DMADL-CTO: Hybrid Distributed Mixed Attention-based deep learning model for Cerebellar Ataxia Detection

Abstract

Cerebellar ataxia (CA) is a disability disease that originated in the cerebellum of the human brain and caused several awkward condition as in-organized body balances, eye movements, inability to gait, and extremities. Because of these symptoms, early detection of CA is mandatorily targeted by the researchers using different conventional methods. These existing detection methods faced various disadvantages, as poor performance effectiveness, complexity problems, increased error rate, and computational time consumption. To overcome these issues in the existing methods, a proposed model is designed as Hybrid Distributed Mixed Attention-based Deep Learning approach enabled Competitive Tuning Optimization (Hybrid DMADL-CTO) model that precisely detects the abnormalities along with its class labels. The developed model possessed a distributive learning approach that captures detective outcomes with minimal operational time. Meanwhile, the model's ability and effectiveness are effectually enhanced by the integration of Competitive tuning optimization (CTO) and mixed attention mechanisms. The developed algorithm eliminates the optimization problem and attained better durability, consistency, and convergence rate significantly. Additionally, the mixed attention mechanism minimized the complexity problem and achieved effective performance efficacy. Thus, the effectiveness of the research model is compared with other related techniques based on the performance metrics, such as accuracy, recall, and precision for which the proposed method acquired 97.67%, 97.99%, and 95.89% respectively.