helpers.py

from functools import lru_cache

import numpy as np
from scipy.stats import binom_test

from constants import NETWORKS_FILE
from utils import load_from_json


@lru_cache(maxsize=1)
def load_networks(datapath, filename=NETWORKS_FILE):
    return load_from_json(datapath / filename)


def get_doctor_differential(li):
    return [val["concept"]["id"] for val in li if val["concept"]["id"] is not None]


def produce_differentials(card):
    return dict(
        [
            [val["user"]["id"], get_doctor_differential(val["doctor_diseases"])]
            for val in card["outcomes"]
            if "doctor_diseases" in val
        ]
    )


def doctor_top_ns(card, true_disease):
    differentials = produce_differentials(card)
    return [
        [key, len(val), 1 if true_disease in val else 0]
        for key, val in differentials.items()
    ]


def mean_list(li):
    if li == []:
        return "none"
    else:
        return np.mean(li)


def bintest(x, y, comf_thresh):
    if comf_thresh == 0:
        return sum(x) >= sum(y)
    return (
        binom_test(sum(x), n=len(x), p=sum(y) / len(y), alternative="greater")
        < comf_thresh
    )