基于RepNet的自闭症健康监测方法研究

Study on the Health Monitoring of Autism Spectrum Disorder Based on RepNet

自闭症谱系障碍(ASD)患者在社交互动、语言交流和兴趣等方面表现出正常行为的偏差。随着自闭症患者数量的增加,迫切需要在早期进行自闭症筛查,以尽早提供专业治疗。目前自闭症筛查主要仰赖家长填写调查问卷和医疗专业人员进行人工观察和评估,这种方式效率较低且需要耗费较长的时间。机器算法在自闭症筛查领域的应用尚未得到广泛采纳。为提高筛查效率和准确性,探讨机器视觉算法RepNet在识别自闭症患者重复动作方面的有效性,使用RepNet算法分析包含重复性和非重复性动作的视频,评估Repnet的准确性;通过分析包含自闭症患儿重复动作的视频,评估RepNet在自闭症重复性动作筛查中的准确性和匹配性。结果表明:RepNet在检测自闭症患者的重复行为方面表现出极高的有效性和准确性。

Autism spectrum disorders(ASD)patients exhibit deviations in social interaction,language communication,and interests,despite displaying normal behavior in these aspects.With the increasing number of individuals diagnosed with autism,there is an urgent need for early autism screening to facilitate timely provision of specialized treatment.Currently,autism screening heavily relies on parents filling out questionnaires and healthcare professionals conducting manual observations and assessments,a process known for its low efficiency and timeconsuming nature.The application of machine algorithms in the field of autism screening has not been widely adopted.To enhance screening efficiency and accuracy,the effectiveness of the machine vision algorithm RepNet in identifying repetitive movements in autism patients is explored;videos containing repetitive and non-repetitive actions are analyzed to evaluate the accuracy of RepNet;through the analysis of videos featuring repetitive movements in autism patients,the RepNet’s accuracy and matching capabilities in screening for repetitive actions associated with autism are assessed.The results show that RepNet exhibits extremely high effectiveness and accuracy in detecting repetitive behaviors in individuals with autism.