Table 4.
Summary of work and predictions relating to the detection of SZ using data from electroencephalogram scans via various artificial intelligence techniques and machine learning algorithms.
| Study | Year | Subjects | Prediction | AI/ML Technique | |
|---|---|---|---|---|---|
| Patients | Control | ||||
| Knott et al. [126] | 1999 | 14 | 14 | at least 89.3% | DFA, Jackknife classification |
| Neuhaus et al. [127] | 2011 | 40 | 40 (matched) | 79.9% (balanced) | SVM (linear, quadratic and radial basis kernels), LDA, Quadratic discriminant analysis (QDA), KNN, naïve Bayes with equal and unequal variances and Mahalanobis classification |
| Iyer et al. [128] | 2012 | 13 | 20 | max 76% (ensemble averaging) 100% (single-trial) |
Random Forest, 10-fold stratified cross-validation |
| Laton et al. [129] | 2014 | 54 | 54 (sex- and age-matched) | up to 84.7% | Naïve Bayes, SVM and decision tree, with two of its improvements: adaboost and Random Forest |
| Neuhaus et al. [130] | 2014 | 144 | 144 (matched) | 74% (balanced) | LDA and QDA (with their diagonal variants), SVM (linear, polynomial, radial basis and multilayer perceptron kernels), Naïve Bayes, KNN (Euclidean and cosine distance measures) and Mahalanobis classification |
| Johannesen et al. [131] | 2016 | 40 | 12 | up to 87% | 1-norm SVM |
| Shim et al. [132] | 2016 | 34 | 34 | Maximum: 88.24% (combined) 80.88% (sensor-level) 85.29% (source-level) |
SVM, Leave-one-out cross-validation |
| Taylor et al. [133] | 2017 | 21 | 22 | 80.84% | SVM, Gaussian processes classifiers, MVPA |
| Krishnan et al. [134] | 2020 | 14 | 14 (sex- and age-matched) | up to 93% | Various, SVM (Radial Basis Function) |