Multi-view representation learning via gcca for multimodal analysis of Parkinson's disease

Juan Camilo Vásquez-Correa, Juan Rafael Orozco-Arroyave, Raman Arora, Elmar Nöth, Najim Dehak, Heidi Christensen, Frank Rudzicz, Tobias Bocklet, Milos Cernak, Hamid R. Chinaei, Julius Hannink, Phani Sankar Nidadavolu, Maria Yancheva, Alyssa Vann, Nikolai Vogler

Abstract

Information from different bio-signals such as speech, handwriting, and gait have been used to monitor the state of Parkinson's disease (PD) patients, however, all the multimodal bio-signals may not always be available. We propose a method based on multi-view representation learning via generalized canonical correlation analysis (GCCA) for learning a representation of features extracted from handwriting and gait that can be used as a complement to speech-based features. Three different problems are addressed: classification of PD patients vs. healthy controls, prediction of the neurological state of PD patients according to the UPDRS score, and the prediction of a modified version of the Frenchay dysarthria assessment (m-FDA). According to the results, the proposed approach is suitable to improve the results in the addressed problems, specially in the prediction of the UPDRS, and m-FDA scores.
BibTeX
@inproceedings{vasquezcorrea2017multi,
  author = {Juan Camilo Vásquez-Correa and Juan Rafael Orozco-Arroyave and Raman Arora and Elmar Nöth and Najim Dehak and Heidi Christensen and Frank Rudzicz and Tobias Bocklet and Milos Cernak and Hamid R. Chinaei and Julius Hannink and Phani Sankar Nidadavolu and Maria Yancheva and Alyssa Vann and Nikolai Vogler},
  title = {{Multi-view representation learning via gcca for multimodal analysis of Parkinson's disease}},
  booktitle = {IEEE International Conference on Acoustics, Speech and Signal Processing},
  year = {2017},
  doi = {10.1109/icassp.2017.7952700},
}

Loading related content...

Figure