TY - JOUR
T1 - Predicting urinary bladder voiding by means of a linear discriminant analysis
T2 - Validation in rats
AU - Tantin, A.
AU - Bou Assi, E.
AU - van Asselt, E.
AU - Hached, S.
AU - Sawan, M.
N1 - Funding Information:
Authors acknowledge the financial support from NSERC of Canada .
Publisher Copyright:
© 2019 Elsevier Ltd
PY - 2020/1
Y1 - 2020/1
N2 - Aims: The objective of this work is to investigate whether changes in bladder pressure's patterns can be used to forecast voiding events in rats with both normal and overactive detrusor. Methods: A voiding forecasting algorithm based on machine learning was developed. Raw pressure curves as well as their corresponding power bands were used as inputs to a linear discriminant analysis classifier. Performance was evaluated on held-out test data and was statistically validated via comparison to random predictors. Results: Using the band-power feature, 93% and 99% of the alarms were respectively generated within 95 s before voiding for normal and hyperactive bladder conditions respectively. The same algorithm was assessed using the band-power feature. It showed performances achieving respective success rates of 99% and 97% for normal and hyperactive bladder condition respectively with alarms generated within 45 s before voiding. Conclusions: We have demonstrated the feasibility of detecting the pre-voiding periods in rats with normal and overactive bladders with a high success rate. Significance: To our knowledge, this is the first study that demonstrates the possibility of predicting voiding in rats with a machine learning algorithm based on a Linear Discriminant Analysis. Our work was compared to other relevant studies and showed better results. With this study, accurate urinary bladder voiding forecasting could be implemented in closed-loop advisory/intervention devices.
AB - Aims: The objective of this work is to investigate whether changes in bladder pressure's patterns can be used to forecast voiding events in rats with both normal and overactive detrusor. Methods: A voiding forecasting algorithm based on machine learning was developed. Raw pressure curves as well as their corresponding power bands were used as inputs to a linear discriminant analysis classifier. Performance was evaluated on held-out test data and was statistically validated via comparison to random predictors. Results: Using the band-power feature, 93% and 99% of the alarms were respectively generated within 95 s before voiding for normal and hyperactive bladder conditions respectively. The same algorithm was assessed using the band-power feature. It showed performances achieving respective success rates of 99% and 97% for normal and hyperactive bladder condition respectively with alarms generated within 45 s before voiding. Conclusions: We have demonstrated the feasibility of detecting the pre-voiding periods in rats with normal and overactive bladders with a high success rate. Significance: To our knowledge, this is the first study that demonstrates the possibility of predicting voiding in rats with a machine learning algorithm based on a Linear Discriminant Analysis. Our work was compared to other relevant studies and showed better results. With this study, accurate urinary bladder voiding forecasting could be implemented in closed-loop advisory/intervention devices.
UR - http://www.scopus.com/inward/record.url?scp=85071602797&partnerID=8YFLogxK
U2 - 10.1016/j.bspc.2019.101667
DO - 10.1016/j.bspc.2019.101667
M3 - Article
AN - SCOPUS:85071602797
VL - 55
JO - Biomedical Signal Processing and Control
JF - Biomedical Signal Processing and Control
SN - 1746-8094
M1 - 101667
ER -