneHotEncoder

# Neural Netwrok libraries
import tensorflow as tf
from tensorflow import keras
#from tensorflow.keras import Sequential
#from tensorflow.keras.layers import Dense


# Creating class for validation
class Heart(BaseModel):
    Age: float
    Anemia: int
    CPK: int
    Diabetes: int
    Ejection_fraction: int
    HBP: int
    Platelets: float
    Serum_Creatinine: float
    Serum_Sodium: int
    Woman_Man: int
    Time: int

# Creating std scaler
sc = StandardScaler()

# Loading the model saved

model = keras.models.load_model("Test_model_Neural_Network")

# List for prediction
pred_label = list()

# Defining data preparation
def preparazione(dati_originali):
    dati_pronti = dati_originali.values
    dati_pronti = sc.fit_transform(dati_originali)
    return dati_pronti

# Defining function for prediction
def predict(dati):
    previsione = model.predict(dati)
    return previsione

# Defining method for converting results
def convert(dati_preparati, lista):
    for i in range(len(dati_preparati)):
        result = pred_label.append(np.argmax(dati_preparati[i]))

    return result

app = FastAPI()

@app.get("/")
def home():
    return {"message":"Welcome to my first API!"}

@app.post("/prediction")
async def get_prediction(heart :Heart):
    input_df = pd.DataFrame([heart.dict()])
    dati_preparati = preparazione(input_df)
    prediction_heart = predict(dati_preparati)
    prediction_label = convert(prediction_heart, pred_label)
    return prediction_label

if __name__ == '__main__':
    uvicorn.run(app, host='127.0.0.1', port=8000)