whanhee

whanhee.py

import pandas as pd
import numpy as np
import json
from simplejson import loads


def get_outcomes():
    """ Get and return ICD codes """""
    f = open('icd_codes.json')
    outcomes_ = json.load(f)
    f.close()
    return loads(outcomes_[0])


def read_admissions():
    admissions_ = pd.read_csv("admissions_2000.csv")
    admissions_['ADATE'] = pd.to_datetime(admissions_['ADATE'], format='%d%b%Y')
    admissions_['DDATE'] = pd.to_datetime(admissions_['DDATE'], format='%d%b%Y')
    return admissions_


def primary(row, outcome=None):
    outcomes = get_outcomes()
    if row["DIAG1"] in outcomes[outcome]["icd9"] and \
            row["DDATE"] < icd_date:
        return 1
    if row["DIAG1"] in outcomes[outcome]["icd10"] and \
            row["DDATE"] >= icd_date:
        return 1
    return 0


def primary_secondary(row, outcome=None, secondary=False):
    """ Get primary and secondary or secondary from DIAG1-10 """
    outcomes = get_outcomes()
    start_number = 1
    if secondary:
         # start from DIAG2 if secondary
        start_number = 2
    diags = ["DIAG" + str(num) for num in np.arange(
        start_number, 11)]

    for diag in diags:
        if row[diag] in outcomes[outcome]["icd9"] and \
                row["DDATE"] < icd_date:
            return 1
        if row[diag] in outcomes[outcome]["icd10"] and \
                row["DDATE"] >= icd_date:
            return 1
    return 0


def simple(aki_primarysecondary):
    """ if aki_primarysecondary is first diag then return 0;
    if it's not, the outcome is first, hence return 1 """ 
    if aki_primarysecondary.iloc[0] == 1:
        return 0
    return 1


def get_first_hosp(df_):
    """ returns a list 1 followed by 0s """
    if len(df_) == 1:
        return 1
    if len(df_) == 2:
        return [1, 0]
    return [1]+[0 for i in range(len(df_)-1)]


outcomes = get_outcomes()
icd_date = pd.Timestamp(year=2015, month=10, day=1)


if __name__ == '__main__':
    # import json
    # print(json.dumps(outcomes, sort_keys=True, indent=4))

    admissions = read_admissions()

    for outcome in outcomes:
        admissions[outcome + "_primary"] = admissions.apply(
            primary, axis=1, outcome=outcome)
        admissions[outcome + "_primarysecondary"] = admissions.apply(
            primary_secondary, axis=1, outcome=outcome)
        admissions[outcome + "_secondary"] = admissions.apply(
            primary_secondary, axis=1, outcome=outcome, secondary=True)

    # ? drop all where outcome_primarysecondary == 0
    # ? drop diag cols

    # aki secondary co_morbidity primary
    co_morbidity = ["diabetes", "csd", "ihd", "pneumonia", "hf", "ami", "cerd", "uti"]
    for d in co_morbidity:
        admissions[d+'_primary_aki_secondary'] = admissions[
            [d+"_primary", "aki_secondary"]].min(axis=1)

    # diabetes or ckd as prior hosp diagnosis
    admissions['diabeteshosp_prior_aki'] = admissions[
        (admissions['aki_primarysecondary'] == 1) | (admissions['diabetes_primarysecondary'] == 1)].sort_values(
        by='ADATE').groupby('QID')['aki_primarysecondary'].transform(simple)
    admissions['ckdhosp_prior_aki'] = admissions[
        (admissions['aki_primarysecondary'] == 1) | (admissions['ckd_primarysecondary'] == 1)].sort_values(
        by='ADATE').groupby('QID')['aki_primarysecondary'].transform(simple)

    # correct primary and secondary so that only the first diag counts
    for outcome in outcomes:
        admissions[outcome + "_primary"] = \
            admissions[admissions[outcome + "_primary"] == 1].sort_values(
                by=['ADATE']).groupby('QID')['ADATE'].transform(get_first_hosp)
        admissions[outcome + "_secondary"] = \
            admissions[admissions[outcome + "_secondary"] == 1].sort_values(
                by=['ADATE']).groupby('QID')['ADATE'].transform(get_first_hosp)

    admissions.sort_values(by=['QID','ADATE']).to_csv("temp.csv", index=False)

New header:

QID,AGE,SEX,RACE,SSA_CNTY_CD,ADATE,DDATE,BENE_DOD,DIAG1,DIAG2,DIAG3,DIAG4,DIAG5,DIAG6,DIAG7,DIAG8,DIAG9,DIAG10,YEAR,all_kidney_primary,all_kidney_primarysecondary,all_kidney_secondary,ckd_primary,ckd_primarysecondary,ckd_secondary,aki_primary,aki_primarysecondary,aki_secondary,glomerular_primary,glomerular_primarysecondary,glomerular_secondary,diabetes_primary,diabetes_primarysecondary,diabetes_secondary,csd_primary,csd_primarysecondary,csd_secondary,ihd_primary,ihd_primarysecondary,ihd_secondary,pneumonia_primary,pneumonia_primarysecondary,pneumonia_secondary,hf_primary,hf_primarysecondary,hf_secondary,ami_primary,ami_primarysecondary,ami_secondary,cerd_primary,cerd_primarysecondary,cerd_secondary,uti_primary,uti_primarysecondary,uti_secondary,diabetes_primary_aki_secondary,csd_primary_aki_secondary,ihd_primary_aki_secondary,pneumonia_primary_aki_secondary,hf_primary_aki_secondary,ami_primary_aki_secondary,cerd_primary_aki_secondary,uti_primary_aki_secondary,diabeteshosp_prior_aki,ckdhosp_prior_aki