Intelligent systems for pediatric rehabilitation

点击翻译按钮获取中文摘要

CHALLENGEMany tens of thousands of Canadian children and youth with severe disabilities still do not have a means of communication despite the advent of numerous technologies that provide alternative access to computers and communication devices. These so-called access technologies tap into everything from muscle activations to brain waves. However, existing technologies fall short in part because they are unaware of the user’s emotional or mental state. For example, when a user becomes frustrated or tired, his or her brain and bodily control signals can change dramatically. As a result the access technology misinterprets the user’s intention, further exacerbating the decline in ability to communicate.PROPOSED RESEARCHThrough this NSERC discovery grant, we will build on recent research by our lab and others, showing that certain dispositions of the user, such as anxiety, intense positive or negative emotions and mental fatigue, can be reliably detected by machine. In particular, the aim of this next phase of the applicant’s research program is to systematically and comprehensively characterize fatigue, frustration and attention, user states that are most pertinent to access technology use. To realize this objective, we will measure brain activity (electrical and blood oxygenation) and temperature distributions in the face while users repeatedly perform a number of activities designed to induce the targeted changes in user state. We select the above measurements because literature has shown that brain and facial temperature modalities are highly sensitive to a user’s emotional response. We will develop models of physiological manifestations of user state using innovative imaging and multivariate pattern discovery techniques, among other methods. Some of the questions we aim to answer will include: (1) What overall patterns of physiological change are most predictive of changes in user state? (2) Are there sequences of physiological changes across measurement modalities which could be helpful in deciphering user states as they evolve? (3) What method of signal/image characteristics and accompanying classification algorithm can discriminate user states in an online fashion?Once we have a mechanism to discern user states, we will investigate ways in which this information could be used to improve communication via an alternative access technology. Specifically, we will start by exploring ways in which knowledge of user state might guide the adaptation of the user interface, for example, by providing helpful prompts to the user. We will also consider ways in which user state might inform the continuous improvement of a machine’s understanding of the user’s control signals.IMPORTANCEThis research is important because: (1) it will yield new analytical methods for the treatment and machine interpretation of multiple physiological measurements; and (2) it stands to vastly improve communication not only for children and youth, but for individuals of all ages who use technology to enable communication. The proposed program will thus strengthen Canada’s international leadership in rehabilitation engineering, in partnership with local and international academic, clinical and industrial stakeholders.ANTICIPATED OUTCOMESThrough this NSERC funding, 10 applied science and engineering doctoral students and 4 post-doctoral fellows will be trained in advanced instrumentation, signal processing and signal classification techniques relating to rehabilitation engineering. These individuals will be well-positioned to take leading roles in academia and industry, and thereby further advance scientific knowledge about brain and body-machine interfaces.