The finding does not mean that parent questionnaires are unreliable or that children’s reports should replace them. Instead, the analysis indicates that asking a child directly may reveal internal and peer-related difficulties that are less visible outside school, while parents may be better positioned to notice conduct problems occurring at home. When child reports were added after parent reports, the statistical model improved substantially more than when parent reports were added after child reports. The combined model explained 21.0% of the variance in NSSI status, but neither informant alone approached definitive predictive accuracy. These results support a complementary, multi-informant approach rather than a contest over which perspective is universally superior.

Researchers collected paired questionnaires in December 2021 from students in grades four through six at eight primary schools in Dongli District, Tianjin. Children completed self-report forms in class with trained teachers and research assistants available for clarification and to preserve privacy, while parents completed questionnaires at home and returned them in sealed envelopes. Children provided assent and parents gave written consent. The investigators excluded incomplete or invalid questionnaires and cases missing either the child or parent Strengths and Difficulties Questionnaire, reducing the initial group of 4,206 students to 1,923 matched parent-child dyads. A complete-case sensitivity analysis produced results consistent with the primary analysis. Because the sampling was by convenience and the participating schools were in one district, the findings describe this particular sample rather than all children in Tianjin or China.

NSSI was measured with the 16-item short version of the Deliberate Self-Harm Inventory, which asks about intentional self-injury during the preceding six months. Each item ranges from never to three or more times, and endorsement of at least one behaviour placed a child in the NSSI group. The most common reported method was carving words, pictures, designs or other marks into the skin, reported by 65.5% of the NSSI group. Cutting the wrist, arms or another body area was reported by 33.9%, and scratching severely enough to cause scarring or bleeding by 30.4%. Biting until the skin broke was reported by 21.7%, while 3.1% reported self-harm that led to hospitalisation or medical treatment. The researchers note that suicidal intent was not assessed separately, so the classification was operational rather than a clinical confirmation that every episode occurred without suicidal intent.

Emotional and behavioural characteristics were assessed with child and parent versions of the Strengths and Difficulties Questionnaire, or SDQ. Its five subscales cover emotional symptoms, conduct problems, hyperactivity and inattention, peer interaction problems and prosocial behaviour. In unadjusted comparisons, children in the NSSI group had higher scores on all four problem subscales in both informants’ reports and lower prosocial-behaviour scores. These group differences do not show that any feature preceded or caused the self-injury, because the survey measured children and behaviours at the same time. The SDQ is a screening instrument rather than a diagnostic assessment, and the study did not test whether its categories corresponded to diagnosed psychiatric disorders.

Multivariable logistic regression, adjusted for age and gender, produced a more nuanced pattern after all five SDQ subscales were entered simultaneously. With standardised scores, each one-standard-deviation increase in parent-reported emotional symptoms was associated with 1.26 times the odds of NSSI, compared with 1.66 for child-reported emotional symptoms. The corresponding odds ratios were 1.25 for parent-reported hyperactivity and 1.39 for child-reported hyperactivity. Child-reported peer interaction problems were associated with NSSI, with an odds ratio of 1.24, whereas parent-reported peer interaction problems were not significant after adjustment. Parent-reported conduct problems were also associated with NSSI, with an odds ratio of 1.18, whereas child-reported conduct problems and prosocial behaviour were not significant after adjustment. These are associations in odds, not changes in an individual child’s probability caused by a symptom score, and the study cannot establish which problems preceded the behaviour.

Formal joint models were then used to compare coefficients for the same SDQ subscale when both child and parent reports were included simultaneously. Wald tests indicated that the associations were statistically significantly larger for child-reported emotional symptoms, hyperactivity and peer interaction problems, with false-discovery-rate-adjusted p-values below 0.001 in each case. The investigators did not find significant differences for conduct problems or prosocial behaviour. Parent-reported peer interaction problems differed between the NSSI and non-NSSI groups in simple comparisons but were no longer significant after adjustment, apparently because their variation overlapped with the other SDQ subscales included in the model. The results therefore describe differences in measured statistical associations; they do not prove that one informant is inherently more accurate, nor do they establish a mechanism for the differences.

Child reports also improved discrimination between children classified with and without NSSI. The child-report model produced an area under the receiver operating characteristic curve, or AUC, of 0.753 with a 95% confidence interval of 0.732 to 0.779, indicating moderate discrimination. The parent-report model yielded an AUC of 0.659, with a 95% confidence interval of 0.638 to 0.694. The difference was statistically significant in a DeLong test for correlated curves, with Z equal to -7.75 and p below 0.001. Nevertheless, an AUC of 0.753 should not be presented as high accuracy or as proof that child reports are more truthful. The models were evaluated in the same cross-sectional sample, with no independent screening validation, so their performance cannot automatically be transferred to another school, region or clinical setting.

Several limitations constrain the interpretation and likely contributed to the apparently stronger child-report results. NSSI and the child SDQ came from the same respondent, creating possible shared-method variance: a child who reports both emotional distress and self-injury may therefore show a stronger association than a parent’s report can reproduce. The DSHI-s was developed for adolescents, and the self-report SDQ is officially recommended for children aged 11 and above, although previous research has reported acceptable psychometric properties in children as young as 8 to 10. Additional validation is therefore needed for these instruments among Chinese children aged 9 to 12. The study also did not independently verify self-injury through clinical interview or another informant, and the functions of the behaviour, such as emotion regulation or social signalling, were not assessed.

The design also prevents conclusions about temporal sequence, causation, prevalence across all Chinese children or the reasons children self-injure. The models adjusted only for age and gender, leaving possible influences such as socioeconomic conditions, parental mental health, academic stress, peer victimisation and parenting style unmeasured. More than half of the initially enrolled students were excluded because complete matched data were unavailable, which may also have introduced selection effects. The study’s clearest conclusion is that, in this particular sample of 9- to 12-year-olds, child and parent questionnaires supplied overlapping but non-identical information: children contributed particularly strong signals about internal distress, hyperactivity and peer problems linked to concurrent self-reported self-injury, while parents contributed additional information that was modest but measurable when added after child reports. Any use of these findings in screening would require further validation rather than treating either report as a standalone diagnostic test.

Subject of Research: Comparison of child and parent reports of emotional and behavioural correlates of self-reported non-suicidal self-injury among preadolescents in Tianjin

Article Title: Emotional and behavioural problems and non-suicidal self-injury in preadolescent children: complementary insights from child and parent reports

Article References: Yang, T., Li, M., Yu, H., Hou, X., Yan, G., Li, B., Ma, H., Sun, L., Yin, H., & Xu, G. (2026). Emotional and behavioural problems and non-suicidal self-injury in preadolescent children: complementary insights from child and parent reports. Frontiers in Psychiatry, 17, Article 1973448. https://doi.org/10.3389/fpsyt.2026.1973448

Image Credits: AI Generated

DOI: 10.3389/fpsyt.2026.1973448

Keywords: preadolescents, self-injury, child self-report, parent report, emotional symptoms, hyperactivity, peer problems, school-based survey, Tianjin, mental health, Emotional, behavioural

The research, led by Elijah Agoalikum of Sichuan Provincial People’s Hospital and the University of Electronic Science and Technology of China, together with colleagues including senior author Michael Maes, used resting-state functional magnetic resonance imaging to compare people with schizophrenia, people with ADHD, and healthy controls. Rather than computing one static connectivity map per participant, the team employed a technique known as dynamic functional network connectivity, or dFNC. The method slices continuous scanning data into overlapping temporal windows and estimates how strongly a set of intrinsic brain networks are synchronized with one another within each window. A clustering algorithm then sorts these moment-to-moment connectivity patterns into a small number of recurring configurations, or states, that the brain revisits again and again across the scan.

In this study, the sliding-window analysis paired with k-means clustering revealed four distinct, recurring connectivity states. Each state represents a characteristic pattern of coordination among major intrinsic networks, including the default-mode network, which is active during introspection and self-referential thought; the fronto-parietal network, which supports cognitive control and working memory; the sensorimotor, visual, and auditory networks, which handle perception and action; the subcortical network, encompassing deep structures such as the basal ganglia and thalamus; and the cerebellar network, long associated with motor coordination but increasingly implicated in cognition and emotion. Once the four states were defined, the researchers quantified how each group used them, measuring the mean dwell time spent in each state, the fraction of total scan time occupied by each state, and the frequency of transitions between states.

The results divided the two disorders along a clear axis of network dynamics. Participants with schizophrenia spent significantly more time, both in average dwell duration and in overall fraction of time, in states characterized by reduced integration between networks and greater segregation, compared with both the ADHD group and healthy controls. In other words, the schizophrenic brain appeared to get stuck in configurations in which major systems operate in relative isolation from one another. This finding resonates with a long-standing hypothesis in schizophrenia research that the illness involves a weakening of long-range connectivity, particularly between the prefrontal cortex and distributed association networks, which may underlie difficulties in coordinating thought, emotion, and behavior.

ADHD presented a different picture. Rather than showing the pronounced entrenchment in segregated states seen in schizophrenia, participants with ADHD spent more time in states reflecting moderate connectivity patterns. The authors interpret this as evidence of altered but more flexible network dynamics relative to schizophrenia. The ADHD brain, in this framing, is not locked into any single maladaptive configuration; instead, its pattern of state occupancy is shifted toward intermediate regimes, suggesting a subtler disruption in how the brain balances stability and flexibility. This distinction is clinically meaningful, because cognitive models of ADHD have long emphasized variability and instability of performance over time, and the dFNC findings provide a neural correlate of that temporal unevenness.

Beyond state occupancy, the study uncovered disorder-specific differences in the strength of connectivity within the states themselves. Schizophrenia showed increased connectivity between cerebellar, cognitive control, and sensory networks compared with ADHD. This may seem paradoxical alongside the finding of reduced integration, but it illustrates an important principle of network neuroscience: connectivity alterations are not uniformly stronger or weaker across the brain. Instead, particular pairs of networks can become hyperconnected while others drift apart, producing a reorganized architecture rather than a simple global deficit. The cerebellum, once viewed as a purely motor structure, has emerged in recent psychiatric imaging as a hub with reciprocal connections to prefrontal and association cortices, and its altered coupling with control and sensory systems in schizophrenia may reflect the wide-ranging cognitive and perceptual disturbances characteristic of the illness.

Perhaps the most striking results are those that link temporal dynamics to lived experience. The researchers correlated the dFNC metrics with clinical symptom scores and found that in schizophrenia, greater mean dwell time and fraction of time in one particular state, designated state 2, were associated with physical anhedonia, the diminished capacity to experience pleasure from tactile, sensory, and bodily sources. Meanwhile, hyperactivity symptoms in the schizophrenia group were associated with the fraction of time spent in state 1. These correlations suggest that the amount of time the brain spends in specific connectivity configurations is not an abstract mathematical curiosity but maps onto concrete symptom dimensions. A patient who lingers longer in a segregated network state may, in effect, be carrying the neural signature of a diminished capacity for sensory pleasure.

This symptom-level mapping carries implications for how psychiatric illness is conceptualized and measured. The authors situate their findings within frameworks, such as the Research Domain Criteria initiative, that view psychopathology as arising from graded disruptions in interacting neural systems rather than from discrete diagnostic categories. In that view, schizophrenia and ADHD are not sealed boxes but overlapping regions in a multidimensional space of brain dynamics, and the same temporal metric, such as dwell time in a segregated state, may index different symptom dimensions in different populations. The study supports this transdiagnostic perspective by showing both shared instability in network dynamics across the two disorders and clearly disorder-specific patterns of state occupancy and connectivity.

The technical pipeline behind these conclusions is worth appreciating. After standard preprocessing of the resting-state fMRI data, the team used group independent component analysis to decompose each participant’s brain activity into spatially independent component networks, separating sources such as the default-mode, fronto-parietal, subcortical, sensorimotor, visual, auditory, and cerebellar networks while suppressing noise from head motion, respiration, and cardiac pulsation. Motion was carefully controlled, with framewise displacement used to account for the fact that participant movement, which itself can differ between clinical groups, can artificially inflate connectivity estimates. Sliding windows were then applied to the component time courses, k-means clustering grouped the windowed connectivity matrices into the four states, and statistical comparisons of dwell time, fraction of time, and transition probabilities between groups were corrected for multiple comparisons using false discovery rate procedures, a standard safeguard against spurious findings when many tests are performed simultaneously.

The authors propose that these dynamic connectivity features could serve as potential biomarkers for symptom dimensions in schizophrenia and ADHD. A biomarker of this kind would not diagnose a disorder in isolation, but it could help clinicians and researchers track illness mechanisms, stratify patients by underlying neural dynamics rather than surface symptoms, and monitor whether treatments are shifting the brain toward healthier patterns of state occupancy. The study also carries a broader message about the moving brain. Static averages, the findings imply, can obscure the very features that distinguish one condition from another: not where the brain goes, but how long it stays and how readily it leaves. As dynamic functional network connectivity matures and larger datasets accumulate, the fleeting choreography of neural networks may become one of psychiatry’s most informative windows into mental illness, revealing that what separates conditions such as schizophrenia and ADHD is not only which networks fail, but how the brain’s moment-to-moment dance among them goes awry.

Subject of Research: Dynamic functional network connectivity alterations in schizophrenia and ADHD measured with resting-state fMRI

Article Title: Network state transitions and connectivity alterations in attention deficit/hyperactivity disorder and schizophrenia: a dynamic fMRI study

Article References: Agoalikum, E., Klugah-Brown, B., Wu, H., Pacheco-Luna, L., Carvalho, A. F., Stoyanov, D., & Maes, M. (2026). Network state transitions and connectivity alterations in attention deficit/hyperactivity disorder and schizophrenia: a dynamic fMRI study. BMC Psychiatry. https://doi.org/10.1186/s12888-026-08577-x

Image Credits: AI Generated

DOI: 10.1186/s12888-026-08577-x

Keywords: dynamic functional network connectivity, schizophrenia, ADHD, resting-state fMRI, brain networks, default-mode network, cerebellum, anhedonia, hyperactivity, state transitions, psychiatry, biomarkers

The research team, led by Esra Hoşoğlu of Necmettin Erbakan University’s Meram Faculty of Medicine, recruited children aged 2 to 6 years from a child and adolescent psychiatry outpatient clinic and from family health centers in Konya. Every ASD diagnosis was confirmed by experienced child and adolescent psychiatrists using DSM-5 criteria. To keep the comparison clean, the researchers excluded children taking appetite-altering medications and those with chronic conditions such as diabetes, cerebral palsy, epilepsy, or gastrointestinal disease, since these factors can independently disrupt eating. Parents whose children had a sibling or parent with autism were also barred from the control group, and families in which a parent was currently receiving psychiatric treatment were excluded from both arms of the study.

The measurement toolkit was comprehensive. Parents completed the Children’s Eating Behavior Questionnaire, a validated 35-item instrument that captures both food-approach tendencies, such as food responsiveness and emotional overeating, and food-avoidant tendencies, such as satiety responsiveness, slowness in eating, emotional undereating, and food fussiness. Child autism severity was rated with the Childhood Autism Rating Scale, while developmental level was gauged using the Ankara Developmental Screening Inventory, which covers language-cognitive, fine motor, gross motor, and social domains. Behavioral problems were assessed with the Aberrant Behavior Checklist, and parental mental health was screened with the Brief Symptom Inventory, a 53-item measure spanning anxiety, depression, somatization, hostility, and negative self-concept.

The statistical strategy was deliberately conservative. Because the mothers and fathers of autistic children were significantly older than the control parents, the researchers used multivariate analysis of covariance to adjust for these baseline imbalances. The group effect on eating behaviors remained stubbornly non-significant: Pillai’s Trace was 0.064 with a p-value of 0.688 in the fully adjusted model, and adding overall developmental level as a further covariate did not change the picture. The team then applied the Benjamini–Hochberg false discovery rate procedure to all within-group analyses, a mathematical safeguard against the false positives that inevitably accumulate when dozens of correlations are tested simultaneously. The study was powered to detect only medium-to-large between-group effects, meaning subtle differences could have slipped through undetected.

Within the autistic group itself, however, the data hinted at meaningful patterns, even if most of them dissolved under strict correction. The most robust finding was a positive correlation between hyperactivity and noncompliance, measured on the Aberrant Behavior Checklist, and satiety responsiveness, with a Spearman coefficient of 0.449 that survived false discovery rate correction in the primary analysis with an adjusted p-value of 0.022. In plain terms, children whose parents described them as more restless and defiant were also reported to feel full more quickly and disengage from meals sooner. The researchers speculate that highly distractible children may be pulled away from their plates by environmental stimuli, leading caregivers to interpret this disengagement as early fullness, a mechanism consistent with prior work on attention-deficit hyperactivity symptoms and mealtime behavior.

Nominally significant, but statistically fragile, associations also emerged involving developmental milestones. Fine motor skills showed positive nominal links to food responsiveness, enjoyment of food, and desire to drink, and a negative nominal link to slowness in eating, with standardized coefficients ranging from negative 0.289 to 0.404. Cognitive-language development nominally predicted lower emotional overeating and higher satiety responsiveness. These relationships make biological sense: self-feeding is a fine motor milestone, and children with better language may communicate hunger and fullness cues more effectively. But after false discovery rate correction across the eight regression models, not a single one of these associations retained statistical significance, with all corrected q-values landing at or above 0.292. The authors are careful to frame these results as exploratory and hypothesis-generating rather than established findings.

Parental mental health, a factor the study was specifically designed to probe, told a similar story of tantalizing signals that failed to solidify. Maternal and paternal depressive symptoms both correlated nominally with children’s emotional overeating, and paternal depression and somatization showed nominal associations with slower eating, yet none withstood correction. The researchers caution that this should not be read as evidence that parental psychology is irrelevant. Prior studies have linked maternal anxiety to restrictive feeding and maternal depression to pressure to eat, and in autism families specifically, higher parental anxiety and depression have been tied to more problematic feeding behaviors and controlling strategies. The present study measured only general psychiatric symptoms, not mealtime-specific emotions or feeding practices, which may be the more proximal levers through which parental distress shapes a child’s relationship with food.

Why did the autistic preschoolers look so much like their typically developing peers? The authors offer several converging explanations. First, early childhood is precisely the developmental window when picky eating peaks in nearly all children, regardless of diagnosis. In typically developing youngsters these behaviors typically fade, whereas in autism, sensory sensitivities and insistence on sameness can cause them to persist and harden. The trajectories of the two groups may simply not have diverged yet, a hypothesis supported by an earlier Italian study that found no eating differences between autistic and non-autistic children aged 2 to 5.9 years but clear differences by ages 6 to 9.5. Second, by excluding children with chronic illness and medication use, the team may have inadvertently filtered out the autistic children with the most severe and complex feeding problems. Third, cultural context matters: research in Türkiye suggests that parents there often continue feeding children well beyond the age typical in Western samples, perceive higher body weight as a sign of health, and favor overprotective feeding styles, all of which could compress group differences on a parent-report instrument.

The study’s limitations deserve honest weight. The sample of 93 children is modest, the post hoc power analysis indicated that only effects of Cohen’s d of 0.59 or larger could be reliably detected, and every measure, from eating behavior to parental symptoms, relied on parent report rather than direct mealtime observation. Notably, previous research comparing questionnaire scores with observed mealtime behavior in autistic children found almost no correspondence, raising the possibility that parental perception, stress, and cultural expectations color the data. The Ankara Developmental Screening Inventory also lacks validated norms for autistic populations, and the cross-sectional design makes it impossible to determine whether behavioral difficulties drive eating problems or vice versa. The researchers themselves emphasize that their findings require replication in larger, longitudinal studies using standardized developmental assessments and multi-method measurement.

Still, the work carries real clinical significance precisely because of when it was conducted. The preschool period is a critical window for establishing eating behaviors that persist from childhood through adolescence into adulthood, and eating problems that emerge at this stage can seed long-term nutritional and weight-related difficulties, whether through the obesity risk associated with food-approach behaviors or the underweight and inadequate intake linked to food avoidance. If the divergence between autistic and typically developing children’s eating patterns genuinely emerges later, then early intervention may have a wide window of opportunity before feeding difficulties consolidate into something more resistant to change. Conversely, if subtle differences were missed for want of statistical power, the study serves as a call for adequately powered longitudinal cohorts that track autistic children from toddlerhood through school age, pairing parent questionnaires with observational mealtime data, sensory processing assessments, and measures of behavioral rigidity. Either way, the message for clinicians is nuanced: a preschool diagnosis of autism does not automatically forecast the severe eating problems so often described in older children, but the hyperactivity-eating connection, however tentative, hints that managing early behavioral dysregulation might protect mealtimes before problems take root.

Subject of Research: Eating behaviors in preschool children with autism spectrum disorder and their developmental, behavioral, and parental correlates

Article Title: Eating behaviors in preschoolers with autism spectrum disorders: associations with developmental and behavioral features and parental mental health

Article References: Hoşoğlu, E., Bilgiç, A., Akça, Ö. F., & Tuzcu, M. S. (2026). Eating behaviors in preschoolers with autism spectrum disorders: associations with developmental and behavioral features and parental mental health. Frontiers in Psychiatry, 17, Article 1902816. https://doi.org/10.3389/fpsyt.2026.1902816

Image Credits: AI Generated

DOI: 10.3389/fpsyt.2026.1902816

Keywords: autism spectrum disorder, preschool children, eating behaviors, food selectivity, satiety responsiveness, parental mental health, hyperactivity, developmental assessment, case-control study, picky eating, Frontiers in Psychiatry, Eating

Selenium is an essential micronutrient that the human body incorporates into a family of proteins known as selenoproteins. These proteins perform a remarkable range of biochemical duties, including antioxidant defense, thyroid hormone metabolism, and immune regulation. The brain is particularly dependent on this system because neural tissue consumes a disproportionate share of the body’s oxygen and generates a steady stream of reactive oxygen species as a byproduct of its relentless metabolic activity. When selenium is scarce, the antioxidant capacity of developing neurons may be compromised at precisely the moment when the brain is wiring itself at its most furious pace. Deficiency has also been linked in animal studies to alterations in neurotransmitter systems, including dopaminergic and serotonergic pathways, which are central to attention regulation and impulse control. This biological plausibility is what motivated the research team to ask whether a mother’s selenium status during pregnancy leaves a measurable trace in her child’s behavior years later.

The study drew on a well-established cohort in Bhaktapur, a city in the Kathmandu Valley where the research group has conducted nutrition studies for more than a decade. In total, 720 mother-infant dyads were enrolled, and maternal plasma selenium concentrations were measured in blood samples taken early in pregnancy. This timing matters. The first trimester and early second trimester are periods of rapid neural tube formation, neuronal proliferation, and the beginnings of synaptic architecture, so any nutritional insult during this window could theoretically have outsized consequences. The researchers then waited, quite literally, for the children to grow. When the children reached four years of age, their mental health was assessed using a parent-reported questionnaire called the Strengths and Difficulties Questionnaire, or SDQ, in its Nepalese version.

The SDQ is one of the most widely used brief behavioral screening instruments in the world. It asks parents to rate their child on a series of attributes that cluster into five subscales: emotional symptoms, conduct problems, hyperactivity-inattention, peer relationship problems, and prosocial behavior. Because the questionnaire had not previously been validated in this specific Nepalese preschool population, the team took the methodologically important step of testing whether the original five-factor structure actually held up in their data. Using confirmatory factor analysis, a statistical technique that evaluates how well observed responses map onto hypothesized underlying dimensions, they found that the five-factor model fit the data well after some modifications. The comparative fit index reached 0.967, comfortably above conventional thresholds, while the root mean square error of approximation was 0.059, indicating acceptable approximation error. In practical terms, this means the Nepalese version of the parent-rated SDQ can be considered a psychometrically sound tool for assessing preschool mental health in this setting, a contribution that will benefit future research in the region regardless of what the selenium analysis showed.

The average maternal plasma selenium concentration in the cohort was 74.7 micrograms per liter, with a standard deviation of 10.2 micrograms per liter. This figure sits in an interesting zone. It is not low enough to constitute overt deficiency by most clinical standards, but it is well below the concentrations typically observed in selenium-replete populations in North America and parts of Europe, where means can exceed 100 micrograms per liter. Nepal’s soil is known to be variable in selenium content, and the predominantly cereal-based local diet does not necessarily compensate for that variability. The researchers therefore had genuine reason to suspect that a meaningful fraction of the pregnant women in the cohort might be experiencing suboptimal selenium status during a critical developmental window.

To analyze the relationship between prenatal selenium and child behavior, the team used generalized linear models, examining maternal plasma selenium both as a continuous variable and in categorized forms, including tertiles that divided the sample into thirds of exposure. The children’s SDQ subscale scores served as the outcomes. The headline finding emerged from the tertile analysis: children whose mothers fell in the lowest selenium tertile during pregnancy had significantly higher hyperactivity-inattention scores at age four than children whose mothers were in the middle tertile. The adjusted regression coefficient was 0.329, with a 95 percent confidence interval running from 0.023 to 0.636, meaning the difference was statistically significant but modest in magnitude, and the confidence interval only narrowly excluded the null value of zero.

Crucially, however, the pattern did not behave the way a classic dose-response relationship would. If selenium deficiency were driving hyperactive behavior in a straightforward manner, one might expect the lowest tertile to differ from the middle tertile, and the middle tertile to differ from the highest, producing a graded gradient across the exposure distribution. Instead, the association appeared only when the lowest and middle tertiles were compared, and no corresponding difference emerged between the middle and highest groups. The analysis of selenium as a continuous variable likewise did not yield significant associations. The authors are appropriately cautious on this point, describing their evidence as limited and noting that the isolated finding lacks the clear exposure-response pattern that would strengthen a causal interpretation. It is possible that a threshold effect exists, in which only selenium status below a certain level matters, but it is equally possible that the result reflects chance, residual confounding, or the many comparisons inherent in examining multiple subscales across multiple exposure parameterizations.

Several other limitations deserve attention. The SDQ was completed by parents, which introduces the possibility of reporting bias, although the validation work strengthens confidence that the instrument measures coherent behavioral dimensions in this population. Plasma selenium reflects relatively recent intake and is influenced by the proteins that transport the element, notably selenoprotein P, so a single early-pregnancy measurement may not perfectly capture selenium status across the full gestational period. The Bhaktapur cohort, while valuable, is drawn from one urbanizing South Asian community, and its selenium distribution differs from that of populations where deficiency is rare, which limits generalizability in both directions. Confounding by overall nutritional quality is also a real concern, since women with low selenium may differ in other dietary and socioeconomic respects that independently influence child development, even though the models adjusted for available covariates.

Still, the study’s strengths are considerable. The prospective design, in which exposure was measured in pregnancy and outcome years later, removes much of the reverse-causation ambiguity that plagues cross-sectional nutrition research. The sample size of 720 dyads is respectable for a birth cohort study of this kind, and the dual contribution of a validated behavioral instrument and a novel exposure-outcome analysis makes the paper useful on two fronts. For selenium biology, the findings add a tentative data point to a small but growing literature linking maternal micronutrient status to child neurobehavioral outcomes, a field that has historically focused more heavily on iron, iodine, and folate. For public health in South Asia, the results raise the possibility that selenium, a mineral rarely included in prenatal supplementation programs, deserves closer scrutiny. The authors themselves conclude that confirmation in further studies is required, and that is the right note: one modest association, in one cohort, at one age, is a signal worth chasing rather than a conclusion worth announcing. Future work with repeated selenium measurements, biomarkers of selenium-dependent enzyme activity, and follow-up into school age will be needed to determine whether the trace element truly shapes the developing mind, or whether this finding dissolves under the weight of replication.

Subject of Research: The association between prenatal maternal selenium concentration and mental health outcomes in four-year-old children in a Nepalese mother-child cohort study.

Article Title: The association between prenatal maternal selenium concentration and mental health in early childhood, a mother-child cohort study

Article References: Ranjitkar, S., Kvestad, I., Chandyo, R. K., Bakken, K. S., Ulak, M., Strand, T. A., Huber, S., Averina, M., & Hysing, M. (2026). The association between prenatal maternal selenium concentration and mental health in early childhood, a mother-child cohort study. PLOS One, 21(10), e0360527. https://doi.org/10.1371/journal.pone.0360527

Image Credits: AI Generated

DOI: 10.1371/journal.pone.0360527

Keywords: selenium, prenatal nutrition, child mental health, hyperactivity, inattention, Nepal, birth cohort, Strengths and Difficulties Questionnaire, micronutrients, neurodevelopment, pregnancy, PLOS One

Psychological resilience, the capacity to maintain or recover healthy functioning in the face of stress and adversity, has long been a central concern of developmental psychology. Children with low resilience-related capacity are at elevated risk for emotional and behavioral difficulties as they grow, and researchers have spent decades trying to identify which factors distinguish children who bend from those who break. What has often been missing, however, is a way to translate that knowledge into practical, scalable screening tools that schools can actually use. The new study set out to fill that gap by asking a deceptively simple question: can machine learning models, trained on ordinary questionnaire data, reliably flag students with relatively low resilience before problems escalate?

The researchers grounded their work in the Promotive and Protective Factors and Processes framework, a widely used model that organizes the influences on resilience into internal characteristics of the child and external resources in the environment. Eleven predictors were measured, spanning both categories, using self-report questionnaires completed by the students themselves and by one of their parents. The measures drew on well-validated instruments, including the Strengths and Difficulties Questionnaire, which captures emotional symptoms, hyperactivity, conduct problems, peer problems, and prosocial behavior, as well as scales assessing temperament, perceived social support, and school climate. This longitudinal design, with data collected at two time points, allowed the team to test whether factors measured early could predict resilience-related capacity months later, a stronger test than a single snapshot.

On the modeling side, the team cast a wide net. Eight algorithms were trained and compared: logistic regression, decision tree, random forest, k-nearest neighbors, support vector machine, AdaBoost, LightGBM, and CatBoost. The latter two belong to the gradient boosting family, methods that build powerful ensembles of weak learners and have dominated machine learning competitions in recent years. Each model was evaluated using ten-fold cross-validation in the training set, a procedure in which the data are repeatedly split so that every case serves as a test for a model trained on the others, reducing the risk that good performance is merely an artifact of one lucky split.

The verdict was striking. Logistic regression, the simplest and most transparent of the eight, achieved the best performance for the study’s screening-oriented goal, posting the highest area under the receiver operating characteristic curve at 0.827, along with the best recall of 0.743 and the best F1 score of 0.621. When the final model was applied to an independent test set it had never seen, it held up well, achieving an AUC of 0.829, accuracy of 0.730, precision of 0.500, recall of 0.729, and an F1 score of 0.593. In practical terms, the model correctly identified roughly three out of four children with low resilience-related capacity, though half of its positive flags were false alarms, a trade-off that may be acceptable in a screening context where a flag leads to supportive follow-up rather than a diagnosis.

The result echoes a growing body of evidence in psychology and forensic science that complex, high-capacity models do not automatically outperform well-specified simpler ones, particularly on tabular data of modest size. Gradient boosting methods can capture intricate nonlinear interactions, but they also risk overfitting when the number of informative features is limited and the sample, while respectable at over a thousand participants, is not enormous. Logistic regression, by contrast, imposes a simple linear structure that appears to match the underlying signal in this dataset. For schools considering such tools, the simplicity bonus is doubled: the model that predicts best is also the one whose inner workings are easiest to explain to teachers, parents, and ethics boards.

That explainability was not left to chance. The researchers supplemented traditional feature importance rankings with SHAP, or SHapley Additive exPlanations, a technique borrowed from cooperative game theory that assigns each feature a contribution value for every individual prediction. Rather than treating the model as a black box, SHAP analysis reveals how much each factor pushed a given child’s predicted risk up or down, and in which direction. This dual approach, combining global feature importance with case-level explanations, represents the current best practice in what has become known as explainable artificial intelligence, and it is what allowed the team to move beyond raw prediction scores to substantive psychological insight.

What emerged from those analyses was a clear hierarchy of predictors. The five most important were hyperactivity, negative affectivity, peer support, emotional symptoms, and prosocial behavior. Notably, four of the five are internal characteristics of the child or closely tied to the child’s interpersonal world, rather than broad environmental factors such as family socioeconomic status or school climate. The authors interpret this pattern as suggesting that relatively low resilience-related capacity among Chinese elementary school students under ordinary developmental pressures is particularly associated with internal vulnerabilities and lower levels of individual and interpersonal protective resources. In other words, the children most likely to struggle are those carrying temperamental and emotional burdens, and lacking the social scaffolding of supportive peers and their own habits of helpful, cooperative behavior.

Each of these top predictors points toward a concrete avenue for intervention. Hyperactivity and emotional symptoms are already frequent targets of school-based mental health programs, and the findings reinforce the value of identifying children with elevated scores on these dimensions early. Negative affectivity, a temperamental tendency toward distress, is harder to change but can inform how adults interpret and respond to a child’s reactions. Peer support and prosocial behavior, meanwhile, are genuinely modifiable: schools can cultivate inclusive classroom climates, structured cooperative activities, and opportunities for children to practice helping others, which prior research links to stronger well-being and resilience. The reciprocity is intuitive, since children who help others tend to build the friendships that later buffer stress.

The study’s limitations are worth keeping in view. The sample comprised Chinese elementary school students, and cultural context shapes both how resilience is expressed and how questionnaire items are answered, so generalization to other populations requires further testing. The predictors relied on self- and parent-report questionnaires rather than clinical interviews, and the model’s precision of 0.500 means any real-world deployment would need to treat predictions as prompts for human judgment, not verdicts. The researchers also note that data are available from the first author on reasonable request, supporting scrutiny of the work. Still, the study was approved by the Institutional Review Board of the School of Psychology at Capital Normal University, informed consent was obtained from all participating parents and children, and the work was supported by the National Natural Science Foundation of China and a Beijing Educational Science Planning grant.

For a field increasingly enamored with ever-larger models, the message of this research is refreshingly grounded. Predicting which children are at risk of low resilience does not require exotic algorithms; it requires good measures, a longitudinal design, honest validation on held-out data, and explanation tools that make the model’s reasoning legible. The fact that hyperactivity, negative affectivity, peer support, emotional symptoms, and prosocial behavior rose to the top gives schools a short, actionable checklist, and the fact that a transparent logistic regression did the predicting means the checklist comes with an interpretable engine behind it. As machine learning continues to seep into education and mental health, this study offers a template for doing it responsibly: prioritize the screening goal, compare simple and complex models on equal footing, and demand explanations before trusting the predictions.

Subject of Research: Explainable machine learning prediction of psychological resilience in Chinese elementary school students

Article Title: Explainable prediction of psychological resilience among Chinese elementary school students using machine learning

Article References: Li, T., Wang, M., & Li, Y. (2026). Explainable prediction of psychological resilience among Chinese elementary school students using machine learning. Current Psychology, 45(20), Article 1597. https://doi.org/10.1007/s12144-026-10158-w

Image Credits: AI Generated

DOI: 10.1007/s12144-026-10158-w

Keywords: psychological resilience, machine learning, logistic regression, SHAP, elementary school students, screening, hyperactivity, peer support, emotional symptoms, prosocial behavior, longitudinal study, child mental health

The study drew on the Adolescent Brain Cognitive Development Study, or ABCD, the largest long-term study of brain development and child health in the United States. From this cohort the researchers analyzed data from 6,001 children, combining each child’s genome with multimodal magnetic resonance imaging and detailed behavioral assessments. The genetic measure at the heart of the work is the polygenic risk score, a single number that aggregates the small effects of thousands of common genetic variants spread across the genome. Individually, these variants contribute almost nothing; summed together, they capture a meaningful slice of inherited liability. Polygenic risk scores for ADHD have previously been shown to predict attention problems in population samples and to relate to white matter microstructure and subcortical connectivity, but earlier efforts typically examined one imaging modality at a time, leaving the joint structure-function picture incomplete.

To overcome that limitation, the team deployed a supervised multimodal fusion framework called MCCAR, a method designed to find patterns that covary together across different types of brain data while remaining anchored to an external variable of interest, in this case the ADHD polygenic risk score. Rather than asking which brain regions differ between children with and without an ADHD diagnosis, the approach asks a subtler question: which coordinated combinations of brain features, spanning functional activity and gray matter volume, vary in step with inherited genetic risk across the entire population? This genetically anchored strategy is powerful because it does not depend on diagnostic categories, which are themselves noisy and heterogeneous, and because it can detect risk-related brain organization in children who may never meet clinical criteria for the disorder.

The signatures that emerged were strikingly concentrated in two systems long implicated in ADHD neurobiology. The first comprises hubs of the cognitive control network, including prefrontal regions that govern attention, planning, impulse inhibition and the flexible allocation of mental resources. Decades of lesion studies, pharmacology and functional imaging have established that catecholamines, particularly dopamine and norepinephrine, tune the gain of dorsolateral prefrontal circuits, and that stimulant medications act on precisely these systems. The second set of regions lies in the midbrain, in nuclei that include the ventral tegmental area and neighboring dopaminergic cell groups, the source neurons that manufacture and project dopamine to the striatum and prefrontal cortex. Crucially, the genetic risk signal was expressed in both directions of brain organization: the affected regions showed aberrations in functional activity, the moment-to-moment fluctuations measured by resting-state fMRI, and in gray matter volume, the structural tissue measure derived from anatomical scans.

Reproducibility, the perennial Achilles heel of brain-wide association studies, received unusually rigorous treatment. A landmark 2022 analysis in Nature showed that many celebrated brain-behavior correlations collapse when sample sizes fall below thousands of participants, fueling skepticism about small neuroimaging findings. The new study confronted this problem head-on. The researchers confirmed that their identified signatures held up across ancestry-defined subsamples and across sex-stratified groups, demonstrating that the patterns were not artifacts of population structure or of the genetic differences between demographic strata that can contaminate polygenic score analyses. This kind of internal replication, performed within a single large cohort but across independent strata, provides a level of confidence that single-split analyses rarely achieve, and it addresses one of the most common criticisms leveled at imaging genetics research.

Specificity was tested with equal care. Psychiatric disorders are genetically entangled: genome-wide analyses have revealed substantial sharing of common variant liability among ADHD, autism, depression, bipolar disorder and schizophrenia, which means a brain signature tied to ADHD genes might plausibly also track risk for related conditions. To rule this out, the team compared their ADHD polygenic risk-associated brain pattern against data spanning eleven psychiatric disorders. The signature proved highly specific to ADHD, distinguishing it from the multimodal patterns associated with the other conditions. That result matters for the emerging transdiagnostic framework in psychiatry, which emphasizes dimensions and shared mechanisms over diagnostic silos. The findings suggest that while some neural features may be shared across neurodevelopmental conditions, the circuit pattern coupled to ADHD’s inherited risk is distinctive enough to serve as a disorder-anchored marker.

The final and arguably most consequential test was predictive. If a genetically informed brain signature truly forms part of the biological pathway from genes to symptoms, it should carry information about behavior. Across the discovery cohort, the signature robustly predicted scores on the two cardinal behavioral phenotypes of ADHD, inattention and hyperactivity. The team then took the pattern derived in ABCD and applied it to the independent ADHD-200 cohort, a publicly available multisite dataset assembled from a completely different set of scanners, protocols and participants. The predictive relationship held, underscoring that the brain signature is not a quirk of one dataset but a generalizable neural correlate of ADHD’s behavioral traits. This discovery-to-validation arc, spanning two independent cohorts, is exactly the structure that critics of neuroimaging psychiatry have demanded for years.

Mechanistically, the convergence on midbrain dopaminergic nuclei and cognitive control hubs reads like a vindication of the disorder’s oldest neurobiological hypothesis. Since the 1950s, the paradoxical calming effect of stimulants in hyperactive children has pointed toward dopamine, and modern work has refined that idea into models in which altered dopaminergic signaling disrupts the prefrontal circuits that normally filter distractions, suppress premature responses and sustain goal-directed effort. The new study adds a genetic bridge: common variants that collectively raise ADHD liability appear to leave their imprint on exactly the structures that produce and respond to dopamine, and on the control networks those structures regulate. The study’s authors frame the work as offering key insights into developing genetically anchored, ADHD-specific brain signatures with transdiagnostic potential, bridging polygenic risk to symptomatology.

The practical implications are still some distance away, but they are coming into view. Polygenic risk scores for ADHD currently explain only a modest fraction of variance, and no responsible clinician will diagnose a child from a genome and a brain scan. Yet markers of this kind could eventually help stratify children before symptoms fully emerge, identify those most likely to benefit from specific interventions, and provide objective endpoints for treatment trials. The researchers have made their tools available: the core fusion code of MCCAR has been released and integrated into the public Fusion ICA Toolbox, and the analysis scripts supporting the study are posted on GitHub, lowering the barrier for other teams to replicate and extend the approach. Genotype, imaging and behavioral data from ABCD remain accessible through the NIMH Data Archive, and the ADHD-200 data are openly available, so the entire analytical chain can be independently audited.

What the study ultimately delivers is a proof of concept that the longest-standing puzzle in ADHD biology, the link between inherited risk and observable behavior, can be traced through the brain with methods rigorous enough to survive modern reproducibility standards. By fusing genetics and multimodal imaging under a supervised framework, and by validating across ancestries, sexes, diagnostic categories and independent cohorts, the work sketches a concrete biological itinerary: from thousands of common variants, to dopaminergic midbrain nuclei and prefrontal control hubs, to the everyday struggles with attention and impulse that define the condition. For a disorder that affects millions of children worldwide and has too often been dismissed as either overdiagnosed or underexplained, that itinerary is more than a technical achievement. It is a step toward grounding ADHD in measurable biology, on the disorder’s own terms.

Subject of Research: Linking ADHD polygenic risk to reproducible multimodal neuroimaging signatures that predict core behavioral phenotypes

Article Title: ADHD polygenic risk converges on reproducible neuroimaging signatures predicting core behavioral phenotypes

Article References: Shan, X., Xu, M., Zhi, D., Qi, S., Jiang, R., Fu, Z., Stevens, M., Pearlson, G., & Sui, J. (2026). ADHD polygenic risk converges on reproducible neuroimaging signatures predicting core behavioral phenotypes. Nature Mental Health. https://doi.org/10.1038/s44220-026-00720-7

Image Credits: AI Generated

DOI: 10.1038/s44220-026-00720-7

Keywords: ADHD, polygenic risk score, neuroimaging, multimodal fusion, ABCD study, dopamine, cognitive control network, gray matter volume, inattention, hyperactivity, ADHD-200, reproducibility

Earlier animal work had already hinted that overexposure to digital-style stimulation very early in life, before weaning, could produce hyperactivity, social deficits and disrupted neuroplasticity. What remained unclear was whether the childhood period, after pups are weaned but before adolescence, carries a similar risk. This matters because human screen exposure frequently intensifies during exactly this stage, when children begin choosing their own content and spending longer stretches with tablets, televisions and gaming devices. The Iranian team, led by Amirreza Hosseinzadeh and corresponding author Monireh Mansouri, together with Hamidreza Pouretemad and Mozhan Parsa, designed their experiment to test that specific post-weaning window under tightly controlled laboratory conditions.

The experimental protocol was deliberately simple in concept but rigorous in execution. Male rats, beginning at postnatal day 22, the point of weaning, were placed in front of a display presenting colored lights paired with cartoon sounds for six hours every single day, continuing until postnatal day 52, which corresponds roughly to adolescence in the rat. This sustained regimen of excessive audiovisual stimulation, abbreviated EAVS by the authors, was intended to model, in a controlled fashion, the kind of high-intensity, fast-paced sensory input that digital media delivers to developing children. The animals were then put through a battery of standard behavioral assays designed to probe the domains most commonly affected in autism spectrum disorder: social interaction, repetitive behavior, and activity levels.

The behavioral results were unambiguous. Rats that had experienced the daily audiovisual overstimulation showed markedly increased repetitive behaviors, one of the diagnostic hallmarks of autism spectrum conditions. They also demonstrated impaired social interaction, engaging less and differently with conspecifics than their normally reared peers. In addition, the exposed animals were hyperactive, moving with an intensity and persistence not seen in controls. Together, the three behavioral changes map onto the classic triad that researchers look for when building animal models of autism-like phenotypes, and their co-occurrence after a purely environmental manipulation is the most provocative aspect of the study.

Behavior alone, however, does not explain mechanism, so the team turned to quantitative neuroanatomy. Using three-dimensional stereological measurement, a technique that allows unbiased estimation of structure volumes and cell numbers from systematically sampled tissue sections, the researchers examined the amygdala, the almond-shaped complex in the medial temporal lobe that assigns emotional significance to sensory input and is heavily implicated in social behavior. What they found was significant enlargement of the amygdala overall, along with an increased number of neurons in two of its key subregions: the basolateral nucleus, which integrates sensory and cortical information feeding into the amygdala, and the central nucleus, which serves as the main output station driving emotional and autonomic responses.

This pattern of structural plasticity is technically significant because the amygdala occupies an unusual position in autism research. Human neuroimaging studies of autism spectrum disorder have variously reported amygdala enlargement in young children with the condition, and altered connectivity between the amygdala and social brain networks is a recurring finding. The fact that a purely environmental input, six hours per day of colored lights and cartoon sounds, was sufficient to drive both volume increases and neuron number increases in the basolateral and central subregions suggests that developmental sensory overstimulation can push the amygdala along a growth trajectory that mirrors, at least superficially, the structural differences observed in clinical populations. The authors are careful, however, to frame this as an autism-like phenotype rather than a model of autism itself.

That caution is important and the researchers state it explicitly. Rodents do not voluntarily choose their media, so the model cannot capture the motivational and behavioral complexity of human device use, where children actively seek out screens and their exposure patterns vary enormously. What the model does offer is experimental control: a defined stimulus, a defined duration, a defined developmental window, and a homogenous genetic background. Within those constraints, the study isolates the neurobiological consequences of sensory overstimulation per se, free from the confounds of family environment, socioeconomic status, or pre-existing developmental differences that complicate every human screen-time study. It is precisely this control that human correlational research can never achieve.

The timing of the manipulation also carries an implicit message about brain development. Postnatal day 22 to 52 in the rat encompasses a period of intense experience-dependent plasticity, when circuits in the limbic system and cortex are being refined by environmental input. The amygdala, in particular, continues to mature through this window, and the increased neuron counts observed in the basolateral and central nuclei indicate that the overstimulation did not merely alter existing neurons but was associated with changes in neuronal population size, whether through altered neurogenesis, altered survival, or shifts in the timing of developmental cell loss. Untangling which of those mechanisms is responsible will be a natural target for follow-up work, as will determining whether the effects persist into adulthood or can be reversed by returning the animals to normal environments.

For the broader public conversation about children and screens, the study does not say that tablets cause autism. Autism spectrum disorder is a strongly heritable, multifactorial condition, and no single environmental exposure can be called its cause. What the research does support is a narrower and still consequential claim: that the developing brain, through at least the childhood years, is structurally and behaviorally responsive to the intensity and character of its sensory environment, and that sustained, excessive exposure to fast, colorful, noisy audiovisual input can bias that development toward patterns of behavior and brain structure that resemble aspects of autism. Previous work had established this vulnerability for the pre-weaning period; this study extends it into the post-weaning phase, closing a gap that parents and pediatricians had every reason to wonder about.

The study was funded by the Iran Cognitive Sciences and Technologies Council and approved by the Ethics Committee of Shahid Beheshti University, conducted in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals. Its publication as an open-access article ensures that the full methods, stereological protocols and raw behavioral data are available for scrutiny and replication. As debate over childhood screen exposure continues to intensify worldwide, work of this kind provides the kind of mechanistic, experimentally controlled evidence that observational studies alone cannot supply, and it will likely fuel further research into how the timing, duration and content of sensory stimulation interact to shape the social and emotional brain during development.

Subject of Research: Effects of post-weaning excessive audiovisual stimulation on autism-like behavior and amygdala structure in rats

Article Title: Post-weaning audiovisual overstimulation induces autism-like phenotypes and amygdala structural plasticity in rats

Article References: Hosseinzadeh, A., Mansouri, M., Pouretemad, H., & Parsa, M. (2026). Post-weaning audiovisual overstimulation induces autism-like phenotypes and amygdala structural plasticity in rats. BMC Neuroscience. https://doi.org/10.1186/s12868-026-01041-2

Image Credits: AI Generated

DOI: 10.1186/s12868-026-01041-2

Keywords: autism spectrum disorder, audiovisual overstimulation, amygdala, brain development, rats, screen exposure, behavioral neuroscience, stereology, social interaction, repetitive behavior, hyperactivity, childhood

The study, led by Walter E. Kaufmann of the Department of Human Genetics at Emory University School of Medicine and Paul S. Horn of the Division of Neurology at Cincinnati Children’s Hospital Medical Center, together with the FORWARD Consortium, drew on the pediatric FORWARD database, a multi-site registry and longitudinal clinical database established to study fragile X syndrome across specialty clinics in the United States. The analytical sample comprised 1,072 males and 338 females, an unusually large cohort for a rare genetic disorder, allowing the researchers to examine patterns of co-occurrence with a statistical power that smaller, single-center studies could not achieve.

The team focused on the eight most common behavioral comorbidities identified by clinicians in these specialty fragile X clinics: attention problems, anxiety, hyperactivity, autism spectrum disorder, mood disturbances, irritability and aggression, self-injurious behavior, and sleep problems. Importantly, the researchers treated these as clinician-identified behavioral concerns rather than formal diagnoses, with one exception. Autism spectrum disorder was the only comorbidity defined using explicit DSM-5 criteria, reflecting the particular diagnostic challenges that autism presents in fragile X syndrome, where the social and repetitive features of the syndrome itself can blur the boundary between behavioral phenotype and co-occurring condition.

Across the cohort, attention problems and anxiety emerged as the two most frequent behavioral comorbidities, and, notably, also as the mildest in terms of functional impact. By contrast, the cluster of disruptive behaviors captured under the umbrella of irritability, aggression, agitation, and self-injurious behavior, while less universally present, carried substantially greater weight in terms of impairment. Roughly half of the children affected by any given behavioral comorbidity showed functional impairment from that condition, a proportion that underscores how often these difficulties cross the threshold from trait to clinically meaningful problem in fragile X syndrome.

The most striking finding concerned co-occurrence. Among children with fragile X syndrome, 69 percent were reported to have impairing behavioral comorbidities occurring together, and 39 percent of those children presented with more than two comorbidities simultaneously. Children with multiple co-occurring conditions showed the greatest overall impairment, a dose-response-like pattern suggesting that the cumulative behavioral load, rather than any single diagnosis, is what most powerfully determines how much a child’s daily functioning is compromised. For families and clinicians, this finding reframes fragile X syndrome not as a condition with one dominant behavioral complication but as one in which multiple behavioral challenges frequently compound one another.

To dissect these patterns, the researchers deployed an unusually rich statistical toolkit. Chi-square analyses quantified how frequently pairs of comorbidities appeared together. Polychoric and polyserial correlations, methods designed to estimate the association between ordinal and continuous variables that underlie observed categorical ratings, allowed the team to measure the strength of association between comorbidities while correcting for the fact that co-occurrence is partly driven by simple frequency. Mann-Whitney tests compared behavioral scale scores between groups. This distinction between frequency-driven co-occurrence and genuine statistical association proved crucial: the researchers found that while which comorbidities co-occur is influenced by how common each one is, the strength of the association between any pair was relatively independent of frequency, implying that some pairings reflect shared underlying mechanisms rather than mere arithmetic.

Two of those strong associations stood out because they replicate well-documented patterns in the general, non-fragile X population. The first was the pairing of attention problems with hyperactivity, a combination familiar from attention deficit-hyperactivity disorder in typical development. The second was the coupling of anxiety with mood disturbances, echoing the extensive literature on comorbidity between anxiety and mood disorders in the general psychiatric population. That these canonical pairings also hold in fragile X syndrome suggests that at least some of the architecture of behavioral comorbidity in this genetic condition mirrors the structure seen in the broader population, even though the underlying biology is profoundly different.

The team also examined how standardized behavioral scale scores related to the clinician-identified comorbidities. Scale scores broadly aligned with other severity parameters, lending convergent validity to the clinician-based classification. However, correlations between comorbidities and specific scales tracked the relevance of the behaviors each scale evaluates, rather than behaving as simple proxies for severity. In other words, a given scale correlated most strongly with the comorbidities whose symptom content it directly captures, a finding that helps clinicians interpret behavioral checklist data in fragile X syndrome with appropriate nuance.

Perhaps the most unexpected result of the entire analysis was the discovery that all eight behavioral comorbidities were strongly correlated with the Sensory problems scale. Sensory difficulties, including hyperarousal to sound, touch, and visual stimuli, have been increasingly recognized as central features of fragile X syndrome, with prior work from the FORWARD registry documenting their prevalence and their links to hyperarousal. But the uniformity of the association across such clinically distinct comorbidities, from attention deficits to aggression to sleep disturbance, was not anticipated. The finding raises the possibility that sensory dysregulation functions as a transdiagnostic factor in fragile X syndrome, a common physiological substrate that feeds into or amplifies virtually every major behavioral problem the condition produces. If confirmed by future longitudinal work, this would carry direct therapeutic implications: interventions targeting sensory hypersensitivity and hyperarousal might yield benefits across a wide swath of behavioral comorbidities simultaneously, rather than requiring separate treatments for each condition.

The authors caution that their data are cross-sectional and clinic-based, drawn from children evaluated at specialty fragile X clinics and therefore not necessarily representative of the full population spectrum of the disorder, which includes individuals with milder presentations never referred to specialty care. Even so, the scale of the FORWARD database and the consistency of the findings give the profiles considerable weight. The research team suggests that the reported profiles of impairing behavioral comorbidities could assist clinicians in the early identification of behavioral symptoms in children with fragile X syndrome and in refining the management of those symptoms, and, they add, perhaps in other neurodevelopmental disorders as well, where multi-comorbidity burdens are similarly common and similarly under-characterized. The study was supported by cooperative agreements and a contract funded by the Centers for Disease Control and Prevention, and de-identified data from the FORWARD project are available under the project’s data sharing plan and CDC policies, opening the door for further analyses of how these behavioral trajectories unfold over time.

Subject of Research: Behavioral comorbidities and their co-occurrence patterns in children with fragile X syndrome

Article Title: Profile of Behavioral Comorbidities in Children With Fragile X Syndrome

Article References: Kaufmann, W. E., Horn, P. S., FORWARD Consortium, Berry-Kravis, E., Velinov, M., Talboy, A. L., Sherman, S. L., Kaufmann, W. E., Schuster, M., Tartaglia, N., Filipink, R. A., Budimirovic, D. B., Barbouth, D., Lightbody, A., Reiss, A., Delahunty, C. M., Hagerman, R. J., Hessl, D., Erickson, C. A., … Morris, S. M. (2026). Profile of Behavioral Comorbidities in Children With Fragile X Syndrome. Journal of Autism and Developmental Disorders. https://doi.org/10.1007/s10803-026-07525-8

Image Credits: AI Generated

DOI: 10.1007/s10803-026-07525-8

Keywords: fragile X syndrome, behavioral comorbidities, anxiety, attention problems, autism spectrum disorder, sensory problems, FORWARD registry, intellectual disability, hyperactivity, self-injurious behavior, neurodevelopmental disorders, FMR1

The findings arrive amid a widening scientific debate over fluoride, a substance added to community water supplies at 0.7 milligrams per litre to prevent tooth decay, yet one that occurs naturally in groundwater at levels that can vary dramatically with local geology. In rural New Hampshire, where many families rely on private, unregulated wells drilled into fluoride-bearing rock, concentrations can be far higher than in fluoridated municipal systems; roughly thirteen percent of wells in regions underlain by felsic igneous rock exceed the U.S. Environmental Protection Agency’s secondary maximum contaminant level of 2.0 milligrams per litre. Against this backdrop, the research team set out to test whether prenatal fluoride exposure is linked not to intelligence, which has dominated the literature, but to the social-emotional and behavioural dimensions of child development that have received far less scrutiny.

Methodologically, the study distinguishes itself through its choice of exposure biomarker. Most previous prospective studies relied on maternal urinary fluoride, which reflects intake over only about six hours—a snapshot that can fluctuate with tea drinking, meal composition, and water consumption on a given day. Toenails, by contrast, grow slowly and integrate fluoride exposure over approximately three to six months, capturing cumulative intake from all sources: drinking water, beverages prepared with fluoridated water, seafood, tea, and the inadvertent swallowing of fluoridated dental products. Because fluoride readily crosses the placenta and penetrates the blood-brain barrier, the developing fetal brain is directly exposed, making the timing and magnitude of maternal exposure biologically consequential.

The researchers analyzed toenail clippings collected from women at roughly 28 weeks of gestation and again about five weeks after delivery, yielding estimates of fluoride exposure during early pregnancy and mid-to-late pregnancy respectively. Samples were analyzed at the Indiana University School of Dentistry using a hexamethyldisiloxane microdiffusion procedure coupled with fluoride ion-selective electrode detection, and specimens weighing less than 2.5 milligrams were excluded after validation work showed that smaller samples produce unreliable results. Household tap water collected at enrollment was analyzed with a fluoride ion-selective electrode as well. The final analytic samples comprised 358 mother-child pairs with toenail fluoride data and 381 pairs with water fluoride data, drawn from a cohort recruited between 2009 and 2013 from prenatal clinics serving women using private water systems.

Children’s behaviour was assessed at age five with the Behavior Assessment System for Children, Second Edition, Parent Rating Scale for preschoolers, a validated instrument that generates age-standardized T-scores across clinical scales such as hyperactivity, aggression, attention problems, anxiety, depression, and withdrawal, along with composite indices of externalizing problems, internalizing problems, and overall behavioural symptoms. Scores of 60 or above fall into the at-risk range. In this sample, most children scored near the normative mean, with between six and twenty-eight percent in the at-risk range depending on the scale. The analysis was preregistered, and the modeling strategy used generalized estimating equations to estimate the odds of at-risk outcomes per one microgram per gram increase in toenail fluoride, with interaction terms testing for differences by child sex and exposure window.

The headline result was statistically robust: each one microgram per gram increase in toenail fluoride reflecting early pregnancy was associated with thirty-nine percent higher odds of scoring in the at-risk range for externalizing problems, with a confidence interval running from fifteen to sixty-seven percent above baseline. The late-pregnancy estimate was similar in magnitude, at thirty percent higher odds, though its confidence interval narrowly included the null. On the individual clinical scales, early-pregnancy fluoride was linked to twenty percent higher odds of at-risk hyperactivity, and the late-pregnancy estimate of twenty-four percent approached significance. The pattern suggests that no single trimester stands out as uniquely vulnerable; rather, fluoride exposure across pregnancy appears relevant to externalizing behaviour.

The sex-specific findings added a further layer of complexity. Among girls, each one microgram per gram increase in early-pregnancy toenail fluoride was associated with thirty-seven percent higher odds of at-risk depressive symptoms, while no association appeared in boys—a statistically significant interaction. Suggestive but imprecise signals also emerged for anxiety and withdrawal in girls in relation to late-pregnancy exposure. This pattern is intriguing because prior fluoride studies, particularly those examining intelligence outcomes, have often reported stronger effects in boys. The authors note that internalizing and externalizing problems may share transdiagnostic roots, including emotion dysregulation, and that sex-specific biological responses to prenatal exposures could shape which behavioural domain is affected.

Plausible biological mechanisms are beginning to take shape in the literature. Fluoride has been implicated in disruption of the thyroid axis, a concern in pregnancy because the fetus depends entirely on maternal thyroid hormones during the first trimester, and even mild maternal thyroid dysfunction has been linked to both internalizing and externalizing problems in children. Fluoride exposure has also been shown to induce oxidative stress in animal and human studies, a pathway independently associated with behavioural problems in offspring. Low iodine status, which compromises thyroid hormone synthesis, may exacerbate fluoride’s effects, and emerging evidence suggests thyroid regulation and oxidative stress responses differ by fetal sex. None of these mechanisms is confirmed as the driver of the observed associations, but together they offer a coherent framework for future investigation.

Why did water fluoride fail to predict outcomes while toenail fluoride succeeded? The authors argue that toenails capture total intake from all dietary and product sources, whereas a single tap-water measurement reflects only one exposure pathway and ignores individual differences in consumption volume. This mismatch can produce exposure misclassification that biases associations toward the null. Median water fluoride in the sample was about 0.2 milligrams per litre—comparable to other communities with naturally occurring fluoride—but values ranged from 0.02 to 13.8 milligrams per litre, and correlations between water and toenail measures, while positive, were moderate. The authors also caution that some participants may have reduced tap-water consumption after learning their wells exceeded arsenic guidelines during the study.

The study’s limitations temper its implications. The cohort was predominantly white, college-educated, married, and relatively affluent, and relied on unregulated private wells, so the findings cannot be assumed to extend to populations drinking optimally fluoridated municipal water. Postnatal fluoride exposure was not measured, toenail growth rates vary between individuals, and only a small fraction of children scored in the at-risk range, limiting statistical power. Yet extensive sensitivity analyses—adjusting for co-occurring metals, imputing missing data, removing influential observations, and modeling outcomes continuously—left the primary conclusions intact. The authors call for longitudinal follow-up, validation of toenail fluoride reference ranges in pregnant populations, and exposome-wide approaches that jointly model fluoride alongside the many other prenatal environmental exposures that shape the developing brain.

Subject of Research: The association between prenatal fluoride exposure measured in maternal toenails and social-emotional and behavioural outcomes in preschool children.

Article Title: Long-term exposure to prenatal fluoride and social-emotional and behavioural outcomes in children in the New Hampshire Birth Cohort Study

Article References: Goodman, C. V., Karagas, M. R., Peacock, J. L., Korrick, S., Flora, D., Lanphear, B., Martinez-Mier, E. A., Tamayo-Cabeza, G., Lippert, F., & Till, C. (2026). Long-term exposure to prenatal fluoride and social-emotional and behavioural outcomes in children in the New Hampshire Birth Cohort Study. Environmental Advances, 25, Article 100753. https://doi.org/10.1016/j.envadv.2026.100753

Image Credits: AI Generated

DOI: 10.1016/j.envadv.2026.100753

Keywords: fluoride, prenatal exposure, child behaviour, hyperactivity, New Hampshire Birth Cohort Study, toenail biomarker, drinking water, neurodevelopment, thyroid, externalizing problems, internalizing problems, private wells