Classification of Autism Spectrum Disorder by lower limb joint angles: machine learning approach
Oral Presentation
Paper ID : 1766-SSRC
Authors
1آموزش و پرورش
2ریاست جمهوری
3هیئت علمی علوم تحقیقات تهران
Abstract
Autism spectrum disorder (ASD) is an early developmental disorder characterized by mutation of enculturation associated with attention deficit disorder in the visual perception of emotional expressions. Research shows that one out of every 100 people has autism. Data analysis and classification of ASD is still challenging due to unsolved issues arising from many severity levels and range of signs and symptoms. So, we used Random Forrest (RF), a supervised machine learning method, in order to classify ASD from healthy children.
Method: In this study, 5 ASD (8.62 ± 2.53 years old) and 5 healthy subjects (9.38 ± 2.81 years old) participated. The subjects walked three times and motion capture was used to record kinematic data during walking in a frequency of 240 Hz. The data was filtered by Butterworth filter with cut of frequency of 10 Hz. Subjects gait cycles were identified and all data were time normalized to 100 points by quadratic interpolation. We used RF to classify the subjects into two groups. The inputs were lower limbs joint angles. We used 80 percent of data for training and 20 percent for testing. The number of trees was 50, the leaf number was 5, and split number was 5. MATLAB (R2022b) was used to create the network.
Results: Firstly, we used ankle, knee, and hip angles as the inputs of RF separately and the accuracy were 72, 73, and 78 respectively. Then, we used all the lower limbs angles as the inputs, the accuracy was 86.
Conclusion: Rf can classify ASD and healthy children with high accuracy. This outcome can very much help in the process of diagnosing ASD, where the whole process can be done in a more time-efficient manner and more accurate diagnosis can be made.
Method: In this study, 5 ASD (8.62 ± 2.53 years old) and 5 healthy subjects (9.38 ± 2.81 years old) participated. The subjects walked three times and motion capture was used to record kinematic data during walking in a frequency of 240 Hz. The data was filtered by Butterworth filter with cut of frequency of 10 Hz. Subjects gait cycles were identified and all data were time normalized to 100 points by quadratic interpolation. We used RF to classify the subjects into two groups. The inputs were lower limbs joint angles. We used 80 percent of data for training and 20 percent for testing. The number of trees was 50, the leaf number was 5, and split number was 5. MATLAB (R2022b) was used to create the network.
Results: Firstly, we used ankle, knee, and hip angles as the inputs of RF separately and the accuracy were 72, 73, and 78 respectively. Then, we used all the lower limbs angles as the inputs, the accuracy was 86.
Conclusion: Rf can classify ASD and healthy children with high accuracy. This outcome can very much help in the process of diagnosing ASD, where the whole process can be done in a more time-efficient manner and more accurate diagnosis can be made.
Keywords