Name: | Description: | Size: | Format: | |
---|---|---|---|---|
47.51 MB | Adobe PDF |
Authors
Abstract(s)
A demência é uma das principais causas de dependência e incapacidade entre as pessoas idosas em todo o mundo. A terapia de reminiscência é uma terapia não farmacológica comummente utilizada nos cuidados com demência devido ao seu valor terapêutico para as pessoas com demência. Esta terapia é útil para criar uma comunicação envolvente entre pessoas com demência e o resto do mundo, utilizando as capacidades preservadas da memória a longo prazo, em vez de enfatizar as limitações existentes por forma a aliviar a experiência de fracasso e isolamento social. As soluções tecnológicas de assistência existentes melhoram a terapia de reminiscência ao proporcionar uma experiência mais envolvente para todos os participantes (pessoas com demência, familiares e clínicos), mas não estão livres de lacunas: a) os dados multimédia utilizados permanecem inalterados ao longo das sessões, e há uma falta de personalização para cada pessoa com demência; b) não têm em conta as emoções transmitidas pelos dados multimédia utilizados nem as reacções emocionais da pessoa com demência aos dados multimédia apresentados; c) a perspectiva dos cuidadores ainda não foi totalmente tida em consideração. Para superar estes desafios, seguimos uma abordagem de concepção centrada no utilizador através de inquéritos mundiais, entrevistas de seguimento, e grupos de discussão com cuidadores formais e informais para informar a concepção de soluções tecnológicas no âmbito dos cuidados de demência. Para cumprir com os requisitos identificados, propomos novos métodos que facilitam a inclusão de emoções no loop durante a terapia de reminiscência para personalizar e diversificar o conteúdo das sessões ao longo do tempo. As contribuições desta tese incluem: a) um conjunto de requisitos funcionais validados recolhidos com os cuidadores formais e informais, os resultados esperados com o cumprimento de cada requisito, e um modelo de arquitectura para o desenvolvimento de soluções tecnológicas de assistência para cuidados de demência; b) uma abordagem end-to-end para identificar automaticamente múltiplas informações emocionais transmitidas por imagens; c) uma abordagem para reduzir a quantidade de imagens que precisam ser anotadas pelas pessoas sem comprometer o desempenho dos modelos de reconhecimento; d) uma técnica de fusão tardia interpretável que combina dinamicamente múltiplos sistemas de recuperação de imagens com base em conteúdo para procurar eficazmente por imagens semelhantes para diversificar e personalizar o conjunto de imagens disponíveis para serem utilizadas nas sessões.
Dementia is one of the major causes of dependency and disability among elderly subjects worldwide. Reminiscence therapy is an inexpensive non-pharmacological therapy commonly used within dementia care due to its therapeutic value for people with dementia. This therapy is useful to create engaging communication between people with dementia and the rest of the world by using the preserved abilities of long-term memory rather than emphasizing the existing impairments to alleviate the experience of failure and social isolation. Current assistive technological solutions improve reminiscence therapy by providing a more lively and engaging experience to all participants (people with dementia, family members, and clinicians), but they are not free of drawbacks: a) the multimedia data used remains unchanged throughout sessions, and there is a lack of customization for each person with dementia; b) they do not take into account the emotions conveyed by the multimedia data used nor the person with dementia’s emotional reactions to the multimedia presented; c) the caregivers’ perspective have not been fully taken into account yet. To overcome these challenges, we followed a usercentered design approach through worldwide surveys, follow-up interviews, and focus groups with formal and informal caregivers to inform the design of technological solutions within dementia care. To fulfil the requirements identified, we propose novel methods that facilitate the inclusion of emotions in the loop during reminiscence therapy to personalize and diversify the content of the sessions over time. Contributions from this thesis include: a) a set of validated functional requirements gathered from formal and informal caregivers, the expected outcomes with the fulfillment of each requirement, and an architecture’s template for the development of assistive technology solutions for dementia care; b) an end-to-end approach to automatically identify multiple emotional information conveyed by images; c) an approach to reduce the amount of images that need to be annotated by humans without compromising the recognition models’ performance; d) an interpretable late-fusion technique that dynamically combines multiple content-based image retrieval systems to effectively search for similar images to diversify and personalize the pool of images available to be used in sessions.
Dementia is one of the major causes of dependency and disability among elderly subjects worldwide. Reminiscence therapy is an inexpensive non-pharmacological therapy commonly used within dementia care due to its therapeutic value for people with dementia. This therapy is useful to create engaging communication between people with dementia and the rest of the world by using the preserved abilities of long-term memory rather than emphasizing the existing impairments to alleviate the experience of failure and social isolation. Current assistive technological solutions improve reminiscence therapy by providing a more lively and engaging experience to all participants (people with dementia, family members, and clinicians), but they are not free of drawbacks: a) the multimedia data used remains unchanged throughout sessions, and there is a lack of customization for each person with dementia; b) they do not take into account the emotions conveyed by the multimedia data used nor the person with dementia’s emotional reactions to the multimedia presented; c) the caregivers’ perspective have not been fully taken into account yet. To overcome these challenges, we followed a usercentered design approach through worldwide surveys, follow-up interviews, and focus groups with formal and informal caregivers to inform the design of technological solutions within dementia care. To fulfil the requirements identified, we propose novel methods that facilitate the inclusion of emotions in the loop during reminiscence therapy to personalize and diversify the content of the sessions over time. Contributions from this thesis include: a) a set of validated functional requirements gathered from formal and informal caregivers, the expected outcomes with the fulfillment of each requirement, and an architecture’s template for the development of assistive technology solutions for dementia care; b) an end-to-end approach to automatically identify multiple emotional information conveyed by images; c) an approach to reduce the amount of images that need to be annotated by humans without compromising the recognition models’ performance; d) an interpretable late-fusion technique that dynamically combines multiple content-based image retrieval systems to effectively search for similar images to diversify and personalize the pool of images available to be used in sessions.
Description
Keywords
tecnologia de assistência recuperação de imagem com base no conteúdo emoções terapia de reminiscência design centrado no utilizador assistive technology content-based image retrieval emotions reminiscence therapy user-centered design