Fig 1.
Phases of the TFFO methodology.
Fig 2.
Accuracy of prediction for each electrode.
The frontal-central scalp area showed the most significant variance between healthy subjects and patients diagnosed with schizophrenia.
Table 1.
Discrimination accuracy per tested methodology.
Fig 3.
Average signal amplitude for healthy subjects and patients diagnosed with schizophrenia within total, early or late sets of events.
The first 5 reactions to the stimulations result in peak responses; the variance between healthy subjects and patients diagnosed with schizophrenia is larger than the variance obtained by the average amplitude of 8 entire event sets. Additionally, the last 5 events of both healthy subjects and patients diagnosed with schizophrenia resulted in almost no reaction and almost zero variance between the two classes.
Fig 4.
Accuracy of prediction as a function of events obtained from each subject’s initial recording of stimulation events.
By using only the first 7 or 8 events from each subject, the prediction accuracy of the methodology is close to optimum.
Fig 5.
Accuracy of methodology prediction as a function of time window within each stimulus recording.
The most variance (or local peaks of methodology accuracy) is best achieved with a ~300 ms window placed in the recording window post-stimulus.
Fig 6.
Accuracy of methodology prediction as a function of the selected frequency band.
The most variance (or local peaks of methodology accuracy) is best achieved using a frequency interval of 15–20 Hz or 22.5–27.5 Hz (mostly Beta2 frequencies). It is clear that the Delta, Theta and Alpha frequency bands possess less discrimination power.