RobIn: A robust interpretable deep network for schizophrenia diagnosis

作者：

Highlights：

• Initial clinical sessions predict schizophrenia diagnosis at six months.

• Squeeze and Excitation adds model robustness for out-of-distribution generalisation.

• Self-attention improves model efficacy and interpretability.

• Traditional machine learning techniques do not pass stress tests.

摘要

•Initial clinical sessions predict schizophrenia diagnosis at six months.•Squeeze and Excitation adds model robustness for out-of-distribution generalisation.•Self-attention improves model efficacy and interpretability.•Traditional machine learning techniques do not pass stress tests.

论文关键词：Schizophrenia,Deep learning,Self-attention,Interpretability,Out-of-distribution

论文评审过程：Received 1 December 2021, Revised 11 February 2022, Accepted 30 March 2022, Available online 11 April 2022, Version of Record 26 April 2022.

论文官网地址：https://doi.org/10.1016/j.eswa.2022.117158