Human - smart rollator interaction for gait analysis and fall prevention using learning methods and the i-walker
- Autores: Atia Cortés Martínez
- Directores de la Tesis: Antonio Benito Martínez Velasco (dir. tes.), Javier Bejar Alonso (dir. tes.)
- Lectura: En la Universitat Politècnica de Catalunya (UPC) ( España ) en 2018
- Idioma: español
- Tribunal Calificador de la Tesis: Alicia Casals (presid.), Aïda Valls Mateu (secret.), Luiza Spiru (voc.)
- Programa de doctorado: Programa de Doctorado en Inteligencia Artificial por la Universidad Politécnica de Catalunya
- Materias:
- Enlaces
- Tesis en acceso abierto en: TDX
- Resumen
The ability to walk is typically related to several bio-mechanical components that are involved in the gait cycle (or stride), including free mobility of joints, particularly in the legs; coordination of muscle activity in terms of timing and intensity; and normal sensory input, such as vision and vestibular system. A walk is composed of the stance and swing phases. The faster we walk, the shorter the stance phase will be. Thus, gait requires input from the brain, spinal cord, peripheral nerves, muscular power and joint and cardiovascular health. Because all of these systems are required to coordinate gait, gait speed is an indicator of the health of many physiological systems. At the same time, a relation between gait and cognition has been widely analysed from the medical point of view, and we can find several reviews in the literature. As people age, they tend to slow their gait speed, and their balance is also affected. Also, the retirement from the working life and the consequent reduction of physical and social activity contribute to the increased incidence of falls in older adults. Moreover, older adults suffer different kinds of cognitive decline, such as dementia or attention problems, which also accentuate gait disorders and its consequences.
Assistive technologies (AT) play a key role in today's' society, especially when it comes to the older adults. ATs have enabled improvements in their Quality of Life, extending their autonomy and community living. This is important, as they can stay active safely and independently. During the last decade, research has focused on developing ATs with a sensor system integrated with the device or located in the human body. Efforts are focused especially on mobility assistance for different targets of people (visual impairment, frailty, chronic diseases or rehabilitation) and activity recognition, which could be used, for instance, to monitor elderly population living in autonomy and community-dwelling.
This thesis proposes a methodology to analyse how do older adults at high risk of falling interact with a smart rollator, the i-Walker, to navigate in indoor, flat environments. The i-Walker is equipped with a set of sensors and actuators and can collect data for long periods of time (several hours). It has already been tested in post-stroke rehabilitation and fall prevention clinical trials with successful results. In this work, we present results on our approach from a narrative perspective. Results are promising since we can relate the data obtained from human-rollator interaction to clinical parameters. The machine learning approach uses the data obtained with the force sensors of the i-Walker based on the interaction of individuals of different ages and health conditions. The analysis complements our extracted gait parameters with biological and clinical data to learn new characteristics of human gait at a stride-to-stride level. We believe that users, caregivers and clinicians would benefit from the new knowledge that the i-Walker can generate from this work.
