Artificial neural network applied like qualifier of symptoms in patients with parkinson’s disease by evaluating the movement of upper-limbs activities

J. P. Bermeo, M. Huerta, M. Bravo, A. Bermeo

Research output: Contribution to conferencePaper

2 Scopus citations

Abstract

© Springer Nature Singapore Pte Ltd. 2019. The Movement Disorder Society (MDS-UPDRS) defines characteristics to qualify various symptoms of PD, the present works propose to apply an Artificial Neural Network ANN to qualify symptoms based upon movement of upper-limbs activities. In this way, a system based on Arduino and Android mobile app were developed, where accelerometers are used to acquire and store the acceleration data from upper-limbs while PD patients were doing three activities: rest sitting, eating and brushing teeth, meanwhile their symptoms were classified by doctor between 0 (normal) to 4 (most severe impairment). After that, store data were processed and estimation on Power Spectral Density (PSD) was done, then this information and doctor’s diagnosis were used into the ANN training to evaluate the symptoms in PD patients. For the ANN training was used back-propagation model and many ANN configurations, until get the best fit between inputs (processed data) and output (doctor’s diagnosis). The results showed that trained ANN can be used like qualifier with a high degree of accuracy over the 90%, for the tests performed. Moreover, even though MSD-UPDRS allows to get an accurate diagnosis, there is not objective, so ANN could be fixed to be completely objective, being a great advantage with manual evaluation.
Original languageEnglish
Pages409-414
Number of pages6
DOIs
StatePublished - 1 Jan 2018
EventIFMBE Proceedings - , Germany
Duration: 1 Jan 2007 → …

Conference

ConferenceIFMBE Proceedings
Country/TerritoryGermany
Period1/01/07 → …

Fingerprint

Dive into the research topics of 'Artificial neural network applied like qualifier of symptoms in patients with parkinson’s disease by evaluating the movement of upper-limbs activities'. Together they form a unique fingerprint.

Cite this