Identification of distinct clinical profiles and trajectories in individuals at high risk of developing psychosis: A latent profile analysis of the north American prodrome longitudinal study consortium-3 dataset.

AIM: People at clinical high risk (CHR) for psychosis are a heterogeneous population in regard to clinical presentation and outcome. It is unclear, however, if their baseline clinical characteristics can be used to construct orthogonal subgroups that differ in their clinical trajectory to provide early identification of individuals in need of tailored interventions. METHODS: We used latent profile analysis (LPA) to determine the number of distinct clinical profiles within the CHR population using the NAPLS-3 dataset, focusing on the clinical features incorporated in the NAPLS psychosis risk calculator (including age, unusual thought content and suspiciousness, processing speed, verbal learning and memory function, social functioning decline, life events, childhood trauma, and family history of psychosis). We then conducted a between-profile comparisons of clinical trajectories based on psychotic and depressive symptoms as well as substance use disorder (SUD) related features over time. RESULTS: Two distinct profiles emerged. One profile, comprising approximately 25% of the sample, was significantly older, displayed better cognitive performance, experienced more types of traumatic and undesirable life events, exhibited a greater decline in functioning in the past year, and was more likely to have relatives with psychosis. This group showed worse positive symptoms and SUD-related features over time, although groups did not differ in the proportion of individuals who developed psychosis. CONCLUSIONS: LPA results suggest CHRs can be segregated into two profiles with different clinical trajectories. Characterizing individuals within these clinical profiles may help understand the divergent outcomes of this population and ultimately facilitate the development of specialized interventions.

Greater Choline-Containing Compounds and Myo-inositol in Treatment-Resistant Versus Responsive Schizophrenia: A 1H-Magnetic Resonance Spectroscopy Meta-analysis.

BACKGROUND: The neurobiology of treatment-resistant schizophrenia (TRS) is poorly understood, and meta-analytic consensus regarding magnetic resonance spectroscopic profiles of glutamate, choline-containing compounds, myo-inositol, and other metabolites in the condition is lacking. METHODS: In this meta-analysis, we examined published findings for N-acetylaspartate, choline-containing compounds (phosphocholine+glycerophosphocholine), myo-inositol, creatine+phosphocreatine, glutamate, and glutamate+glutamine in the anterior cingulate cortex and dorsal striatum in people with TRS versus non-TRS as well as TRS versus healthy control participants (HCs) and TRS versus ultra TRS (i.e., TRS with clozapine resistance). A MEDLINE search revealed 9 articles including 239 people with pooled TRS and ultra TRS, 59 with ultra TRS, 175 with non-TRS, and 153 (HCs) that met meta-analytic criteria. RESULTS: Significant effects included higher anterior cingulate cortex phosphocholine+glycerophosphocholine and myo-inositol in the pooled TRS and ultra TRS group than in both the non-TRS group and HCs as well as higher dorsal striatal phosphocholine+glycerophosphocholine in ultra TRS versus HCs, but no differences in other regional metabolites. CONCLUSIONS: The observed metabolite profile in TRS (higher phosphocholine+glycerophosphocholine and myo-inositol signal) is consistent with the hypothesis that TRS has a neuroinflammatory component, although this meta-analysis is not a critical test of that hypothesis. A similar profile is seen in healthy aging, which is known to involve increased neuroinflammation and glial activation. Because the overall number of datasets was low, however, results should be considered preliminary and highlight the need for additional studies of brain metabolites in TRS and their possible association with inflammatory processes.

Meta-analytic evidence of elevated choline, reduced N-acetylaspartate, and normal creatine in schizophrenia and their moderation by measurement quality, echo time, and medication status.

BACKGROUND: Brain metabolite abnormalities measured with magnetic resonance spectroscopy (MRS) provide insight into pathological processes in schizophrenia. Prior meta-analyses have not yet answered important questions about the influence of clinical and technical factors on neurometabolite abnormalities and brain region differences. To address these gaps, we performed an updated meta-analysis of N-acetylaspartate (NAA), choline, and creatine levels in patients with schizophrenia and assessed the moderating effects of medication status, echo time, measurement quality, and other factors. METHODS: We searched citations from three earlier meta-analyses and the PubMed database after the most recent meta-analysis to identify studies for screening. In total, 113 publications reporting 366 regional metabolite datasets met our inclusion criteria and reported findings in medial prefrontal cortex (MPFC), dorsolateral prefrontal cortex, frontal white matter, hippocampus, thalamus, and basal ganglia from a total of 4445 patient and 3944 control observations. RESULTS: Patients with schizophrenia had reduced NAA in five of the six brain regions, with a statistically significant sparing of the basal ganglia. Patients had elevated choline in the basal ganglia and both prefrontal cortical regions. Patient creatine levels were normal in all six regions. In some regions, the NAA and choline differences were greater in studies enrolling predominantly medicated patients compared to studies enrolling predominantly unmedicated patients. Patient NAA levels were more reduced in hippocampus and frontal white matter in studies using longer echo times than those using shorter echo times. MPFC choline and NAA abnormalities were greater in studies reporting better metabolite measurement quality. CONCLUSIONS: Choline is elevated in the basal ganglia and prefrontal cortical regions, suggesting regionally increased membrane turnover or glial activation in schizophrenia. The basal ganglia are significantly spared from the well-established widespread reduction of NAA in schizophrenia suggesting a regional difference in disease-associated factors affecting NAA. The echo time findings agree with prior reports and suggest microstructural changes cause faster NAA T2 relaxation in hippocampus and frontal white matter in schizophrenia. Separating the effects of medication status and illness chronicity on NAA and choline abnormalities will require further patient-level studies. Metabolite measurement quality was shown to be a critical factor in MRS studies of schizophrenia.

Altered Associations Between Task Performance and Dorsolateral Prefrontal Cortex Activation During Cognitive Control in Schizophrenia.

BACKGROUND: Dysfunctional cognitive control processes are now well understood to be core features of schizophrenia (SZ). A body of work suggests that the dorsolateral prefrontal cortex (DLPFC) plays a critical role in explaining cognitive control disruptions in SZ. Here, we examined relationships between DLPFC activation and drift rate (DR), a model-based performance measure that combines reaction time and accuracy, in people with SZ and healthy control (HC) participants. METHODS: One hundred fifty-one people with recent-onset SZ spectrum disorders and 118 HC participants performed the AX-Continuous Performance Task during functional magnetic resonance imaging scanning. Proactive cognitive control-associated activation was extracted from left and right DLPFC regions of interest. Individual behavior was fit using a drift diffusion model, allowing DR to vary between task conditions. RESULTS: Behaviorally, people with SZ showed significantly lower DRs than HC participants, particularly during high proactive control trial types ("B" trials). Recapitulating previous findings, the SZ group also demonstrated reduced cognitive control-associated DLPFC activation compared with HC participants. Furthermore, significant group differences were also observed in the relationship between left and right DLPFC activation with DR, such that positive relationships between DR and activation were found in HC participants but not in people with SZ. CONCLUSIONS: These results suggest that DLPFC activation is less associated with cognitive control-related behavioral performance enhancements in SZ. Potential mechanisms and implications are discussed.

Spectroscopic meta-analyses reveal novel metabolite profiles across methamphetamine and cocaine substance use disorder.

BACKGROUND: Although proton magnetic resonance spectroscopy (MRS) has been used to study metabolite alterations in stimulant (methamphetamine and cocaine) substance use disorders (SUDs) for over 25 years, data-driven consensus regarding the nature and magnitude of these alterations is lacking. METHOD: In this meta-analysis, we examined associations between SUD and regional metabolites (N-acetyl aspartate (NAA), choline, myo-inositol, creatine, glutamate, and glutamate+glutamine (glx)) in the medial prefrontal cortex (mPFC), frontal white matter (FWM), occipital cortex, and basal ganglia as measured by 1 H-MRS. We also examined moderating effects of MRS acquisition parameters (echo time (TE), field strength), data quality (coefficient of variation (COV)), and demographic/clinical variables. RESULTS: A MEDLINE search revealed 28 articles that met meta-analytic criteria. Significant effects included lower mPFC NAA, higher mPFC myo-inositol, and lower mPFC creatine in SUD relative to people without SUD. mPFC NAA effects were moderated by TE, with larger effects at longer TEs. For choline, although no group effects were observed, effect sizes in the mPFC were related to MRS technical indicators (field strength, COV). No effects of age, sex, primary drug of use (methamphetamine vs. cocaine), duration of use, or duration of abstinence were observed. Evidence for moderating effects of TE and COV may have implications for future MRS studies in SUDs. CONCLUSIONS: The observed metabolite profile in methamphetamine and cocaine SUD (lower NAA and creatine with higher myo-inositol) parallels that observed in Alzheimer's disease and mild cognitive impairment, suggesting these drugs are associated with neurometabolic differences similar to those characterizing these neurodegenerative conditions.

Altered Associations Between Motivated Performance and Frontostriatal Functional Connectivity During Reward Anticipation in Schizophrenia.

BACKGROUND AND HYPOTHESIS: The neuronal mechanisms that underlie deficits in effort cost computation in schizophrenia (SZ) are poorly understood. Given the role of frontostriatal circuits in valence-oriented motivation, we hypothesized that these circuits are either dysfunctional in SZ or do not appropriately predict behavior in SZ when task conditions are difficult and good performance is rewarded. STUDY DESIGN: A total of 52 people with recent onset SZ-spectrum disorders and 48 healthy controls (HCs) performed a 3T fMRI task with 2 valence conditions (rewarded vs neutral) and 2 difficulty conditions. Frontostriatal connectivity was extracted during the cue (anticipatory) phase. Individual behavior was fit using a drift-diffusion model, allowing the performance parameter, drift rate (DR), to vary between task conditions. Three models were examined: A group × condition model of DR, a group × condition model of connectivity, and a regression model of connectivity predicting DR depending on group and condition. STUDY RESULTS: DRs showed the expected positive correlation with accuracy and a negative association with reaction time. The SZ group showed a deficit in DR but did not differ in overall connectivity or show a valence-specific deficit in connectivity. Significant group × valence × difficulty interactions, however, were observed on the relationship between right dorsolateral prefrontal (DLPFC)-striatal connectivity and DR (DLPFC-Caudate: F = 10.92, PFDR = .004; DLPFC-Putamen: F = 5.14, PFDR = .048) driven by more positive relationships between DR and connectivity during cues for the difficult-rewarded condition in HCs compared to SZ. CONCLUSIONS: These findings suggest that frontostriatal connectivity is less predictive of performance in SZ when task difficulty is increased and a reward incentive is applied.

Increased Striatal Presynaptic Dopamine in a Nonhuman Primate Model of Maternal Immune Activation: A Longitudinal Neurodevelopmental Positron Emission Tomography Study With Implications for Schizophrenia.

BACKGROUND: Epidemiological studies suggest that maternal immune activation (MIA) is a significant risk factor for future neurodevelopmental disorders, including schizophrenia (SZ), in offspring. Consistent with findings in SZ research and work in rodent systems, preliminary cross-sectional findings in nonhuman primates suggest that MIA is associated with dopaminergic hyperfunction in young adult offspring. METHODS: In this unique prospective longitudinal study, we used [18F]fluoro-l-m-tyrosine positron emission tomography to examine the developmental time course of striatal presynaptic dopamine synthesis in male rhesus monkeys born to dams (n = 13) injected with a modified form of the inflammatory viral mimic, polyinosinic:polycytidylic acid [poly(I:C)], in the late first trimester. Striatal (caudate, putamen, and nucleus accumbens) dopamine from these animals was compared with that of control offspring born to dams that received saline (n = 10) or no injection (n = 4). Dopamine was measured at 15, 26, 38, and 48 months of age. Prior work with this cohort found decreased prefrontal gray matter volume in MIA offspring versus controls between 6 and 45 months of age. Based on theories of the etiology and development of SZ-related pathology, we hypothesized that there would be a delayed (relative to the gray matter decrease) increase in striatal fluoro-l-m-tyrosine signal in the MIA group versus controls. RESULTS: [18F]fluoro-l-m-tyrosine signal showed developmental increases in both groups in the caudate and putamen. Group comparisons revealed significantly greater caudate dopaminergic signal in the MIA group at 26 months. CONCLUSIONS: These findings are highly relevant to the known pathophysiology of SZ and highlight the translational relevance of the MIA model in understanding mechanisms by which MIA during pregnancy increases risk for later illness in offspring.

Evidence for functional improvement in reward anticipation in recent onset schizophrenia after one year of coordinated specialty care.

BACKGROUND: Motivational impairment associated with deficits in processing the anticipation of future reward is hypothesized to be a cardinal feature of schizophrenia spectrum disorders (SZ). Evidence from short-term follow-up (6-week post-treatment) studies suggests that these deficits may improve or be reversed with treatment, although longer-term outcomes are unknown. Here we examined the one-year trajectory of functional activation in brain circuitry associated with reward anticipation in people with recent onset SZ who participated in coordinated specialty care (CSC) treatment, hypothesizing normalization of brain response mirroring previous short-term findings in first-episode individuals. METHOD: Blood oxygen level-dependent (BOLD) response in the dorsal anterior cingulate cortex, anterior insula, and ventral striatum (VS) associated with reward anticipation during the Incentivized Control Engagement Task (ICE-T) was analyzed in a baseline sample of 49 healthy controls (HCs) and 52 demographically matched people with SZ, with follow-up data available for 35 HCs and 17 people with SZ. RESULTS: In agreement with our hypothesis, significant time × diagnosis interactions were observed across all regions, in which reward anticipation-associated BOLD response increased in SZ to above baseline HC levels at follow-up. Increased VS activation was associated with decreased reality distortion symptoms over the follow-up period. Baseline reward anticipation-associated BOLD response in the right anterior insula was associated with improvement in reality distortion symptoms. CONCLUSIONS: These findings suggest that functional deficits in reward anticipation may be reversed after one year of CSC in recent onset participants with SZ, and that this improvement is associated with reduced positive symptoms in the illness.

Data augmentation with Mixup: Enhancing performance of a functional neuroimaging-based prognostic deep learning classifier in recent onset psychosis.

Although deep learning holds great promise as a prognostic tool in psychiatry, a limitation of the method is that it requires large training sample sizes to achieve replicable accuracy. This is problematic for fMRI datasets as they are typically small due to the considerable time, cost, and resources necessary to obtain them. A recently developed self-supervised learning method called Mixup may help overcome this challenge. In Mixup, the learner combines pairs of training instances to produce a virtual third instance that is a linear combination of the two instances and their labels. This procedure is also well-suited to the coregistered images typically found in fMRI datasets. Here we compared performance of a task fMRI-based deep learner with Mixup vs without Mixup on predicting response to treatment in recent onset psychosis. Whole brain fMRI time series data were extracted from a cognitive control task in 82 patients with recent onset psychosis and used to predict "Improver" (n = 47) vs "Non-Improver" (n = 35) status, with Improver defined as showing a 20 % reduction in total Brief Psychiatric Rating Scale score after 1 year of treatment. Mixup significantly improved performance (accuracy without Mixup: 76.5 % [95 % CI: 75.9-77.1 %]; accuracy with Mixup: 80.1 % [95 % CI: 79.4-80.8 %]). Ablation showed the improvement was due to improvement in both Improvers and Non-Improvers. These results suggest that using Mixup may significantly improve performance and reduce overfitting of fMRI-based prognostic deep learners and may also help overcome the small sample size challenge inherent to many neuroimaging datasets.

Deep Learning in Neuroimaging: Overcoming Challenges With Emerging Approaches.

Deep learning (DL) is of great interest in psychiatry due its potential yet largely untapped ability to utilize multidimensional datasets (such as fMRI data) to predict clinical outcomes. Typical DL methods, however, have strong assumptions, such as large datasets and underlying model opaqueness, that are suitable for natural image prediction problems but not medical imaging. Here we describe three relatively novel DL approaches that may help accelerate its incorporation into mainstream psychiatry research and ultimately bring it into the clinic as a prognostic tool. We first introduce two methods that can reduce the amount of training data required to develop accurate models. These may prove invaluable for fMRI-based DL given the time and monetary expense required to acquire neuroimaging data. These methods are (1) transfer learning - the ability of deep learners to incorporate knowledge learned from one data source (e.g., fMRI data from one site) and apply it toward learning from a second data source (e.g., data from another site), and (2) data augmentation (via Mixup) - a self-supervised learning technique in which "virtual" instances are created. We then discuss explainable artificial intelligence (XAI), i.e., tools that reveal what features (and in what combinations) deep learners use to make decisions. XAI can be used to solve the "black box" criticism common in DL and reveal mechanisms that ultimately produce clinical outcomes. We expect these techniques to greatly enhance the applicability of DL in psychiatric research and help reveal novel mechanisms and potential pathways for therapeutic intervention in mental illness.

Magnetic resonance spectroscopic evidence of increased choline in the dorsolateral prefrontal and visual cortices in recent onset schizophrenia.

A complete characterization of neurometabolite profiles in the dorsolateral prefrontal cortex (DLPFC) in recent onset schizophrenia (SZ) remains elusive. Filling in this knowledge gap is essential in order to better understand how the neurochemistry of this region contributes to SZ pathology. To that end, DLPFC N-acetyl aspartate (NAA), myo-inositol, glutamate, choline, and creatine levels were examined by 3 T magnetic resonance spectroscopy (MRS) in recent onset individuals with SZ (n = 40) and healthy controls (HC) (n = 47). Metabolite levels were also examined in the visual cortex (VC) as a control region. People with SZ showed significantly higher choline in both the DLPFC and VC, but no differences in NAA, myo-inositol, glutamate, or creatine in either region. A trend-level negative correlation was also observed between DLPFC NAA and negative symptoms in SZ. Our results suggest that choline is increased in both the prefrontal and occipital cortices in recent onset SZ, and that DLPFC NAA levels may be inversely related to negative symptoms in the illness. The observed increase in choline-containing compounds in both DLPFC and VC in recent onset SZ could reflect increased membrane remodeling such as occurs in activated microglia and astrocytes in response to neuroinflammation.

Mechanisms underlying dorsolateral prefrontal cortex contributions to cognitive dysfunction in schizophrenia.

Kraepelin, in his early descriptions of schizophrenia (SZ), characterized the illness as having "an orchestra without a conductor." Kraepelin further speculated that this "conductor" was situated in the frontal lobes. Findings from multiple studies over the following decades have clearly implicated pathology of the dorsolateral prefrontal cortex (DLPFC) as playing a central role in the pathophysiology of SZ, particularly with regard to key cognitive features such as deficits in working memory and cognitive control. Following an overview of the cognitive mechanisms associated with DLPFC function and how they are altered in SZ, we review evidence from an array of neuroscientific approaches addressing how these cognitive impairments may reflect the underlying pathophysiology of the illness. Specifically, we present evidence suggesting that alterations of the DLPFC in SZ are evident across a range of spatial and temporal resolutions: from its cellular and molecular architecture, to its gross structural and functional integrity, and from millisecond to longer timescales. We then present an integrative model based upon how microscale changes in neuronal signaling in the DLPFC can influence synchronized patterns of neural activity to produce macrocircuit-level alterations in DLPFC activation that ultimately influence cognition and behavior. We conclude with a discussion of initial efforts aimed at targeting DLPFC function in SZ, the clinical implications of those efforts, and potential avenues for future development.

Improvement in prefrontal thalamic connectivity during the early course of the illness in recent-onset psychosis: a 12-month longitudinal follow-up resting-state fMRI study.

BACKGROUND: Previous research in resting-state functional magnetic resonance imaging (rs-fMRI) has shown a mixed pattern of disrupted thalamocortical connectivity in psychosis. The clinical meaning of these findings and their stability over time remains unclear. We aimed to study thalamocortical connectivity longitudinally over a 1-year period in participants with recent-onset psychosis. METHODS: To this purpose, 129 individuals with recent-onset psychosis and 87 controls were clinically evaluated and scanned using rs-fMRI. Among them, 43 patients and 40 controls were re-scanned and re-evaluated 12 months later. Functional connectivity between the thalamus and the rest of the brain was calculated using a seed to voxel approach, and then compared between groups and correlated with clinical features cross-sectionally and longitudinally. RESULTS: At baseline, participants with recent-onset psychosis showed increased connectivity (compared to controls) between the thalamus and somatosensory and temporal regions (k = 653, T = 5.712), as well as decreased connectivity between the thalamus and left cerebellum and right prefrontal cortex (PFC; k = 201, T = -4.700). Longitudinal analyses revealed increased connectivity over time in recent-onset psychosis (relative to controls) in the right middle frontal gyrus. CONCLUSIONS: Our results support the concept of abnormal thalamic connectivity as a core feature in psychosis. In agreement with a non-degenerative model of illness in which functional changes occur early in development and do not deteriorate over time, no evidence of progressive deterioration of connectivity during early psychosis was observed. Indeed, regionally increased connectivity between thalamus and PFC was observed.

Medial Prefrontal Cortex Glutamate Is Reduced in Schizophrenia and Moderated by Measurement Quality: A Meta-analysis of Proton Magnetic Resonance Spectroscopy Studies.

BACKGROUND: Magnetic resonance spectroscopy studies measuring brain glutamate separately from glutamine are helping elucidate schizophrenia pathophysiology. An expanded literature and improved methodologies motivate an updated meta-analysis examining effects of measurement quality and other moderating factors in characterizing abnormal glutamate levels in schizophrenia. METHODS: Searching previous meta-analyses and the MEDLINE database identified 83 proton magnetic resonance spectroscopy datasets published through March 25, 2020. Three quality metrics were extracted-Cramér-Rao lower bound (CRLB), line width, and coefficient of variation. Pooled effect sizes (Hedges' g) were calculated with random-effects, inverse variance-weighted models. Moderator analyses were conducted using quality metrics, field strength, echo time, medication, age, and stage of illness. RESULTS: Across 36 datasets (2086 participants), medial prefrontal cortex glutamate was significantly reduced in patients (g = -0.19, confidence interval [CI] = -0.07 to -0.32). CRLB and coefficient of variation quality subgroups significantly moderated this effect. Glutamate was significantly more reduced in studies with lower CRLB or coefficient of variation (g = -0.44, CI = -0.29 to -0.60, and g = -0.43, CI = -0.29 to -0.57, respectively). Studies using echo time ≤20 ms also showed significantly greater reduction in glutamate (g = -0.41, CI = -0.26 to -0.55). Across 11 hippocampal datasets, group differences and moderator effects were nonsignificant. Group effects in thalamus and dorsolateral prefrontal cortex were also nonsignificant. CONCLUSIONS: High-quality measurements reveal consistently reduced medial prefrontal cortex glutamate in schizophrenia. Stricter CRLB criteria and reduced nuisance variance may increase the sensitivity of future studies examining additional regions and the pathophysiological significance of abnormal glutamate levels in schizophrenia.

Schizophrenia and bipolar disorder are associated with opposite brain reward anticipation-associated response.

Blunted and exaggerated neuronal response to rewards are hypothesized to be core features of schizophrenia spectrum disorders (SZ) and bipolar disorder (BD), respectively. Nonetheless, direct tests of this hypothesis, in which response between SZ and BD is compared in the same study, are lacking. Here we examined the functional correlates of reward processing during the Incentivized Control Engagement Task (ICE-T) using 3T fMRI. Reward-associated activation was examined in 49 healthy controls (HCs), 52 recent-onset individuals with SZ, and 22 recent-onset individuals with Type I BD using anterior cingulate (ACC), anterior insula, and ventral striatal regions of interest. Significant group X reward condition (neutral vs. reward) interactions were observed during reward anticipation in the dorsal ACC (F(2,120) = 4.21, P = 0.017) and right insula (F(2,120) = 4.77, P = 0.010). The ACC interaction was driven by relatively higher activation in the BD group vs. HCs (P = 0.007) and vs. individuals with SZ (P = 0.010). The insula interaction was driven by reduced activation in the SZ group relative to HCs (P = 0.018) and vs. people with BD (P = 0.008). A composite of reward anticipation-associated response across all associated ROIs also differed significantly by diagnosis (F(1,120) = 5.59, P = 0.02), BD > HC > SZ. No effects of group or group X reward interactions were observed during reward feedback. These results suggest that people with SZ and BD have opposite patterns of activation associated with reward anticipation but not reward receipt. Implications of these findings in regard to Research Domain Criteria-based classification of illness and the neurobiology of reward in psychosis are discussed.

What transcranial direct current stimulation intensity is best for cognitive enhancement?

In a recent study published in The Journal of Neurophysiology, Ehrardt et al. (Ehrhardt SE, Filmer HL, Wards Y, Mattingley JB, Dux PE. J Neurophysiol 125: 385-397, 2021) report that moderate intensity (1 mA/25 cm2) transcranial direct current stimulation (tDCS) is optimal for improving performance on a stimulus-response matching task, as opposed to a lower 0.7 mA/25 cm2 or higher 2 mA/25 cm2 dose. This result suggests that behavioral effects of tDCS do not follow a linear dose-response curve. Potential neurobiological and neurocognitive implications of these findings, as well as suggested directions for future research, are discussed.

Cognitive Subtyping in Schizophrenia: A Latent Profile Analysis.

Cognitive dysfunction is a core feature of schizophrenia. The subtyping of cognitive performance in schizophrenia may aid the refinement of disease heterogeneity. The literature on cognitive subtyping in schizophrenia, however, is limited by variable methodologies and neuropsychological tasks, lack of validation, and paucity of studies examining longitudinal stability of profiles. It is also unclear if cognitive profiles represent a single linear severity continuum or unique cognitive subtypes. Cognitive performance measured with the Brief Assessment of Cognition in Schizophrenia was analyzed in schizophrenia patients (n = 767). Healthy controls (n = 1012) were included as reference group. Latent profile analysis was performed in a schizophrenia discovery cohort (n = 659) and replicated in an independent cohort (n = 108). Longitudinal stability of cognitive profiles was evaluated with latent transition analysis in a 10-week follow-up cohort. Confirmatory factor analysis (CFA) was carried out to investigate if cognitive profiles represent a unidimensional structure. A 4-profile solution was obtained from the discovery cohort and replicated in an independent cohort. It comprised of a "less-impaired" cognitive subtype, 2 subtypes with "intermediate cognitive impairment" differentiated by executive function performance, and a "globally impaired" cognitive subtype. This solution showed relative stability across time. CFA revealed that cognitive profiles are better explained by distinct meaningful profiles than a severity linear continuum. Associations between profiles and negative symptoms were observed. The subtyping of schizophrenia patients based on cognitive performance and its associations with symptomatology may aid phenotype refinement, mapping of specific biological mechanisms, and tailored clinical treatments.

Comparing machine and deep learning-based algorithms for prediction of clinical improvement in psychosis with functional magnetic resonance imaging.

Previous work using logistic regression suggests that cognitive control-related frontoparietal activation in early psychosis can predict symptomatic improvement after 1 year of coordinated specialty care with 66% accuracy. Here, we evaluated the ability of six machine learning (ML) algorithms and deep learning (DL) to predict "Improver" status (>20% improvement on Brief Psychiatric Rating Scale [BPRS] total score at 1-year follow-up vs. baseline) and continuous change in BPRS score using the same functional magnetic resonance imaging-based features (frontoparietal activations during the AX-continuous performance task) in the same sample (individuals with either schizophrenia (n = 65, 49M/16F, mean age 20.8 years) or Type I bipolar disorder (n = 17, 9M/8F, mean age 21.6 years)). 138 healthy controls were included as a reference group. "Shallow" ML methods included Naive Bayes, support vector machine, K Star, AdaBoost, J48 decision tree, and random forest. DL included an explainable artificial intelligence (XAI) procedure for understanding results. The best overall performances (70% accuracy for the binary outcome and root mean square error = 9.47 for the continuous outcome) were achieved using DL. XAI revealed left DLPFC activation was the strongest feature used to make binary classification decisions, with a classification activation threshold (adjusted beta = .017) intermediate to the healthy control mean (adjusted beta = .15, 95% CI = -0.02 to 0.31) and patient mean (adjusted beta = -.13, 95% CI = -0.37 to 0.11). Our results suggest DL is more powerful than shallow ML methods for predicting symptomatic improvement. The left DLPFC may be a functional target for future biomarker development as its activation was particularly important for predicting improvement.