def _get_birth_weight_sum(self, pop: pd.DataFrame, base_filter: QueryString, configuration: Dict,
time_spans: List[Tuple[str, Tuple[pd.Timestamp, pd.Timestamp]]],
age_bins: pd.DataFrame) -> Dict[str, float]:
base_key = utilities.get_output_template(**configuration).substitute(measure='birth_weight_sum')
birth_weight_sum = {}
for year, (year_start, year_end) in time_spans:
year_filter = base_filter.format(start_time=year_start, end_time=year_end)
year_key = base_key.substitute(year=year)
group_birth_weight_sums = utilities.get_group_counts(pop, year_filter, year_key, configuration, age_bins,
lambda df: df[self.birth_weight_pipeline_name].sum())
birth_weight_sum.update(group_birth_weight_sums)
return birth_weight_sum
def _get_births(self, pop: pd.DataFrame, base_filter: QueryString, configuration: Dict,
time_spans: List[Tuple[str, Tuple[pd.Timestamp, pd.Timestamp]]], age_bins: pd.DataFrame,
cutoff_weight: float = None) -> Dict[str, float]:
if cutoff_weight:
base_filter += (
QueryString('{column} <= {cutoff}')
.format(column=f'`{self.birth_weight_pipeline_name}`', cutoff=cutoff_weight)
)
measure = 'low_weight_births'
else:
measure = 'total_births'
base_key = utilities.get_output_template(**configuration).substitute(measure=measure)
births = {}
for year, (year_start, year_end) in time_spans:
year_filter = base_filter.format(start_time=year_start, end_time=year_end)
year_key = base_key.substitute(year=year)
group_births = utilities.get_group_counts(pop, year_filter, year_key, configuration, age_bins)
births.update(group_births)
return births
def on_collect_metrics(self, event: 'Event'):
pop = self.population_view.get(event.index, query='alive == "alive"')
initial_proportion_reduction = pop[
'initial_treatment_proportion_reduction']
fpg = self.fpg(pop.index)
sbp = self.sbp(pop.index)
ldlc = self.ldlc(pop.index)
cvd_score = self.cvd_risk_score(pop.index)
measure_map = list(
zip([
'fpg_person_time', 'sbp_person_time', 'ldlc_person_time',
'cv_risk_score_person_time'
], [fpg, sbp, ldlc, cvd_score]))
adherent = self.is_adherent(pop.index).astype(int)
raw_ldlc = ldlc / (1 - initial_proportion_reduction)
at_target = (ldlc / raw_ldlc <= 0.5).astype(int)
# noinspection PyTypeChecker
step_size = to_years(event.step_size)
age_sex_filter, (ages, sexes) = get_age_sex_filter_and_iterables(
self.config, self.age_bins)
base_key = get_output_template(**self.config).substitute(
year=event.time.year)
base_filter = QueryString(f'alive == "alive"') + age_sex_filter
person_time = {}
for labels, pop_in_group in self.stratifier.group(pop):
for group, age_group in ages:
start, end = age_group.age_start, age_group.age_end
for sex in sexes:
filter_kwargs = {
'age_start': start,
'age_end': end,
'sex': sex,
'age_group': group
}
group_key = base_key.substitute(**filter_kwargs)
group_filter = base_filter.format(**filter_kwargs)
sub_pop = (pop_in_group.query(group_filter) if group_filter
and not pop_in_group.empty else pop_in_group)
for measure, attribute in measure_map:
person_time[group_key.substitute(
measure=measure)] = sum(
attribute.loc[sub_pop.index] * step_size)
adherent_pt = sum(adherent.loc[sub_pop.index] * step_size)
person_time[group_key.substitute(
measure='adherent_person_time')] = adherent_pt
at_target_pt = sum(at_target.loc[sub_pop.index] *
step_size)
person_time[group_key.substitute(
measure='at_target_person_time')] = at_target_pt
treatments = {
group_key.substitute(
measure=f'{treatment}_person_time'): 0.
for treatment in project_globals.TREATMENT
}
treatments.update(
(sub_pop[project_globals.TREATMENT.name].map(
lambda x: group_key.substitute(
measure=f'{x}_person_time')).value_counts() *
step_size).to_dict())
person_time.update(treatments)
self.results.update(
self.stratifier.update_labels(person_time, labels))