OlharTEA: Intelligent System to Assist Health Professionals in Decision Making for Children in ASD
O Autism Spectrum Disorder (ASD), or Transtorno do Espectro Autista (TEA) in Portuguese, is characterized by challenges in the realm of neurodevelopment, affecting approximately one in every 200 individuals. It impacts areas such as socialization, communication, and learning. A person with ASD is born with the disorder, and there is no cure for this condition. This dissertation describes olharTEA, a system developed to utilize the unique way in which the eyes of individuals with ASD react to certain types of images, to assist the medical field in diagnosing children aged five to 12. OlharTEA involves the use of a webcam to capture the focal map of children, comparing it with focal maps from various children previously studied both with and without ASD, provided by a public dataset. The comparison is facilitated through the utilization of 34 classification methods from the Python Sklearn library, which is a part of the artificial intelligence (AI) toolkit. Among these classifiers, the ones with the highest precision were the SVC, KNeighborsClassifier, and HistGradientBoostingClassifier. This enabled olharTEA to make the desired predictions, indicating the likelihood of whether the analyzed child is within or outside the ASD spectrum. Usability tests were conducted involving 13 adults, and their feedback suggested potential adjustments, such as background color, image transition speed, and the placement of olharTEA's control buttons. The average overall usability satisfaction was 4.69 out of a maximum possible score of 5. With the development of olharTEA, it becomes evident that there are possibilities for various studies related to the use of computers in diagnosing various diseases and disorders, such as ASD.