Analysis of Electrocardiogram Signal and Ammonia Concentration For Clustering ASD Condition

One of the health disorders that has increased in prevalence in recent years is Autism or Autism Spectrum Disorder (ASD), a neurodevelopmental disorder that occurs from birth. The difficulty of the diagnosis and monitoring process, which is still based only on psychological conditions, is also a contributing factor to the increasing prevalence of ASD. In contrast, several diagnostic biomarkers have been developed, including electrocardiogram signal and ammonia concentration. However, its use has not been optimal. This study will analyze the two clinical indicators above to classify ASD conditions as an alternative reference in ASD monitoring. The analysis was carried out in two main stages using data taken from ASD subjects and normal subjects. First, pre-processing includes characterization, taking the characteristics of the baseline heart rate approximated by the trendline equation of the ECG signal and the maximum value of the NH3 concentration, and extraction using the Principal Component Analysis method. Second, the processing includes finding the optimal number of clusters and the clustering process using the Fuzzy C-Means (FCM) method. The search for the most optimal number of clusters uses the Fuzzy Subtractive Clustering (FSC) method and validates using the Xie-Beni Index. At the same time, a confusion matrix will test the clustering process. The study results found four clusters as the most optimal number of classes, namely normal, mild ASD, moderate ASD, and severe ASD. Then, for the accuracy of clustering measured using the confusion matrix method, the accuracy results are around 81.08%.