Tesis:
Automatic psychomotor function quantification in Parkinson’s disease via natural interaction with digital devices
- Autor: ARROYO-GALLEGO, Teresa
- Título: Automatic psychomotor function quantification in Parkinson’s disease via natural interaction with digital devices
- Fecha: 2020
- Materia: Sin materia definida
- Escuela: FACULTAD DE INFORMATICA
- Departamentos: AEROTECNIA
- Acceso electrónico: http://oa.upm.es/64517/
- Director/a 1º: LEDESMA CARBAYO, María Jesús
- Director/a 2º: GIANCARDO, Luca
- Resumen: This PhD thesis focuses on the study and development of new diagnostic techniques for evaluating psychomotor degradation in subjects affected by neurodegenerative diseases and movement disorders, specifically Parkinson’s disease. We propose a novel method based on leveraging user-device interaction as a source of information to detect and monitor psychomotor impairment associated with this type of pathology. The objective is to collaborate in the development of a new tool for the quantification of psychomotor damage that could be used in conjunction of current clinical standards. The proposed method for evaluating user-device interaction is based on capturing and analyzing the temporal information associated with each keystroke during a typing routine, whether on physical keyboards or touch screens. The main research tasks have focused on the implementation and evaluation of different algorithms for extracting biometric information related to the psychomotor function from typing activities that are part of users’ daily routine. The activities have been developed within the framework of a study on Parkinson’s disease as part of a scientific project of the M+Vision consortium ("neuroQWERTY"). In this context, the goal is to develop a tool that provides an objective evaluation of the psychomotor function, while being accessible and unobtrusive to the user. Based on the analysis of the timing patterns built from fingers interactions with mechanical and touchscreen keyboards, we developed a series of algorithms that introduce a collection of digital markers to generate computational outcomes measurements for the management of PD. Specifically, we have validated the feasibility of this approach to detect early signs of PD via mechanical and touchscreen typing. Also, we have developed an algorithm to identify response to medication through the analysis of longitudinal patterns of mechanical keyboard typing. These findings have a huge impact on the monitoring of the disease, since it brings us closer to an automatic quantification of the progress of the symptoms. Having detailed information on the evolution of the disease is crucial to progress towards increasingly personalized therapies, as well as improving our knowledge about the effect of these therapies and, consequently, promoting more efficient drug development.