Individualized prediction models in ADHD: a systematic review and meta-regression

Molecular Psychiatry, Journal Year: 2024, Volume and Issue: 29(12), P. 3865 - 3873

Published: May 23, 2024

Abstract There have been increasing efforts to develop prediction models supporting personalised detection, prediction, or treatment of ADHD. We overviewed the current status science in ADHD by: (1) systematically reviewing and appraising available models; (2) quantitatively assessing factors impacting performance published models. did a PRISMA/CHARMS/TRIPOD-compliant systematic review (PROSPERO: CRD42023387502), searching, until 20/12/2023, studies reporting internally and/or externally validated diagnostic/prognostic/treatment-response Using meta-regressions, we explored impact affecting area under curve (AUC) assessed study risk bias with Prediction Model Risk Bias Assessment Tool (PROBAST). From 7764 identified records, 100 were included (88% diagnostic, 5% prognostic, 7% treatment-response). Of these, 96% validated, respectively. None was implemented clinical practice. Only 8% deemed at low bias; 67% considered high bias. Clinical, neuroimaging, cognitive predictors used 35%, 31%, 27% studies, The increased those including, compared not (β = 6.54, p 0.007). Type validation, age range, type model, number predictors, quality, other alter AUC. Several developed support diagnosis However, predict outcomes response limited, none is ready for implementation into use which may be combined seems improve A new generation research should address these gaps by conducting replicable, models, followed research.

Language: Английский

---

Can we diagnose mental disorders in children? A large‐scale assessment of machine learning on structural neuroimaging of 6916 children in the adolescent brain cognitive development study

JCPP Advances, Journal Year: 2023, Volume and Issue: 3(4)

Published: June 28, 2023

Abstract Background Prediction of mental disorders based on neuroimaging is an emerging area research with promising first results in adults. However, the unique demographic children underrepresented and it doubtful whether findings obtained adults can be transferred to children. Methods Using data from 6916 aged 9–10 multicenter Adolescent Brain Cognitive Development study, we extracted 136 regional volume thickness measures structural magnetic resonance images rigorously evaluate capabilities machine learning predict 10 different psychiatric disorders: major depressive disorder, bipolar disorder (BD), psychotic symptoms, attention deficit hyperactivity (ADHD), oppositional defiant conduct post‐traumatic stress obsessive‐compulsive generalized anxiety social disorder. For each performed cross‐validation assessed models discovered a true pattern via permutation testing. Results Two detected statistical significance when using advanced that (i) allow for non‐linear relationships between neuroanatomy (ii) model interdependencies disorders, (iii) avoid confounding due sociodemographic factors: ADHD (AUROC = 0.567, p 0.002) BD 0.551, 0.002). In contrast, traditional perform consistently worse only 0.529, Conclusion While modest absolute classification performance does not warrant application clinic, our provide empirical evidence embracing explicitly accounting complexities discover patterns would remain hidden models.

Language: Английский

---

Differential gray matter correlates and machine learning prediction of abuse and internalizing psychopathology in adolescent females

Scientific Reports, Journal Year: 2025, Volume and Issue: 15(1)

Published: Jan. 3, 2025

Childhood abuse represents one of the most potent risk factors for development psychopathology during childhood, accounting 30–60% onset. While previous studies have separately associated reductions in gray matter volume (GMV) with childhood and internalizing (IP), it is unclear whether IP differ their structural abnormalities, which GMV features are related to at individual level. In a pooled multisite, multi-investigator sample, 246 child adolescent females between ages 8–18 were recruited into interpersonal violence (IPV) and/or (i.e. posttraumatic stress disorder (PTSD), depression, anxiety). Youth completed assessments IP, history, underwent high resolution T1 MRI. First, we characterized how differences exposure depend on presence or absence using voxel-based morphometry (VBM). Next, trained convolutional neural networks predict experience estimated strength direction importance each feature making individual-level predictions Shapley values. values aggregated across entire cohort, top 1% clusters highest reported. At group-level, VBM analyses identified widespread decreases prefrontal cortex, insula, hippocampus youth while was specifically increased cingulate cortex supramarginal gyrus. Further, interactions severity ventral dorsal anterior thalamus. After extensive training, model tuning, evaluation, performed above chance when predicting (63% accuracy) experiences (71% level individual. Interestingly, regions had degree overlap group-level patterns. We unique correlates both group overlap, providing evidence that trauma may uniquely jointly impact neurodevelopment. Feature learning offer power novelty beyond traditional approaches identification biomarkers movement towards individualized diagnosis treatment.

Language: Английский

---

A NEUROANATOMIA FUNCIONAL E NOVAS PERSPECTIVAS PARA PSIQUIATRIA INFANTIL: UMA REVISÃO SISTEMÁTICA

Revista Foco, Journal Year: 2025, Volume and Issue: 18(3), P. e7900 - e7900

Published: March 5, 2025

INTRODUÇÃO: O diagnóstico das psicopatologias é baseado em aspectos clínicos e autorreferidos bastante heterogêneos inespecíficos, sendo um desafio sobretudo na psiquiatria infantil. Diante disso, muitas pesquisas buscam, através da neuroanatomia funcional, critérios objetivos que colaborem prática clínica. OBJETIVO: Reunir estudos exploram a aplicabilidade funcional distúrbios neuropsiquiátricos MÉTODO: Selecionou-se artigos nas bases de dados PubMed, BVS SCIELO, seguindo os PRISMA conforme elegibilidade: disponibilidade integralmente plataforma digital, originais, datados entre 2019 2023. RESULTADOS: Foram selecionados 17 após aplicação dos elegibilidade, retirada duplicatas avaliação, partir leitura títulos, resumos texto completo com maior ênfase relação nos infância adolescência. DISCUSSÃO: Embora muitos contribuam para compreensão inspirem seu uso clínico, esses ainda apresentam grandes desafios fundamentação seus resultados. CONCLUSÃO: A colabora o entendimento promove novas perspectivas infantil ao possibilitar aprimoramento tratamento individualizado.

---

Reliable multimodal brain signatures predict mental health outcomes in children

Biological Psychiatry Cognitive Neuroscience and Neuroimaging, Journal Year: 2025, Volume and Issue: unknown

Published: March 1, 2025

Inter-individual brain differences likely precede the emergence of mood and anxiety disorders, however, specific alterations remain unclear. While many studies focus on a single imaging modality in isolation, recent advances multimodal image analysis allow for more comprehensive understanding complex neurobiology that underlies mental health. In large population-based cohort children from Adolescent Brain Cognitive Development (ABCD) study (N > 10K), we applied data-driven linked independent component to identify variations cortical structure white matter microstructure together predict longitudinal behavioural health symptoms. were examined sub-sample twins depending presence at-risk behaviours. Two signatures at age 9-10y predicted symptoms 9-12y, with small effect sizes. Cortical association, limbic default mode regions peripheral higher depression across two split-halves. The signature differed amongst symptom trajectories related emotion-regulation network functional connectivity. Linked subcortical structures projection tract variably inhibition, sensation seeking, psychosis severity over time male participants. These patterns significantly different between pairs discordant self-injurious behaviour. Our results demonstrate reliable, childhood, before disorders tend emerge, lay foundation long-term outcomes offer targets early identification at-risk.

Language: Английский

---

Predicting children's emotional and behavioral difficulties at age five using pregnancy and newborn risk factors: Evidence from the UK Household Longitudinal Study

Journal of Affective Disorders, Journal Year: 2025, Volume and Issue: unknown

Published: May 1, 2025

Childhood emotional and behavioral difficulties have a profound impact on later life outcomes, making it crucial to identify early-life risk factors that predict difficulties. However, much of the existing research has concentrated diagnosing, rather than predicting, difficulties, often focused adolescents younger children. This study employs machine learning (ML) techniques construct an interpretable predictive model using data from UK Household Longitudinal Study, aiming key influence children's during childhood. We examined maternal habits pregnancy parent-reported birth, breastfeeding regulatory problems newborn stage. Our findings highlighted lack breastfeeding, low birthweight smoking as three most significant predictors Other important were related infant problems. Heterogeneity analysis revealed gender differences in power, with being stronger predictor for boys, amount fussing infancy having greater girls. highlights importance comprehensive prenatal postnatal care, advocates early screening calls gender-specific approaches assessing addressing

Language: Английский

---

Longitudinal sex-at-birth and age analyses of cortical structure in the ABCD Study^®

bioRxiv (Cold Spring Harbor Laboratory), Journal Year: 2024, Volume and Issue: unknown

Published: June 11, 2024

Abstract While the brain continues to develop during adolescence, such development may depend on sex-at-birth. However, elucidation of differences be hindered by analytical decisions (e.g., covariate selection address body/brain-size differences) and typical reporting cross-sectional data. To further evaluate adolescent cortical development, we analyzed data from Adolescent Brain Cognitive Development Study SM , whose cohort 11,000+ youth participants with biannual neuroimaging collection can facilitate understanding neuroanatomical change a critical developmental window. Doubly considering individual in context group-level effects, regional changes thickness, sulcal depth, surface area, volume between two timepoints (∼2 years apart) 9-to 12-year-olds assigned male or female First, conducted linear mixed-effects models gauge how controlling for intracranial volume, whole-brain (WBV), summary metric mean thickness) influenced interpretations age-dependent change. Next, evaluated relative thickness area as function sex-at-birth age. Here, showed that WBV (thickness, volume) total were more optimal covariates; different covariates would have substantially altered our overall sex-at-birth-specific development. Further, provided evidence suggest aggregate is changing generally comparable across those sex-at-birth, corresponding happening at slightly older ages Overall, these results help elucidate trajectories early adolescence. Significance Statement most brain’s happens life, much it still Because many factors alter trajectories, important shape/timing (i.e., what constitutes development). affected choose analyze them. way researchers brain/body size affects interpret variation over time. consider similar patterns simply groups. These support relatively novel analyzing

Language: Английский

---

The Transition from Homogeneous to Heterogeneous Machine Learning in Neuropsychiatric Research

Biological Psychiatry Global Open Science, Journal Year: 2024, Volume and Issue: 5(1), P. 100397 - 100397

Published: Sept. 26, 2024

Language: Английский

---

Effectiveness of ML with Neuroimaging Data in Detecting Individuals/Children with ASD

Published: Sept. 27, 2024

Language: Английский

---

Data Analysis Frameworks for Investigating Behavioural Differences

Published: Jan. 1, 2023

This chapter provides an introduction to methods of data analysis that are commonly applied in developmental psychopathology studies including path analysis, cluster structural equation modelling, and machine learning. It emphasises the value meta-analysis address synthesis results numerous comments on "reproducibility crisis" has emerged science recent years its relevance for psychopathology.

Language: Английский

---