def process_physiology(self):
_d1 = map_devices(self.datasets['vital_signs'],
self.datasets['device_types'])
_d2 = map_devices(self.datasets['blood_pressure'],
self.datasets['device_types'])
bp_types = np.array(['diastolic_value', 'systolic_value'])
df = [_d1]
for bp_type in bp_types:
cols = _d2.columns[~_d2.columns.str.startswith(bp_type)]
_tmp = _d2[cols].copy()
_tmp.source = bp_types[bp_types != bp_type][0].replace('_value', '_bp')
_tmp.columns = _d1.columns
df.append(_tmp)
df = pd.concat(df)
dtypes = {
'device_type': 'category',
'patient_id': 'category',
'home_id': 'category',
'unit': 'category',
'source': 'category',
'value': 'float'
}
df = df.astype(dtypes).reset_index(drop=True)
df = localize_time(df, ['start_date'])
return df
def process_kitchen(self):
# TODO: add sensor type as a column
activity = self.datasets['activity']
activity = localize_time(activity, ['start_date'])
doors = self.datasets['doors']
appliances = map_devices(self.datasets['appliances'],
self.datasets['device_types'])
doors = doors[(doors.location_name == 'fridge door')
& (doors.transition == 'opened>closed')]
activity = activity[activity.location_name.isin(['kitchen',
'dining room'])]
activity.insert(0, 'device_type', 'pir')
doors.insert(0, 'device_type', 'door')
appliances.value = appliances.value.astype(str)\
.replace({'microwave-use': 'oven-use',
'toaster-use': 'oven-use', 'iron-use': np.nan,
'multi-use socket-use': np.nan})
doors = doors[['patient_id', 'start_date', 'device_type','location_name']]\
.rename({'location_name': 'activity'}, axis=1)
activity = activity[['patient_id', 'start_date', 'device_type', 'location_name']]\
.rename({'location_name': 'activity'}, axis=1)
appliances = appliances[['patient_id', 'start_date', 'device_type', 'value']]\
.rename({'value': 'activity'}, axis=1)
kitchen = pd.concat([doors, activity, appliances.dropna()])
kitchen = kitchen[kitchen.patient_id != ''].sort_values(
['patient_id', 'start_date'])
kitchen = process_transition(kitchen, ['patient_id'], 'start_date',
'activity', ['device_type'])
return kitchen.reset_index(drop=True)
def process_motion(self):
activity = self.datasets['activity']
activity = localize_time(activity, ['start_date'])
entryway = self.datasets['entryway']
bed_occupancy = self.datasets['bed_occupancy']
fact = ['patient_id', 'location_name', 'start_date']
motion = pd.concat([activity[fact], entryway[fact], bed_occupancy[fact]]).\
sort_values(['patient_id', 'start_date'])
mapping = {
'bathroom1': 'Bathroom',
'WC1': 'Bathroom',
'kitchen': 'Kitchen',
'hallway': 'Hallway',
'corridor1': 'Hallway',
'dining room': 'Lounge',
'living room': 'Lounge',
'lounge': 'Lounge',
'bedroom1': 'Bedroom',
'front door': 'Front door',
'back door': 'Back door'
}
motion.location_name = motion.location_name.replace(mapping)
motion = motion[~motion.location_name.
isin(['office', 'conservatory', 'study', 'cellar'])]
return motion.reset_index(drop=True)
def process_physiology(self):
df = self.datasets['observation']
device_type = self.datasets['device_type']
devices = [
'8310-5', '150456', '55284-4', '29463-7', '251837008', '163636005',
'8462-4', '8480-6', '8867-4'
]
df = df.query('type in @devices')
mapping = device_type.set_index('code').display.to_dict()
mapping = {**mapping, **{'8867': 'raw_heart_rate'}}
columns = [
'datetimeObserved', 'type', 'subject', 'valueQuantity', 'valueUnit'
]
df = df[columns].assign(source=df.type.replace(mapping))
mapper = {
'Body temperature': 'raw_body_temperature',
'Total body water': 'raw_total_body_water',
'Systolic blood pressure': 'systolic_bp',
'Diastolic blood pressure': 'diastolic_bp',
'MDC_PULS_OXIM_SAT_O2': 'raw_oxygen_saturation',
'O/E - muscle mass': 'raw_body_muscle_mass',
'Body weight': 'raw_body_weight'
}
df.source = df.source.replace(mapper)
df.columns = [
'start_date', 'device_type', 'patient_id', 'value', 'unit', 'source'
]
df = localize_time(df, ['start_date'])
return df
def process_light(self):
df = self.datasets['observation']
columns = ['datetimeObserved','type','subject','location','valueQuantity','valueUnit']
df = df.query('type == "56242006"')[columns]
df.columns = ['start_date', 'device_type', 'patient_id','location_name','value', 'unit']
df = df.assign(source='raw_light')
df = localize_time(df,['start_date'])
return df
def process_doors(self):
groupby = ['patient_id', 'location_name']
datetime = 'start_date'
value = 'value'
covariates = ['location_name']
df = self.datasets['door']
df = localize_time(df, ['start_date'])
doors = process_transition(df, groupby, datetime, value, covariates)
return doors.drop('location_name', axis=1).reset_index()
def process_motion(obj):
df = obj.datasets['observation']
df = (df[['subject', 'location', 'datetimeObserved',
'type']].query('type == "272149007-2"').dropna(
subset=['location']).drop(columns='type'))
df.columns = ['patient_id', 'location_name', 'start_date']
df = localize_time(df, ['start_date'])
df = df.drop_duplicates().reset_index()
return df
def process_habitat(self):
df = map_devices(self.datasets['environmental'],
self.datasets['device_types'])
df = df[df.location_name != ''].reset_index(drop=True)
dtypes = {
'home_id': 'category',
'location_id': 'category',
'unit': 'category',
'location_name': 'category',
'source': 'category',
'value': 'float'
}
df = df.astype(dtypes)
df = localize_time(df, ['start_date'])
return df
def process_doors(self):
df = self.datasets['observation']
columns = ['subject', 'location', 'datetimeObserved', 'valueState', 'type']
df = (df[columns].query('type == "224751004"').dropna(
subset=['location']).drop(columns='type'))
df.columns = ['patient_id', 'location_name', 'start_date', 'value']
df.value = df.value.replace({'Close': 'closed', 'Open': 'opened'})
groupby = ['patient_id', 'location_name']
datetime = 'start_date'
value = 'value'
covariates = None
df = localize_time(df, ['start_date'])
doors = process_transition(df, groupby, datetime, value, covariates)
doors = doors.reset_index()
doors.location_name = doors.location_name.str.lower()
return doors
def process_sleep(self):
"""process_sleep force sleep_mat to 1min frequency and localise time
:return: loclaized sleep metrics timeseries
:rtype: pd.DataFrame
"""
sleep_mat = self.datasets['sleep_mat']
sleep_mat.snoring = sleep_mat.snoring.astype(bool)
sleep_mat = (sleep_mat.set_index('start_date').groupby(
['patient_id', 'home_id']).resample('1T').agg({
'snoring':
'sum',
'heart_rate':
'mean',
'respiratory_rate':
'mean'
}).dropna().reset_index())
sleep_mat = localize_time(sleep_mat, ['start_date'])
return sleep_mat
def process_sleep(self):
sleep_mat = self.datasets['sleep_mat']
sleep_mat = localize_time(sleep_mat, ['start_date'])
return sleep_mat