Machine Learning Approaches: From Theory to Application in Schizophrenia

Table 2

Comparison between three state-of-the-art studies. Performance evaluation.

Author	Sample size	MRI technique	Best accuracy (%)
[23]	SCZ = 23 (A) 46 (B); HC = 38 (A) 41 (B)	Structural T1-w	90.2 (A) 90.8 (B)
[25]	ARMS = 45; HC = 75	Structural T1-w	78÷94
[35]	SCZ = 20; HC = 20	fMRI	87

In the studies in the first and last rows, SVM was used to compare subjects affected by schizophrenia (SCZ) with healthy controls (HC). The study in the second row applied SVM to identify individuals in different at-risk mental states (ARMS) of psychosis. The last column shows the performance of each algorithm in terms of accuracy.