| Name: | Description: | Size: | Format: | |
|---|---|---|---|---|
| 1.56 MB | Adobe PDF |
Authors
Advisor(s)
Abstract(s)
O declínio irreversivelmente progressivo das funções fisiológicas é conhecido como
envelhecimento. Entre essas alterações está o envelhecimento cerebral, que leva ao
declínio cognitivo e à promoção da expressão da demência. Isso afeta diretamente a
memória, a aprendizagem e as habilidades motoras, o que resulta em uma diminuição na
eficiência da marcha. O objetivo foi investigar a viabilidade de identificar e classificar o risco
de Demência com base na análise das variáveis cinemáticas relacionadas com a marcha de
idosos, utilizando algoritmos de machine learning. Este estudo observacional transversal
examinou uma amostra de 59 indivíduos com idades entre 60 ± 8 anos. Foram divididos em
um grupo de idosos institucionalizados (GI) com 26 participantes e outro grupo não
institucionalizado (GNI) com 33 participantes, todos residentes na cidade de Bragança. Os
dados de marcha foram avaliados através de uma caminhada de 10 metros, capturados em
vídeo e posteriormente analisados pelo software Kinovea. O estado cognitivo foi avaliado
por meio do questionário Mini Exame do Estado Mental (MEEM). Para a análise estatística,
o programa PythonTM foi utilizado para criar um modelo de algoritmos de machine learning
para classificar o risco de demência associado aos idosos de acordo com suas variáveis de
marcha. Os resultados deste estudo mostraram que os modelos algorítmicos alcançaram
um desempenho global de 74,6%, com o algoritmo Ada Boost a liderar com 83,5%. A
validação cruzada dos algoritmos revelou uma precisão global de 72%, com o Classificador
de Vetores de Suporte a apresentar o melhor desempenho individual 80%, indicando que
este modelo respondeu corretamente a 80% das classificações nos diferentes subconjuntos
de dados. Portanto, concluímos que a avaliação da marcha, combinada com algoritmos de
machine learning, evidenciou a relação entre as variáveis da marcha e a demência,
tornando-se uma técnica segura e eficiente para a classificação da demência.
The irreversibly progressive decline in physiological functions is known as aging. Among these changes is brain aging, which leads to cognitive decline and the promotion of dementia expression. This directly affects memory, learning, and motor skills, which results in a decrease in gait efficiency. The aim was to investigate the feasibility of identifying and classifying the risk of dementia based on the analysis of kinematic variables related to gait in the elderly, using machine learning algorithms. This cross-sectional observational study examined a sample of 59 individuals aged 60 ± 8 years. They were divided into a group of institutionalized elderly (GI) with 26 participants and another non-institutionalized group (GNI) with 33 participants, all living in the city of Bragança. Gait data were collected through a 10- meter walk, captured on video and later analyzed by the Kinovea software. Cognitive status was assessed using the Mini Mental State Examination (MMSE) questionnaire. For the statistical analysis, the PythonTM program was used to create a model of machine learning algorithms to classify the risk of dementia associated with the elderly according to their gait variables. The results of this study showed that the algorithmic models achieved an overall performance of 74.6%, with the Ada Boost algorithm leading with 83.5%. Cross-validation of the algorithms revealed an overall accuracy of 72%, with the Support Vector Classifier showing the best individual performance of 80%, indicating that this model correctly responded to 80% of the classifications in the different subsets of data. Therefore, we conclude that gait assessment, combined with machine learning algorithms, evidenced the relationship between gait characteristics and dementia, making it a safe and efficient technique for dementia classification.
The irreversibly progressive decline in physiological functions is known as aging. Among these changes is brain aging, which leads to cognitive decline and the promotion of dementia expression. This directly affects memory, learning, and motor skills, which results in a decrease in gait efficiency. The aim was to investigate the feasibility of identifying and classifying the risk of dementia based on the analysis of kinematic variables related to gait in the elderly, using machine learning algorithms. This cross-sectional observational study examined a sample of 59 individuals aged 60 ± 8 years. They were divided into a group of institutionalized elderly (GI) with 26 participants and another non-institutionalized group (GNI) with 33 participants, all living in the city of Bragança. Gait data were collected through a 10- meter walk, captured on video and later analyzed by the Kinovea software. Cognitive status was assessed using the Mini Mental State Examination (MMSE) questionnaire. For the statistical analysis, the PythonTM program was used to create a model of machine learning algorithms to classify the risk of dementia associated with the elderly according to their gait variables. The results of this study showed that the algorithmic models achieved an overall performance of 74.6%, with the Ada Boost algorithm leading with 83.5%. Cross-validation of the algorithms revealed an overall accuracy of 72%, with the Support Vector Classifier showing the best individual performance of 80%, indicating that this model correctly responded to 80% of the classifications in the different subsets of data. Therefore, we conclude that gait assessment, combined with machine learning algorithms, evidenced the relationship between gait characteristics and dementia, making it a safe and efficient technique for dementia classification.
Description
Keywords
Risco de demência Análise Marcha Pessoas idosas Algoritmos Machine learning