Integration of a Heart Rate Prediction Model into a Personal Health Record to Support the Telerehabilitation Training of Cardiopulmonary Patients - Biomedical Engineering Systems and Technologies - page 305

We also included enviro
For this purpose we procure
by a weather station in Han
chose temperature, humidi
added them to the training d
Only fully completed tr
course showing the charac
(see Fig. 1) were included
load plateau at a certain le
phase. The third phase (loa
the fourth phase (cool dow
also excluded sessions, wh
the training load, because
data. This could also be an
training load was too high o
ing schedule.
To automatically and ro
previously mentioned criter
over time to detect whether
Additionally, a session had
training sessions not fulfilli
of 1201 to 668 (325 F, 343
trainings per patient).

onmental variables, which could have a possible influen
ed data from the German weather service that was recor
nnover (station id: 2014), where the training took place.
ity and air pressure as main descriptors for weather
data.
raining sessions with a specific phase, and a typical l
cteristics of a successful four-phase rehabilitation train
into the dataset. The first phase (warm up) consists o
evel. This load increases stepwise over time in the seco
ad phase) shows a constant load for at least 10 minutes
wn) the load is reduced stepwise until it reaches null.
ere a monitoring physician interfered by de- or increas
the reasons of such changes were not documented in
indicator for training under suboptimal conditions, e.g.
or low for the patient due to an inadequately adapted tra

nce.
rded
We
and

load
ning
of a
ond
s. In
We
sing
our
the
ain-

obustly extract the sessions from the dataset matching
ria, we calculated the difference derivate of the load val
r the load was in- or decreasing during a training sessi
d to last from 12 to 26 minutes in total. After discarding
ing these criteria, we reduced the above mentioned num
3 M) training sessions from 115 patients (in mean 5.8±

the
lues
ion.
g all
mber
4.5

Fig. 2. Sample training sessio
RMSE is calculated for the fou

on for scenario four with measured and predicted heart rate.
ur training phases.

The