def recode(age,
female,
ethnicity,
smoking,
sbp,
cvdHist,
bpld,
statin,
anticoag,
hba1c,
tchol,
hdl,
creat,
acr,
target="CHF"):
coefInfo = CHD_INFO
if target == "MI":
coefInfo = MI_INFO
if target == "STROKE":
coefInfo = STROKE_INFO
"""
Calculate the survival value
"""
xFeat = np.array([
clean_age(age), female, ethnicity, smoking,
clean_bp(sbp), cvdHist, bpld, statin, anticoag,
clean_hba1c(hba1c),
clean_tot_chol(tchol),
clean_hdl(hdl), creat,
clean_acr(acr)
])
return cox_surv(xFeat, coefInfo["coef"], coefInfo["s0"], coefInfo["const"])
def fremantle(age, male, cvd, hba1c, acr, hdl_mmol, seurope, aboriginal):
"""
Calculate the risk for cardiovascular disease
using the coefficients from the Fremantle Cohort
Parameters
----------
age : numeric
Age of subject
isMale : boolean or int
Subject is male (True or False)
cvd: boolean or int
Previous history of CVD (True or False)
hba1c : numeric
Hba1c (%) of subject
acr : numeric
Urinary albumin : creatinine ratio in mg/mmol
hdl : numeric
High density lipid cholestrol in mmol/L
seurope : bool or int
Subject is Southern European (True or False)
aboriginal : bool or int
Subject is Indigenous Australian (True or False)
"""
xFeat = np.array([
clean_age(age), male, cvd,
np.log(clean_hba1c(hba1c)),
np.log(clean_acr(acr)),
np.log(clean_hdl(hdl_mmol, meas="mmol")), seurope, aboriginal
])
return cox_surv(xFeat, FREMANTLE_COEF, FREMANTLE_SM, FREMANTLE_CONST)
def frs_simple(female, age, bmi, sbp, htn, smk, diab):
"""
10-year risk calculated using the Simple Non-Laboratory
Framingham Risk Score (FRS) Calculation.
Parameters
----------
female : boolean
age : numeric
Age of subject
bmi : numeric
BMI of subject
sbp : numeric
Systolic blood pressure of subject
ht_treat : bool or int
Treatment for hypertension (True or False)
smk : bool or int
Subject is smoker (True or False)
diab : bool or int
Subject has diabetes (True or False)
"""
xFeat = np.array([np.log(clean_age(age)),
np.log(clean_bmi(bmi)),
np.log(clean_bp(sbp))*(1-htn),
np.log(clean_bp(sbp))*htn,
smk,
diab])
genderInfo = NONLAB_MEN
if female:
genderInfo = NONLAB_WOMEN
return cox_surv(xFeat, genderInfo["coef"],
genderInfo["s0"], genderInfo["const"])
def hkdr_chd(age, female, cur_smoker, diab_dur, egfr, acr, nonhdl_mmol):
"""
Calculate the risk for coronary heart disease
using the coefficients from the HKDR CHD Cohort
Parameters
----------
age : numeric
Age of subject
isFemale : boolean or int
Subject is female (True or False)
curSmoke: boolean or int
Previous history of CVD (True or False)
diabDur : numeric
Nubmer of years of diabetes
egfr : numeric
Estimated Glomerular Filteration Rate
acr : numeric
Urinary albumin : creatinine ratio in mg/mmol
nonHDL : numeric
Non-HDL cholesterol (mmol/L)
"""
xFeat = np.array([
clean_age(age), female, cur_smoker,
clean_diab_dur(diab_dur),
np.log10(clean_egfr(egfr)),
np.log10(1 + clean_acr(acr)),
clean_nonhdl(nonhdl_mmol, meas="mmol")
])
return cox_surv(xFeat, HKDR_CHD["coef"], HKDR_CHD["sm"], HKDR_CHD["const"],
HKDR_CHD["shrink"])
def dmcx(age, egfr, tchdl, acr, smoker, diab_dur, female, sbp, dbp, hba1c,
htn_med, bmi, insulin, aGlucose):
"""
Calculate the risk for cardiovascular disease
using the coefficients from the DMCX Cohort
Parameters
----------
age : numeric
Age of subject
"""
# do some preprocessing
age = clean_age(age)
egfr = clean_egfr(egfr)
sbp = clean_bp(sbp)
dbp = clean_bp(dbp)
bmi = clean_bmi(bmi)
hba1c = clean_hba1c(hba1c)
tchdl = clean_tchdl(tchdl)
xFeat = np.array([
age, egfr >= 60 and egfr < 90, egfr >= 30 and egfr < 60, egfr < 30,
tchdl,
np.log(clean_acr(acr) + 1), smoker,
clean_diab_dur(diab_dur), sbp, hba1c, htn_med, dbp, bmi, insulin,
dbp**2, bmi**2, sbp**2, hba1c**2, age * tchdl, age * hba1c,
age * smoker, aGlucose
])
coefInfo = MALE_DCMX
if female:
coefInfo = FEMALE_DCMX
return cox_surv(xFeat, coefInfo["coef"], coefInfo["sm"], coefInfo["const"])
def hkdr_hf(female, age, bmi, hba1c, acr, hb, chdHist):
"""
Calculate the risk for heart failure
using the coefficients from the HKDR Cohort
Parameters
----------
age : numeric
Age of subject
bmi : numeric
BMI of the subject (in kg/m^2)
hba1c: numeric
HBA1C (%)
acr : numeric
Urinary albumin : creatinine ratio in mg/mmol
hb : numeric
Blood Hemoglobin (g/dl)
chdHist : boolean or int
Subject had CHD (true or False)
"""
baseSurv = HKDR_HF["male_sm"]
if female:
baseSurv = HKDR_HF["female_sm"]
xFeat = np.array([
clean_age(age),
clean_bmi(bmi),
clean_hba1c(hba1c),
np.log10(1 + clean_acr(acr)),
clean_hb(hb), chdHist
])
return cox_surv(xFeat, HKDR_HF["coef"], baseSurv, HKDR_HF["const"],
HKDR_HF["shrink"])
def hkdr_stroke(age, hba1c, acr, chd):
xFeat = np.array(
[clean_age(age),
clean_hba1c(hba1c),
np.log10(clean_acr(acr)), chd])
return cox_surv(xFeat, HKDR_STROKE["coef"], HKDR_STROKE["sm"],
HKDR_STROKE["const"])
def pce(female, ac, age, tot_chol, hdl, sbp, smoker, htn, diab, risk=5):
if risk not in [5, 10]:
raise NotImplementedError("Does not support risk that is not 5 or 10")
baseSurv = "s10"
if risk == 5:
baseSurv = "s5"
# figure out what the betas are
cohortInfo = WHITE_MALE
if female and ac:
cohortInfo = BLACK_FEMALE
elif female:
cohortInfo = WHITE_FEMALE
elif ac:
cohortInfo = BLACK_MALE
age = clean_age(age)
tot_chol = clean_tot_chol(tot_chol)
hdl = clean_hdl(hdl)
sbp = clean_bp(sbp)
xFeat = np.array([
np.log(age),
np.log(age)**2,
np.log(tot_chol),
np.log(tot_chol) * np.log(age),
np.log(hdl),
np.log(hdl) * np.log(age),
np.log(sbp) * (1 - htn),
np.log(age) * np.log(sbp) * (1 - htn),
np.log(sbp) * htn,
np.log(age) * np.log(sbp) * htn, smoker, smoker * np.log(age), diab
])
s = cox_surv(xFeat, cohortInfo["coef"], cohortInfo[baseSurv],
cohortInfo["const"])
return s
def dial(is_male,
age,
bmi,
cur_smoke,
sbp,
non_hdl,
hba1c,
egfr,
microalbumin,
macroalbumin,
diab_dur,
cvd_hist,
insulin,
hz_treat=0,
high_risk_county=False):
# fix the age to be within 34-94
age = np.clip(clean_age(age), 34, 94)
diab_dur = int(round(clean_diab_dur(diab_dur)))
bmi = clean_bmi(bmi)
sbp = clean_bp(sbp)
non_hdl = clean_nonhdl(non_hdl, meas="mmol")
egfr = clean_egfr(egfr)
hba1c = clean_hba1c(hba1c, meas="mmol")
xFeat = np.array([
is_male, age * is_male, bmi - 30, bmi**2 - 30**2, cur_smoke,
age * cur_smoke, sbp - 140, sbp**2 - 140**2, non_hdl - 3.8,
non_hdl**2 - 3.8**2, hba1c - 50, hba1c**2 - 50**2, egfr - 80,
egfr**2 - 80**2, microalbumin, macroalbumin, diab_dur, cvd_hist,
age * cvd_hist, insulin, age * insulin, hz_treat, high_risk_county
])
# look-up the age-specific value
s0 = AGE_S0[age]
return cox_surv(xFeat, DIAL_COEF, s0)
def qdiabetes(age,
male,
bmi,
diab_dur,
ac,
easian,
hba1c,
tchdl,
sbp,
heavy_smoke,
moderate_smoke,
light_smoke,
prev_smoke,
afib,
cvd,
renal,
tYear=5,
dmt1=False):
genderInfo = FEMALE_CCF
fractalFunc = _frac_poly_female
if male:
genderInfo = MALE_CCF
fractalFunc = _frac_poly_male
return _survival(clean_age(age), clean_bmi(bmi),
clean_diab_dur(diab_dur, 0), ac, easian,
clean_hba1c(hba1c, meas="mmol"), clean_tchdl(tchdl),
clean_bp(sbp), heavy_smoke, moderate_smoke, light_smoke,
prev_smoke, afib, cvd, renal, dmt1, genderInfo, tYear,
fractalFunc)
def ukpds(ageDiab, age, female, ac, smoking, hba1c, sbp, tchdl, tYear=10):
"""
Calculate the number of years to forecast the risk.
"""
xFeat = np.array([clean_age(age)-55,
female,
ac,
bool(smoking),
clean_hba1c(hba1c)-6.72,
(clean_bp(sbp) - 135.7)/10,
np.log(clean_tchdl(tchdl))-1.59])
q = Q_0 * np.prod(np.power(BETA, xFeat))
uscore = 1 - np.exp(-q * D**(age-ageDiab)* (1-D**tYear)/ (1 - D))
return max(uscore, 0.0)
def frs_primary(female, age, tot_chol, hdl, sbp, htn, smk, diab):
"""
"""
xFeat = np.array([np.log(clean_age(age)),
np.log(clean_tot_chol(tot_chol)),
np.log(clean_hdl(hdl)),
np.log(clean_bp(sbp))*(1-htn),
np.log(clean_bp(sbp))*htn,
smk,
diab])
genderInfo = LAB_MEN
if female:
genderInfo = LAB_WOMEN
return cox_surv(xFeat, genderInfo["coef"],
genderInfo["s0"], genderInfo["const"])
def aric(age, male, cauc, tc, hdl, sbp, htn, smoke):
# set the coeff based on male or female
genderInfo = FEMALE_INFO
if male:
genderInfo = MALE_INFO
age = clean_age(age)
tc = clean_tot_chol(tc)
hdl = clean_hdl(hdl)
xFeat = np.array([
age, age**2, cauc, tc >= 200 and tc <= 279, tc >= 280, hdl < 45,
hdl >= 45 and hdl <= 49,
clean_bp(sbp), htn, smoke
])
return cox_surv(xFeat, genderInfo["coef"], genderInfo["sm"],
genderInfo["xBetaMed"])
