app.py

from flask import Flask, request, jsonify
import pickle
import gdown
import numpy as np

app = Flask(__name__)

# Load model
gdown.download_folder("https://drive.google.com/drive/folders/1BBA-8nXp9-JKOvNHFtMUg2o9CkQ2I2ft?usp=share_link", quiet=True, use_cookies=False)
filename = 'model_weights/analyzer_mlp_weights.sav'
loaded_model = pickle.load(open(filename, 'rb'))

@app.route('/analyzer', methods=['POST'])
def hello_world():
    # Getting parameters from json input body
    age = request.get_json()["age"]
    weight = request.get_json()["weight"]
    last_steps = request.get_json()["last_steps"]
    steps_goal = request.get_json()["steps_goal"]
    bcs_index = request.get_json()["bcs_index"]

    avg_steps = sum(last_steps)/len(last_steps)
    goal_weight = (100/(((bcs_index - 5) * 10) + 100)) * weight

    STEPS_SCORE = min(1, avg_steps/steps_goal)
    WEIGHT_SCORE = min(goal_weight, weight) / max(goal_weight, weight)

    # pounded formula
    # weight for diff from current weight to goal weight -> 0.8
    # weight for diff from current average steps per day to goal -> 0.2

    WEIGHT_FOR_WEIGHT_SCORE = 0.8
    WEIGHT_FOR_STEPS_SCORE = 0.2
    HEALTH_SCORE = WEIGHT_FOR_WEIGHT_SCORE * WEIGHT_SCORE + WEIGHT_FOR_STEPS_SCORE * STEPS_SCORE

    return jsonify({"health_score":HEALTH_SCORE})

@app.route('/analyzer/v2', methods=['POST'])
def hello_world_2():
    # Getting parameters from json input body
    age = request.get_json()["age"]
    weight = request.get_json()["weight"]
    last_steps = request.get_json()["last_steps"]
    steps_goal = request.get_json()["steps_goal"]
    bcs_index = request.get_json()["bcs_index"]

    # Get params for input
    meta_passos = steps_goal
    peso_ideal = (100/(((bcs_index - 5) * 10) + 100)) * weight
    indice_peso = 1 - ((np.abs(peso_ideal - weight) / peso_ideal))
    indice_passos = min(1, max(0, np.mean(last_steps)/meta_passos))
    
    pred = loaded_model.predict([[indice_peso, indice_passos]])[0]

    return jsonify({"health_score":pred})