Structure Polish Academy of Sciences

Researchers have taken full advantage of the benefits brought about by the artificial intelligence and used artificial neural networks to detect schizophrenia and autism. Their findings has been recently published in Cognitive Computation journal. Among the authors are scientists from two institute of the Polish Academy of Sciences.

Scientists explored the limits of automated detection of autism spectrum disorder (ASD) and schizophrenia (SCZ). They investigated the effectiveness of several baseline approaches, e.g. bag of words and dictionary-based vectors, followed by a machine learning model.

The authors of the article include: Dr. Aleksander Wawer from the Institute of Computer Science of the Polish Academy of Sciences, Dr. Łukasz Okruszek and Dr. Justyna Sarzyńska-Wawer from the Institute of Psychology of the Polish Academy of Sciences and Dr. Izabela Chojnicka from the Faculty of Psychology at the University of Warsaw.

Scientists used textual data obtained from interviews conducted with the diagnostic tools commonly used in clinical psychiatry, and then applied selected deep learning methods to text representation and inference. As clinicians typically provide small datasets, scientists focused on the effectiveness of few-shot learning methods (dedicated to low data size). They compared effectiveness of automated methods (in distinguishing people with SCZ and ASD from healthy people) with the methods traditionally used by psychiatrists (diagnostic tests). Interestingly, automatic methods outperformed human raters. According to the researchers, few-shot learning methods revealed promising results for the diagnosis of schizophrenia. The work of Polish researchers is a crucial contribution to automated diagnostics, especially for such fields as psychiatry, which still lacks objective tools supporting diagnostics.

The article "Single and Cross-Disorder Detection for Autism and Schizophrenia" has been published in Cognitive Computation journal.