Machine learning in medical diagnosis. A case study in the identification of Autism Spectrum Disorder from ocular behaviour.

Despite recent advances, autism diagnosis remains a complex challenge due to the need for specialized medical resources, time, and materials. This often leads to late diagnoses, even in adulthood, hindering e ff ective interventions. On the other hand, the field of artificial intelligence and machine learning has seen remarkable progress. These techniques have opened up new opportunities in various areas, including medical diagnosis and Autism Spectrum Disorder (ASD). The primary objective of this article is to provide a general overview of the applicability of machine learning techniques in medical diagnosis, using a specific case of ASD as an example. A classification model based on the XGBoost algorithm has been developed, achieving a sensitivity of 82 % and a specificity of 74 % when classifying individual samples. Furthermore, by combining this model with a majority voting algorithm, highly noteworthy classification results are obtained in the test set.