Correlation of frontal lobe electroencephalogram asymmetry and cognitive prognosis in schizophrenia

Abstract

BACKGROUND

Schizophrenia is a debilitating psychiatric disorder marked by cognitive impairments that affect functional outcomes. However, the neurobiological underpinnings of these deficits, particularly in relation to frontal lobe electroencephalography (EEG) asymmetry, remain insufficiently understood. This study aims to examine the association between frontal lobe EEG asymmetry and cognitive prognosis in individuals with schizophrenia.

AIM

To explore the association between frontal lobe EEG asymmetry and cognitive prognosis in patients with schizophrenia.

METHODS

This retrospective study analyzed data from 104 patients with schizophrenia who received treatment between 2020 and 2023. The patients were divided into good (n = 52) and poor (n = 52) cognitive prognosis groups based on post-treatment scores on the Measurement and Treatment Research to Improve Cognition in Schizophrenia Consensus Cognitive Battery. Frontal EEG asymmetry was assessed under eyes-open and eyes-closed conditions, and frontoparietal network (FPN) connectivity was evaluated using resting-state functional magnetic resonance imaging. Statistical correlations between EEG asymmetry, FPN connectivity, and cognitive outcomes were explored.

RESULTS

Notable differences in EEG asymmetry were observed under eyes-open conditions. Patients in the good cognitive prognosis group demonstrated mean eyes-open frontal alpha asymmetry values of -0.09 at F4-F3 and -0.155 at F6-F5, whereas those in the poor prognosis group showed values of -0.10 and -0.183, respectively. These differences were significantly correlated with cognitive outcomes (P = 0.0482 and P = 0.0111, respectively). Furthermore, left and right FPN connectivity were negatively correlated with poor cognitive prognosis (P = 0.015 and P = 0.008, respectively). However, no significant difference in laterality index was observed.

CONCLUSION

The findings suggest that eyes-open frontal EEG asymmetry and FPN connectivity are associated with cognitive prognosis in individuals with schizophrenia.

Key Words: Schizophrenia; Cognitive prognosis; Electroencephalogram asymmetry; Frontal lobe; Frontoparietal connectivity; Neurophysiology

Core Tip: This study investigates the correlation between frontal lobe electroencephalogram asymmetry and cognitive prognosis in patients with schizophrenia. By focusing on frontal alpha asymmetry and frontoparietal network connectivity, we explore novel neurophysiological biomarkers for predicting cognitive outcomes. Our findings suggest that frontal alpha asymmetry under eyes-open conditions may serve as a critical indicator of cognitive recovery post-treatment. These insights could enhance personalized treatment plans and improve prognostic evaluations in clinical settings.

INTRODUCTION

Schizophrenia is a severe psychiatric disorder characterized by a constellation of symptoms, including delusions, hallucinations, cognitive impairments, and negative affective states[1]. These symptoms impair an individual’s functional capacity and quality of life[2]. Among these, cognitive dysfunction has garnered increasing attention[3]. Cognitive deficits in schizophrenia are profound, affecting working memory, executive function, attention, and social cognition, and are considered critical determinants of functional outcomes[4]. However, the neurobiological mechanisms underlying these cognitive impairments remain inadequately understood, highlighting the need for further research[5].

Electroencephalography (EEG) offers a noninvasive method for examining brain activity and has been extensively used in research on neurophysiological changes in schizophrenia[6]. Among the EEG-derived metrics, frontal alpha asymmetry (FAA) has received growing attention. FAA reflects the distribution of alpha band power (8-12 Hz) between the right and left frontal cortices[7] and has traditionally been associated with affective processing, motivational states, and approach-avoidance behavior, particularly in clinical populations with mood disorders[8]. Emerging evidence suggests that FAA is relevant to schizophrenia, but its role in predicting cognitive outcomes within this population remains largely understudied[9].

The frontal lobes, particularly the prefrontal cortex, play a critical role in cognitive functions, such as working memory, planning, decision-making, and social interactions[10]. Dysregulated activity in these regions has been implicated in diverse cognitive deficits observed in individuals with schizophrenia[11]. Current neurobiological models suggest that disrupted dopaminergic signaling, a core feature of schizophrenia pathophysiology, contributes to prefrontal cortex dysfunction, thereby exacerbating cognitive symptoms[12]. Asymmetries in EEG signals, which may reflect lateralized metabolic activity or dysfunction, can hold prognostic significance for cognitive recovery following schizophrenia treatment[13].

Despite these insights, the relationship between frontal EEG asymmetry and cognitive outcomes in schizophrenia remains relatively underexplored[14]. Such asymmetry may reflect compensatory neural mechanisms or disruptions in typical hemispheric functioning that are critical for various cognitive domains[15]. However, few studies have systematically examined how specific patterns of frontal asymmetry relate to cognitive prognoses in schizophrenia, particularly in the context of pharmacological interventions[16].

The conceptual framework guiding this study posits that frontal EEG asymmetry (FAA) reflects an imbalance in prefrontal metabolic activity between hemispheres, and disrupted frontoparietal network (FPN) connectivity signifies inefficiencies in frontoparietal information transfer. Collectively, these neural markers may limit cognitive plasticity during treatment. This model aligns with existing theories of dopaminergic dysfunction in schizophrenia and extends prior research on FAA in mood disorders by applying it to the context of cognitive prognosis[17].

This study aimed to investigate the association between frontal lobe EEG asymmetry and cognitive prognosis in patients with schizophrenia. The analysis focused on relative alpha power across several frontal regions, including the mid-frontal, lateral-frontal, anterior temporal, prefrontal, and medial prefrontal cortices, under both eyes-open and eyes-closed conditions. Through a comprehensive examination of these relationships, we sought to identify EEG-based biomarkers that may serve as indicators of cognitive prognosis in schizophrenia.

MATERIALS AND METHODS

Study design

This study conducted a retrospective analysis of 104 patients diagnosed with schizophrenia who received treatment at Ningbo Psychiatric Hospital between January 2020 and December 2023. The objective was to assess cognitive prognosis one year after treatment using the Measurement and Treatment Research to Improve Cognition in Schizophrenia Consensus Cognitive Battery (MCCB). Based on the median post-treatment MCCB scores, patients were classified into two groups: The good cognitive prognosis group (n = 52) and poor cognitive prognosis group (n = 52). The sample size (n = 104) was determined on the basis of prior studies reporting moderate effect sizes (Cohen’s d = 0.3-0.5) in FAA differences among schizophrenia subgroups. A post hoc power analysis indicated that the study had 80% power to detect significant FAA differences at α value of 0.05, assuming a standard deviation of 0.03.

Inclusion and exclusion criteria

Inclusion criteria: Participants were eligible for inclusion when they met the following criteria: (1) Age of 18 years or older; (2) Diagnosis of schizophrenia according to the International Classification of Diseases, Tenth Revision criteria for schizophrenia; (3) Undisrupted adherence to prescribed medication regimens without interruption; (4) Right-handedness; (5) Complete clinical documentation; and (6) Availability for regular follow-up assessments.

Exclusion criteria: Participants were excluded from the study when they met any of the following conditions: (1) Presence of neurological disorders, such as cerebrovascular disease and epilepsy, and a history of brain injury resulting in coma; (2) Dependence on substances, including alcohol and heroin, or diagnosis of a personality disorder; (3) Pregnancy or breastfeeding; (4) Presence of serious physical illnesses, such as cardiac, hepatic, and renal failure, or any form of cancer; and (5) Inability to cooperate with EEG assessments.

Clinical treatment

Upon admission, all patients were enrolled in a standardized one-year treatment regimen. The specific antipsychotic medications administered varied across participants, guided by clinical judgment and individualized treatment needs. Initial assessments were conducted to evaluate medication tolerability prior to initiating long-term therapy. Owing to the retrospective design of the study, variability was present in both the types and dosages of antipsychotic medications. For instance, some patients received intramuscular injections of risperidone microspheres, while others were treated with oral formulations such as olanzapine or aripiprazole. Patients newly prescribed these medications underwent a short-term trial period for the assessment of tolerability before the continuation of the prescribed regimen. For each participant, the treatment regimen remained consistent throughout the study period. The inclusion criteria did not restrict the type of antipsychotic medication, as long as the same medication was administered continuously during the treatment period. This approach was designed to minimize potential confounding effects related to medication variability, thereby allowing us to focus on the primary research objective.

Positive and negative syndrome scale

The Positive and Negative Syndrome Scale (PANSS) was employed for the assessment of the severity of schizophrenia symptoms prior to treatment. The PANSS comprises 30 items divided into three subscales: Positive symptoms, negative symptoms, and general psychopathology. Each item is rated on a seven-point scale, ranging from 1 (absent) to 7 (extremely severe). The positive symptom subscale includes features, such as delusions and hallucinations, whereas the negative symptom subscale captures impairments, such as emotional blunting and social withdrawal. The general psychopathology subscale encompasses symptoms, including anxiety and depression. PANSS scores were derived from structured professional interviews and clinical observations conducted over the preceding week. Higher total and subscale scores reflect greater symptom severity, whereas lower scores indicate milder or absent symptoms.

MCCB

Following one year of treatment, the MCCB was administered to evaluate patients’ cognitive function prognosis. The MCCB includes 10 subtests that assess seven key cognitive domains: (1) Speed of processing: Include the Brief Assessment of Cognition in Schizophrenia Symbol Coding and Trail Making Test A; (2) Attention/vigilance: Assessed by the Continuous Performance Test-Identical Pairs; (3) Working memory: Measured through the Wechsler Memory Scale-Third Edition Spatial Span and Letter-Number Span; (4) Verbal learning: Evaluated using the Hopkins Verbal Learning Test-Revised; (5) Visual learning: Assessed by the Brief Visuospatial Memory Test-Revised; (6) Reasoning and problem solving: Measured with the Neuropsychological Assessment Battery Mazes; and (7) Social cognition: Includes Managing Emotions and the Penn Emotion Recognition Test.

Raw scores from each subtest were converted into standardized scores and subsequently into T-scores with established normative conversion tables. These T-scores were then aggregated to generate a composite cognitive function score. Higher composite scores reflect better cognitive functioning. All the participants’ MCCB scores were computed, with individuals scoring above the median classified into the good cognitive prognosis group and those scoring below the median placed in the poor cognitive prognosis group.

EEG acquisition, preprocessing, and FAA calculation

EEG data were recorded using a Nihon Kohden EEG-1200C system (Nihon Kohden Corporation, Tokyo, Japan) while participants were seated comfortably in a sound-attenuated room. Recordings were performed using the system’s integrated amplifier and a standard headcap equipped with gold-cup electrodes arranged according to the international 10-20 system. EEG signals were recorded from 19 standard scalp locations. For ocular artifact monitoring, disposable electrodes were placed: Vertically above and below the left eye (vertical electrooculography), and horizontally at the outer canthi of both eyes (horizontal electrooculography). All signals were referenced to linked mastoids (A1/A2), with the ground electrode positioned on the forehead (FPz). Electrode impedance was maintained below 10 kΩ throughout the recording. Data was acquired at a sampling rate of 500 Hz with an analog bandpass filter of 0.5-120 Hz.

Resting-state EEG was recorded for five minutes under both eyes-open and eyes-closed conditions. Data preprocessing and analysis were performed using the system’s native NeuroWorkbench software (Nihon Kohden Corporation, Tokyo, Japan). Ocular artifacts (blinks/saccades) were corrected using built-in regression-based algorithms in NeuroWorkbench, triggered by vertical electrooculography excursions exceeding ± 80 μV. Continuous data were bandpass-filtered between 1.0-70.0 Hz (infinite impulse response Butterworth, zero-phase shift) and segmented into 2.0-second epochs. Epochs containing artifacts (amplitude > ± 100 μV or gradient > 50 μV/sample) were automatically rejected. Fast Fourier transform was applied to 30 randomly selected artifact-free epochs per participant/condition to compute absolute power across standard frequency bands: Delta (1-4 Hz), theta (4-8 Hz), alpha (8-13 Hz), beta (13-30 Hz), and gamma (30-70 Hz).

FAA was calculated using six homologous frontal electrode pairs: F4-F3, F8-F7, FP2-FP1, AF4-AF3, F6-F5, and F2-F1. These electrode pairs were selected based on established protocols in depression research, where FAA has been extensively examined. Particular emphasis was placed on mid-frontal (F4-F3) and lateral-frontal (F8-F7) sites, as these regions are considered reliable indicators of FAA associated with affective processing and depressive phenotypes. Additional electrode pairs (FP2-FP1, AF4-AF3, F6-F5, F2-F1) were included to enable a more comprehensive mapping of frontal subregions. Alpha-band power (8-13 Hz; PSα) was extracted, and FAA was computed as the difference in log₁₀-transformed alpha power between right and left homologous sites: FAA = log₁₀(PSα_r) - log₁₀(PSα_l). More negative FAA values indicate relatively greater alpha power in the left hemisphere, reflecting reduced cortical activation (i.e., hypoactivation) in the left frontal region. The term “greater asymmetry” refers to deviations from zero (i.e., larger |FAA| values), with directionality (negative/positive) indexing lateralized activation patterns. Eyes-open and eyes-closed conditions were analyzed separately to account for state-dependent neural dynamics. Additional analyses of delta-band activity confirmed minimal residual ocular artifact, supporting the reliability of the alpha-band measures.

FPN analysis via resting-state functional magnetic resonance imaging

All magnetic resonance imaging (MRI) data were acquired on a united imaging uMR 586 3.0T scanner (United Imaging Healthcare, Shanghai, China) using the system’s 64-channel phased-array head-neck coil and native acquisition software (R001). Structural images were obtained with a three-dimensional magnetization-prepared rapid gradient echo sequence (echo time = 3.8 milliseconds, repetition time = 2000 milliseconds, inversion time = 990 milliseconds, field of view = 256 mm × 256 mm, acquisition matrix = 256 × 256, spatial resolution = 1.0 mm × 1.0 mm × 1.0 mm, 192 sagittal slices, GRAPPA acceleration factor = 2). Resting-state functional MRI data were collected using a single-shot gradient-echo echo-planar imaging pulse sequence (echo time = 30 milliseconds, repetition time = 2500 milliseconds, field of view = 220 mm × 220 mm, matrix = 64 × 64, voxel size = 3.44 mm × 3.44 mm × 4.0 mm, 35 axial slices, flip angle = 90°, bandwidth = 2232 Hz/Px, GRAPPA acceleration factor = 2). During the 8-minute resting-state functional MRI scan, participants maintained eyes-open fixation on a white cross centered on a black background.

Functional data preprocessing and analysis were performed using FSL 5.0.5 with uMR-specific B0 field correction based on system phase maps. Functional images underwent motion correction using MCFLIRT, geometric distortion correction (B0 unwarping), spatial smoothing with a 6 mm full width at half maximum Gaussian kernel, and high-pass temporal filtering (0.01-0.1 Hz). Structural-functional registration was achieved via boundary-based registration, followed by nonlinear normalization to MNI 152 standard space (FNIRT) with 3-mm isotropic resampling. To mitigate the potential confounding effects of head motion, strict motion exclusion criteria were applied: No participant exhibited head motion exceeding 1.5 mm in absolute mean displacement, in accordance with established thresholds in prior studies. The average head motion was 0.31 ± 0.08 mm in the healthy control group and 0.29 ± 0.07 mm in the schizophrenia group, with no statistically significant difference observed between the two groups.

Group-level Independent Component Analysis was conducted using the Multivariate Exploratory Linear Optimized Decomposition into Independent Components Version 3.14, with temporal concatenation applied across all participants. Bayesian dimensionality estimation identified 43 distinct spatiotemporal components. Template matching against FSL’s standard network templates confirmed the FPN components. Subject-specific spatial maps and time courses were derived via dual regression. From these data, mean connectivity values were extracted within gray matter-masked left and right FPNs. The laterality index (LI) of FPN connectivity was calculated using the following formula: LI = (left FPN - right FPN)/(left FPN + right FPN). Positive LI values indicate left-hemispheric dominance in FPN connectivity, whereas negative values indicate right-hemispheric dominance.

Statistical analysis

Statistical analyses were performed using SPSS version 29.0 (SPSS Inc., Chicago, IL, United States). Categorical variables were presented as n (%) and analyzed using the χ² test under appropriate conditions. Specifically, the standard χ² test was applied when the total sample size was ≥ 40 and the expected frequency (T) in all cells was ≥ 5. When the sample size was ≥ 40 but the expected frequency in any cell ranged from 1 to < 5, a continuity-corrected χ² test was used. For sample size of < 40 or when any expected frequency was < 1, Fisher’s exact test was employed to ensure statistical validity.

Continuous variables were assessed for normality using the Shapiro-Wilk test. Variables that followed a normal distribution were reported as mean ± SD, while non-normally distributed data were presented as medians and interquartile ranges [median (25^th percentile, 75^th percentile)] and analyzed using the Wilcoxon rank-sum test. A P value of < 0.05 was considered indicative of statistical significance. Correlation analyses were conducted using Pearson’s correlation for normally distributed continuous variables and Spearman’s correlation for ordinal or non-normally distributed variables.

RESULTS

Basic data

The two groups demonstrated comparable demographic and clinical characteristics prior to treatment. The mean age was 38.56 ± 4.96 years in the good cognitive prognosis group and 37.97 ± 4.68 years in the poor prognosis group, with no significant difference observed (P = 0.5341; Table 1). Gender distribution was similar, with 29 females and 27 males in the good prognosis group and 27 females and 25 males in the poor prognosis group (P = 0.694). Body mass index did not differ significantly between groups (21.86 ± 2.65 kg/m²vs 21.93 ± 2.34 kg/m²; P = 0.873), nor did years of education (13.86 ± 1.42 years vs 13.75 ± 1.53 years; P = 0.691). Histories of smoking and alcohol use and the presence of comorbidities, such as hypertension, diabetes, and depression, showed no significant group differences. Marital status and ethnicity were similarly distributed. The duration of illness averaged 5.27 ± 1.52 years in the good prognosis group and 5.23 ± 1.49 years in the poor prognosis group (P = 0.894). Furthermore, baseline clinical symptom scores, including positive, negative, and general psychopathology subscales, did not differ significantly between groups. Overall, there were no statistically significant differences in demographic variables or pre-treatment clinical profiles between the two groups.

Table 1 Comparison of demographic characteristics between two groups.

	Good cognitive prognosis group (n = 52)	Poor cognitive prognosis group (n = 52)	t/χ²	P value
Age (years)	38.56 ± 4.96	37.97 ± 4.68	0.624	0.5341
Female/male	29	27	0.155	0.694
Body mass index (kg/m2)	21.86 ± 2.65	21.93 ± 2.34	0.161	0.873
Years of education (years)	13.86 ± 1.42	13.75 ± 1.53	0.398	0.691
Smoking history (yes/no)	16	18	0.175	0.676
Drinking history (yes/no)	15	17	0.181	0.671
Complication
Hypertension (yes/no)	12	11	0.056	0.813
Diabetes (yes/no)	4	5	0.000	1.000
Depression (yes/no)	14	15	0.048	0.827
Marital status (married/unmarried)	28	24	0.615	0.433
Ethnicity (Han/other)	47	45	0.377	0.539
Course of disease	5.27 ± 1.52	5.23 ± 1.49	0.133	0.894
Pathogeny			0.253	0.969
Family genetic history	9	11
Psychological factor	21	20
Environmental factor	18	17
Internal factors	4	4
Clinical scale score before treatment
Positive symptom score	17.18 ± 2.65	17.97 ± 2.81	1.475	0.1433
Negative symptom score	19.71 ± 1.93	20.32 ± 1.85	1.645	0.103
General pathological symptom evaluation	36.36 ± 7.06	37.65 ± 7.82	0.883	0.3793

Asymmetry of EEG

For the F4-F3, the good cognitive prognosis group demonstrated a mean eyes-open FAA of -0.09 ± 0.02, compared to -0.10 ± 0.03 in the poor prognosis group, a difference that reached statistical significance (P = 0.0482; Cohen’s d = 0.35; Figure 1A and B). By contrast, the eyes-closed FAA values did not differ significantly between groups, with means of -0.02 ± 0.003 and -0.021 ± 0.005 for the good and poor prognosis groups, respectively (P = 0.219; Cohen’s d = 0.21). These findings suggest a significant group difference in frontal EEG asymmetry under eyes-open conditions, potentially indicative of state-dependent neural dynamics associated with cognitive prognosis in schizophrenia.

Figure 1 Comparison of asymmetry of electroencephalography. A and B: In lateral prefrontal cortex between two groups. The frontal alpha asymmetry (FAA) values were calculated using the difference in alpha power between right (F4) and left (F3) frontal regions; C and D: In anterior temporal region between two groups. The FAA values were calculated using the difference in alpha power between right (F8) and left (F7) frontal regions; E and F: In frontal lobe between two groups. The FAA values were calculated using the difference in alpha power between right (FP2) and left (FP1) frontal regions; G and H: In medial prefrontal cortex between two groups. The FAA values were calculated using the difference in alpha power between right (AF4) and left (AF3) frontal regions; I and J: In middle frontal gyrus between two groups. The FAA values were calculated using the difference in alpha power between right (F6) and left (F5) frontal regions; K and L: In prefrontal cortex between two groups. The FAA values were calculated using the difference in alpha power between right (F2) and left (F1) frontal regions. Statistical comparisons were performed using independent samples t-tests. FAA: Frontal alpha asymmetry; NS: No statistically significant difference. ^aP < 0.05.

For the F8-F7 eyes-open FAA, the good cognitive prognosis group exhibited a mean value of -0.278 ± 0.08, while the poor prognosis group showed a comparable mean of -0.275 ± 0.06 (P = 0.8292; Cohen’s d = 0.04; Figure 1C and D). Similarly, under eyes-closed conditions, FAA values were -0.070 ± 0.008 for the good prognosis group and -0.060 ± 0.007 for the poor prognosis group (P = 0.4991; Cohen’s d = 0.13). These findings indicate that EEG asymmetry in the anterior temporal region was comparable between groups, with no statistically significant differences observed.

For the FP2-FP1 eyes-open FAA, the good cognitive prognosis group had a mean value of -0.143 ± 0.03, while the poor prognosis group showed a similar mean of -0.141 ± 0.02 (P = 0.69; Cohen’s d = 0.07; Figure 1E and F). Under eyes-closed conditions, FAA values were -0.01 ± 0.003 for the good prognosis group and 0.004 ± 0.001 for the poor prognosis group (P = 0.1523; Cohen’s d = 0.28). These results suggest that EEG asymmetry in the prefrontal cortex does not significantly differ between the two cognitive prognosis groups.

For the AF4-AF3 eyes-open FAA, the good cognitive prognosis group had a mean value of -0.095 ± 0.02, compared to -0.100 ± 0.03 in the poor prognosis group (P = 0.3197; Cohen’s d = 0.18; Figure 1G and H). Under eyes-closed conditions, the FAA was -0.029 ± 0.003 for the good prognosis group and -0.030 ± 0.005 for the poor prognosis group (P = 0.219; Cohen’s d = 0.21). These findings indicate that EEG asymmetry in the medial prefrontal cortex was comparable between the two groups, with no statistically significant differences observed.

For the F6-F5, the good cognitive prognosis group exhibited a mean eyes-open FAA of -0.155 ± 0.05, whereas the poor prognosis group showed a mean of -0.183 ± 0.06 (P = 0.0111; Cohen’s d = 0.51; Figure 1I and J). In contrast, the eyes-closed FAA did not differ significantly between the groups, with mean values of -0.039 ± 0.04 for the good prognosis group and -0.050 ± 0.04 for the poor prognosis group (P = 0.1639; Cohen’s d = 0.27). These results suggest a significant difference in EEG asymmetry within the middle frontal gyrus under eyes-open conditions, potentially linked to cognitive outcomes.

For the F2-F1, the good cognitive prognosis group had a mean eyes-open FAA of -0.051 ± 0.006, compared to -0.049 ± 0.004 in the poor prognosis group (P = 0.0482; Cohen’s d = 0.35; Figure 1K and L). Conversely, the eyes-closed FAA did not differ significantly between groups, with mean values of -0.018 ± 0.003 for the good prognosis group and -0.019 ± 0.004 for the poor prognosis group (P = 0.1523; Cohen’s d = 0.28). These findings indicate a significant difference in EEG asymmetry in the lateral prefrontal cortex under eyes-open conditions, which appears to be associated with cognitive prognosis.

Bilateral FPN connectivity and LI

The good cognitive prognosis group demonstrated higher mean connectivity in the left FPN, with a value of 14.32 ± 2.94 compared with 12.71 ± 3.47 in the poor prognosis group (P = 0.012; Table 2). Similarly, right FPN connectivity was also greater in the good prognosis group, with a mean of 12.81 ± 2.68 vs 11.28 ± 3.31 in the poor prognosis group (P = 0.011). However, no significant difference was found in the LI of the FPN, with both groups exhibiting comparable values (0.06 ± 0.01 in the good prognosis group and 0.06 ± 0.02 in the poor prognosis group; P = 0.685). These results suggest that while bilateral FPN connectivity is associated with cognitive prognosis, the degree of hemispheric dominance, as measured by the LI, was not.

Table 2 Comparison of bilateral frontoparietal networks connectivity and laterality index between two groups.

	Good cognitive prognosis group (n = 52)	Poor cognitive prognosis group (n = 52)	t	P value
Left FPN1	14.32 ± 2.94	12.71 ± 3.47	2.554	0.012
Right FPN2	12.81 ± 2.68	11.28 ± 3.31	2.602	0.011
LI of FPN3	0.06 ± 0.01	0.06 ± 0.02	1413.500	0.685

¹Mean connectivity of left frontoparietal network.

²Mean connectivity of right frontoparietal network.

³Laterality index of mean connectivity of bilateral frontoparietal networks.

FPN: Frontoparietal network; LI: Laterality index.

Correlation analysis

A negative correlation was found between eyes-open F4-F3 FAA and poor cognitive prognosis [rho = -0.203, P = 0.039, 95% confidence interval (CI): -0.371 to -0.024], indicating that greater left-sided asymmetry in this region is associated with worse outcomes (Table 3). A stronger negative correlation was observed for eyes-open F6-F5 FAA (rho = -0.244, P = 0.013, 95%CI: -0.401 to -0.073). By contrast, eyes-open F2-F1 FAA demonstrated a positive correlation (rho = 0.203, P = 0.039, 95%CI: 0.024-0.371). Furthermore, both left and right FPN connectivity showed significant negative correlations with poor cognitive prognosis. The left FPN had a rho value of -0.239 (P = 0.015, 95%CI: -0.401 to -0.062), while the right FPN a rho value of -0.258 (P = 0.008, 95%CI: -0.421 to -0.082). These results underscore the potential prognostic relevance of frontal EEG asymmetry and FPN connectivity in schizophrenia.

Table 3 Correlation analysis between frontal lobe electroencephalography asymmetry and poor cognitive function prognosis in schizophrenia patients.

	rho	P value
Eyes-open F4-F3	-0.203	0.039
Eyes-open F6-F5	-0.244	0.013
Eyes-open F2-F1	0.203	0.039
Left FPN1	-0.239	0.015
Right FPN2	-0.258	0.008

¹Mean connectivity of left frontoparietal network.

²Mean connectivity of right frontoparietal network.

FPN: Frontoparietal network.

DISCUSSION

Our study reveals significant associations between frontal EEG asymmetry, FPN connectivity, and cognitive prognosis in individuals with schizophrenia. Specifically, eyes-open FAA at key frontal electrode pairs (F4-F3, F6-F5, and F2-F1), along with reduced bilateral FPN connectivity, differentiated patients with poorer cognitive outcomes following treatment. These findings contribute to a deeper understanding of the neurophysiological mechanisms underlying cognitive recovery in schizophrenia and highlight potential biomarkers for prognosis and therapeutic targeting.

The observed negative correlation between eyes-open F4-F3 FAA (reflecting relative left frontal hypoactivation) and cognitive prognosis is consistent with established models of prefrontal dysfunction in schizophrenia[18]. Dysregulation of dopaminergic pathways, which is a central pathophysiological mechanism[19], disrupts frontal cortical efficiency, particularly impacting lateral prefrontal regions essential for executive functions[20]. Notably, the state-dependent nature of this association, emerging only during eyes-open conditions, aligns with prior evidence that environmental engagement modulates frontal asymmetry[21]. This underscores the relevance of ecologically valid neural states in the identification and interpretation of cognitive biomarkers.

While FAA has been extensively investigated in affective disorders[22], its prognostic utility for cognitive outcomes in schizophrenia under eyes-open conditions represents a novel application. Our electrode-specific analysis diverges from recent studies emphasizing EEG microstates[23] or large-scale network measures[24], suggesting that FAA captures localized and complementary prognostic information. Significant asymmetry at F6-F5 and F2-F1 further highlights the involvement of lateral and dorsolateral prefrontal regions, which are key hubs for working memory and attentional control, functions strongly associated with real-world functional outcomes[25,26].

The finding of reduced bilateral FPN connectivity in the poor prognosis group reinforces a network-level perspective. The FPN is integral to cognitive control and adaptive processing[27,28]. Its impaired connectivity supports the “dysconnection hypothesis” of schizophrenia, which posits inefficient integration across distributed brain networks[29]. Notably, the absence of significant differences in the LI, despite bilateral reductions, suggests that the overall integrity of bilateral FPNs rather than hemispheric dominance is paramount for cognitive recovery.

These patterns likely reflect intertwined mechanisms: Perturbed dopaminergic signaling may underlie both frontal asymmetry and FPN dysconnectivity, impairing neural synchrony and efficiency[30]. The FAA findings could indicate either a primary deficit in left frontal engagement or a maladaptive right-hemisphere compensation attempt[31]. This variance may represent heterogeneity in neuroplastic responses to illness and treatment[32]. Our findings suggest that eyes-open FAA measures (F4-F3, F6-F5, and F2-F1) and FPN connectivity could serve as adjunctive prognostic tools. Integrating these markers may help stratify patients early in terms of cognitive recovery potential, which could guide the allocation of intensive cognitive remediation resources toward individuals with poor neurophysiological prognoses[33]. These neurophysiological measures may also become targets for neuromodulation interventions, such as transcranial magnetic stimulation or transcranial direct current stimulation, with the goal of normalizing asymmetry or enhancing connectivity. However, translating this into clinical practice requires addressing key challenges related to reliability, accessibility, and the standardization of protocols.

While our study offers valuable insights into the relationship between frontal lobe EEG asymmetry and cognitive function prognosis in schizophrenia, several limitations must be acknowledged. First, the relatively small sample size may constrain subgroup analyses and limit the generalizability of the findings to the broader schizophrenia population. The retrospective design and naturalistic medication regimen introduce potential variability; future prospective studies that control for antipsychotic type and dosage are necessary. The use of a median split to categorize cognitive prognosis may impose arbitrary thresholds. Alternative approaches, such as modeling continuous MCCB scores or applying machine learning-based clustering, may better capture the heterogeneity of cognitive outcomes in schizophrenia. Importantly, the cross-sectional nature of the neurophysiological assessment precludes causal inference regarding whether these markers represent stable traits, consequences, or dynamic states. Longitudinal studies tracking EEG measures and cognitive performance over time are essential[34]. The limited spatial resolution of EEG may also hinder precise localization of frontal asymmetry to specific prefrontal subregions; multimodal approaches such as simultaneous EEG-functional MRI could enhance anatomical specificity[35]. Finally, the focus on resting-state activity may not adequately reflect dysfunction under cognitive load. Task-based paradigms could provide complementary insights into state-dependent neural mechanisms[36]. Addressing these limitations in future research will be critical to validating and extending the current findings.

Moreover, these insights underscore the need for ongoing longitudinal studies to examine how neural and cognitive functions interact throughout the course of treatment and disease progression. Such research can determine whether the observed EEG asymmetries are modifiable through targeted interventions aimed at restoring normative brain activity or whether they represent stable traits of the disorder. Exploring therapeutic strategies that address neurotransmitter imbalances or enhance network connectivity may ultimately contribute to improved treatment efficacy and cognitive recovery in individuals with schizophrenia.

CONCLUSION

This study demonstrates that state-dependent eyes-open frontal EEG asymmetry, particularly at F4-F3, F6-F5, and F2-F1, and reduced bilateral FPN connectivity are associated with poor cognitive prognosis in schizophrenia. These findings extend the application of FAA to cognitive outcome prediction and link it with network-level dysfunction. The results highlight the potential of these markers as adjunctive prognostic tools and therapeutic targets, but important limitations related to study design, sample characteristics, and measurement methodology must be considered. Future large-scale, longitudinal, and multimodal studies that incorporate task-based paradigms and control for treatment variables are needed to validate these markers, clarify their temporal dynamics, and assess their clinical utility in guiding personalized intervention strategies. This study contributes to ongoing efforts to integrate neurophysiological metrics into the prognostic assessment of cognitive function in schizophrenia.