Predictive assessment of autism using unsupervised machine learning models

Abstract

The application of different artificial intelligence models in clinical decision support systems has been a research topic which mainly focuses on the diagnosis method. In this paper we describe the application of unsupervised machine learning models in decision supportive tools for predictive grading of autistic disorder. We used competitive learning networks and unsupervised data clustering methods to model the differential grading in childhood autistic rating scale CARS-based assessment. Modelling of conventional score-based assessment using unsupervised learning methods is the novelty in this work. Self-organisation feature map SOM with single input and four output units perform with a predictive ability of 100% during resubstitution testing.