![](bg6.png)
tion distinctly different from the one they were trained
on. The decrease in the performance was not substan-
tial, which may suggest that supervised fall detection
methods such as a CNN or SVM generalise well, or
that the population cohorts are not particularly differ-
ent.
ReN and 1NN personalised novelty detectors per-
form better than supervised methods applied across
population cohorts, but nevertheless more data per in-
dividual is needed to be able to evaluate whether this
could be true for 1SVM. It is known that some clas-
sifiers need a very large amount of training data to
achieve good performance (for example a CNN). The
performance ranking of the algorithms may change
when the algorithms are trained on a much bigger co-
hort.
We have not found sufficient evidence to prove
that novelty hybrid methods outperform novelty de-
tectors. Further experiments with varied amounts
of features extracted in the pre-training phase and a
larger amount of data are required.
Another interesting future avenue to explore
would be using domain adaptation as proposed in
(Ganin et al., 2016). The labelled fall and ADL data
from young participants could be used for training
alongside unlabelled data from target population, so
that the neural network could learn features that are
indiscriminative with respect to the shift between the
two population cohorts but discriminative between
falls and ADLs.
REFERENCES
Albert, M. V., Kording, K., Herrmann, M., and Jayaraman,
A. (2012). Fall classification by machine learning us-
ing mobile phones. PloS one, 7(5):e36556.
Bergstra, J., Breuleux, O., Bastien, F., Lamblin, P., Pascanu,
R., Desjardins, G., Turian, J., Warde-Farley, D., and
Bengio, Y. (2010). Theano: A cpu and gpu math com-
piler in python. In Proc. 9th Python in Science Conf,
pages 1–7.
Bourke, A., Obrien, J., and Lyons, G. (2007). Evaluation of
a threshold-based tri-axial accelerometer fall detection
algorithm. Gait & posture, 26(2):194–199.
Chen, J., Kwong, K., Chang, D., Luk, J., and Bajcsy, R.
(2006). Wearable sensors for reliable fall detection. In
Engineering in Medicine and Biology Society, 2005.
IEEE-EMBS 2005. 27th Annual International Confer-
ence of the, pages 3551–3554. IEEE.
Ganin, Y., Ustinova, E., Ajakan, H., Germain, P.,
Larochelle, H., Laviolette, F., Marchand, M., and
Lempitsky, V. (2016). Domain-adversarial training of
neural networks. Journal of Machine Learning Re-
search, 17(59):1–35.
Igual, R., Medrano, C., and Plaza, I. (2013). Challenges,
issues and trends in fall detection systems. Biomed.
Eng. Online, 12(66):1–66.
Kangas, M., Konttila, A., Lindgren, P., Winblad, I., and
J
¨
ams
¨
a, T. (2008). Comparison of low-complexity fall
detection algorithms for body attached accelerome-
ters. Gait & posture, 28(2):285–291.
Lee, R. Y. and Carlisle, A. J. (2011). Detection of falls us-
ing accelerometers and mobile phone technology. Age
and ageing, page afr050.
Lisowska, A., Wheeler, G., Ceballos Inza, V., and Poole,
I. (2015). An evaluation of supervised, novelty-based
and hybrid approaches to fall detection using silmee
accelerometer data. In Proceedings of the IEEE Inter-
national Conference on Computer Vision Workshops,
pages 10–16.
Lutze, R. and Waldh
¨
or, K. (2016). Smartwatch based
tumble recognitiona data mining model comparision
study. In e-Health Networking, Applications and
Services (Healthcom), 2016 IEEE 18th International
Conference on, pages 1–6. IEEE.
Medrano, C., Igual, R., Plaza, I., and Castro, M. (2014a).
Detecting falls as novelties in acceleration patterns ac-
quired with smartphones. PloS one, 9.
Medrano, C., Igual, R., Plaza, I., Castro, M., and Fardoun,
H. M. (2014b). Personalizable smartphone application
for detecting falls. In Biomedical and Health Infor-
matics (BHI), 2014 IEEE-EMBS International Con-
ference on, pages 169–172. IEEE.
Noury, N., Rumeau, P., Bourke, A.,
´
OLaighin, G., and
Lundy, J. (2008). A proposal for the classification and
evaluation of fall detectors. Irbm, 29(6):340–349.
Ojetola, O. (2013). Detection of Human Fal ls using Wear-
able Sensors. PhD thesis, Coventry University.
Pedregosa, F., Varoquaux, G., Gramfort, A., Michel, V.,
Thirion, B., Grisel, O., Blondel, M., Prettenhofer, P.,
Weiss, R., Dubourg, V., et al. (2011). Scikit-learn:
Machine learning in python. The Journal of Machine
Learning Research, 12:2825–2830.
Suzuki, T., Tanaka, H., Minami, S., Yamada, H., and Miy-
ata, T. (2013). Wearable wireless vital monitoring
technology for smart health care. In Medical Informa-
tion and Communication Technology (ISMICT), 2013
7th International Symposium on, pages 1–4. IEEE.
Zhang, T., Wang, J., Xu, L., and Liu, P. (2006). Fall de-
tection by wearable sensor and one-class svm algo-
rithm. In Intelligent Computing in Signal Processing
and Pattern Recognition, pages 858–863. Springer.
HEALTHINF 2018 - 11th International Conference on Health Informatics
82