Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2023 May 26;66(1):e46.
doi: 10.1192/j.eurpsy.2023.2410.

A multivariate approach to investigate the associations of electrophysiological indices with schizophrenia clinical and functional outcome

Luigi Giuliani  1 , Nikolaos Koutsouleris  2 , Giulia Maria Giordano  1 , Thomas Koenig  3 , Armida Mucci  1 , Andrea Perrottelli  1 , Anne Reuf  2 , Mario Altamura  4 , Antonello Bellomo  4 , Roberto Brugnoli  5 , Giulio Corrivetti  6 , Giorgio Di Lorenzo  7 , Paolo Girardi  8 , Palmiero Monteleone  9 , Cinzia Niolu  7 , Silvana Galderisi  1 , Mario Maj  1 ; Italian Network for Research on Psychoses
Affiliations

A multivariate approach to investigate the associations of electrophysiological indices with schizophrenia clinical and functional outcome

Luigi Giuliani et al. Eur Psychiatry. .

Abstract

Background: Different electrophysiological (EEG) indices have been investigated as possible biomarkers of schizophrenia. However, these indices have a very limited use in clinical practice, as their associations with clinical and functional outcomes remain unclear. This study aimed to investigate the associations of multiple EEG markers with clinical variables and functional outcomes in subjects with schizophrenia (SCZs).

Methods: Resting-state EEGs (frequency bands and microstates) and auditory event-related potentials (MMN-P3a and N100-P3b) were recorded in 113 SCZs and 57 healthy controls (HCs) at baseline. Illness- and functioning-related variables were assessed both at baseline and at 4-year follow-up in 61 SCZs. We generated a machine-learning classifier for each EEG parameter (frequency bands, microstates, N100-P300 task, and MMN-P3a task) to identify potential markers discriminating SCZs from HCs, and a global classifier. Associations of the classifiers' decision scores with illness- and functioning-related variables at baseline and follow-up were then investigated.

Results: The global classifier discriminated SCZs from HCs with an accuracy of 75.4% and its decision scores significantly correlated with negative symptoms, depression, neurocognition, and real-life functioning at 4-year follow-up.

Conclusions: These results suggest that a combination of multiple EEG alterations is associated with poor functional outcomes and its clinical and cognitive determinants in SCZs. These findings need replication, possibly looking at different illness stages in order to implement EEG as a possible tool for the prediction of poor functional outcome.

Keywords: EEG; functional outcome; machine learning; schizophrenia.

PubMed Disclaimer

Conflict of interest statement

The authors declare none.

Figures

Figure 1.
Figure 1.
Experimental design of the machine-learning pipelines used to train and cross-validate the unimodal and stacked classifiers. We used nested, repeated cross-validation to train and validate the four individual machine-learning classifiers, consisting of an outer 10-fold cross-validation cycle (CV2), which provided validation participants for computing an unbiased estimate of predictor generalisability to new patients, and an inner 10-fold cross-validation cycle (CV1), which delivered training participants to the multivariate pattern analysis pipeline as well as test participants for features and parameters optimisation. The same nested cross-validation structure was applied to the stacked machine-learning classifier, obtained by combining unimodal classifiers’ outputs within the machine-learning environment. CV, cross-validation; NN, nearest neighbor; SVM, support vector machine.
Figure 2.
Figure 2.
Projection of illness-related and functioning variables, measured at baseline (left) and follow-up (right), to four factors, using Non-Negative Matrix Factorization.
Figure 3.
Figure 3.
Composition of predictive variable sets selected by the unimodal machine-learning classifiers: frequency bands (A), microstates (B), MMN-P3a (C), and N100-P3b (D). The features were first ranked according to the selection probability measured across all inner-cycle training partitions. Variables ranking among the top 10% of selected features were marked with red and listed with their selection probability (psel) and correlation with the classifier’s outcome (Spearman’s ρ).
Figure 4.
Figure 4.
Contribution (Spearman’s ρ) of each individual EEG data modality to the global classifier’s decisions.
See this image and copyright information in PMC

References

    1. Fusar-Poli P, Estrade A, Stanghellini G, Venables J, Onwumere J, Messas G, Gilardi L, Nelson B, Patel V, Bonoldi I, Aragona M, Cabrera A, Rico J, Hoque A, Otaiku J, Hunter N, Tamelini MG, Maschiao LF, Puchivailo MC, Piedade VL, Keri P, Kpodo L, Sunkel C, Bao J, Shiers D, Kuipers E, Arango C, Maj M. The lived experience of psychosis: a bottom-up review co-written by experts by experience and academics. World Psychiatry. 2022;21(2):168–88. - PMC - PubMed
    1. Killaspy H, Harvey C, Brasier C, Brophy L, Ennals P, Fletcher J, et al. Community-based social interventions for people with severe mental illness: a systematic review and narrative synthesis of recent evidence. World Psychiatry. 2022;21(1):96–123. - PMC - PubMed
    1. GBD 2016 Disease and Injury Incidence and Prevalence Collaborators. Global, regional, and national incidence, prevalence, and years lived with disability for 328 diseases and injuries for 195 countries, 1990-2016: a systematic analysis for the global burden of disease study 2016. Lancet. 2017;390(10100):1211–59. - PMC - PubMed
    1. Galderisi S, Rossi A, Rocca P, Bertolino A, Mucci A, Bucci P, Rucci P, Gibertoni D, Aguglia E, Amore M, Bellomo A, Biondi M, Brugnoli R, Dell–Osso L, De Ronchi D, Di Emidio G, Di Giannantonio M, Fagiolini A, Marchesi C, Monteleone P, Oldani L, Pinna F, Roncone R, Sacchetti E, Santonastaso P, Siracusano A, Vita A, Zeppegno P, Maj M, Italian Network for Research on Psychoses. The influence of illness-related variables, personal resources and context-related factors on real-life functioning of people with schizophrenia. World Psychiatry. 2014;13(3):275–87. - PMC - PubMed
    1. Galderisi S, Rucci P, Kirkpatrick B, Mucci A, Gibertoni D, Rocca P, et al. Interplay among psychopathologic variables, personal resources, context-related factors, and real-life functioning in individuals with schizophrenia: a network analysis. JAMA Psychiatry. 2018;75(4):396–404. - PMC - PubMed

Publication types