Reliable and generalizable brain-based predictions of cognitive functioning across common psychiatric illness

Chopra, Sidhant; Dhamala, Elvisha; Lawhead, Connor; Ricard, Jocelyn A; Orchard, Edwina R; An, Lijun; Chen, Pansheng; Wulan, Naren; Kumar, Poornima; Rubenstein, Arielle; Moses, Julia; Chen, Lia; Aquino, Kevin; Fornito, Alex; Harpaz-Rotem, Illan; Germine, Laura; Baker, Justin T.; Yeo, B. T. Thomas; Holmes, Avram J.

doi:10.1101/2022.12.08.22283232

Cited by 2 publications

(2 citation statements)

References 100 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The datasets in the current study comprised relatively healthy participants. Meta-matching might be potentially useful for psychiatric populations (Chopra et al, 2022). Including psychiatric datasets to the base model training might further improve generalization to new datasets by increasing the diversity of the source datasets.…”

Section: Discussionmentioning

confidence: 99%

Multilayer meta-matching: translating phenotypic prediction models from multiple datasets to small data

Chen,

An,

Wulan

et al. 2023

Preprint

Self Cite

View full text Add to dashboard Cite

Resting-state functional connectivity (RSFC) is widely used to predict phenotypic traits in individuals. Large sample sizes can significantly improve prediction accuracies. However, for studies of certain clinical populations or focused neuroscience inquiries, small-scale datasets often remain a necessity. We have previously proposed a “meta-matching” approach to translate prediction models from large datasets to predict new phenotypes in small datasets. We demonstrated large improvement of meta-matching over classical kernel ridge regression (KRR) when translating models from a single source dataset (UK Biobank) to the Human Connectome Project Young Adults (HCP-YA) dataset. In the current study, we propose two meta-matching variants (“meta-matching with dataset stacking” and “multilayer meta-matching”) to translate models from multiple source datasets across disparate sample sizes to predict new phenotypes in small target datasets. We evaluate both approaches by translating models trained from five source datasets (with sample sizes ranging from 862 participants to 36,834 participants) to predict phenotypes in the HCP-YA and HCP-Aging datasets. We find that multilayer meta-matching modestly outperforms meta-matching with dataset stacking. Both meta-matching variants perform better than the original “meta-matching with stacking” approach trained only on the UK Biobank. All meta-matching variants outperform classical KRR and transfer learning by a large margin. In fact, KRR is better than classical transfer learning when less than 50 participants are available for finetuning, suggesting the difficulty of classical transfer learning in the very small sample regime. The multilayer meta-matching model is publicly available at GITHUB_LINK.

show abstract

Section: Discussionmentioning

confidence: 99%

Multilayer meta-matching: translating phenotypic prediction models from multiple datasets to small data

Chen,

An,

Wulan

et al. 2023

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…New methods for external replication, such as meta-matching, have shown promise in leveraging neural network models that were trained using large samples that could be applied to behavioral prediction in smaller samples (Chopra et al 2022. Recently, metamatching was trained on the UK Biobank and then applied to small clinical samples where large data sets are not possible (Chopra et al 2022. It was found that meta-matching increased the prediction of brain connectivity to cognitive function.…”

Section: Replication Testingmentioning

confidence: 99%

Poverty, Brain Development, and Mental Health: Progress, Challenges, and Paths Forward

Monk

Hardi

2023

Annu. Rev. Dev. Psychol.

View full text Add to dashboard Cite

Poverty is associated with changes in brain development and elevates the risk for psychopathology in childhood, adolescence, and adulthood. Although the field is rapidly expanding, there are methodological challenges that raise questions about the validity of current findings. These challenges include the interrelated issues of reliability, effect size, interindividual heterogeneity, and replicability. To address these issues, we propose a multipronged approach that spans short-, medium-, and long-term solutions, including changes to data pipelines along with more comprehensive data acquisition of environment, brain, and mental health. Additional suggestions are to use open science approaches, more robust statistical analyses, and replication testing. Furthermore, we propose increased integration between advanced analytical approaches using large samples and neuroscience models in intervention research to enhance the interpretability of findings. Collectively, these approaches will expand the application of neuroimaging findings and provide a foundation for eventual policy changes designed to improve conditions for children in poverty. Expected final online publication date for the Annual Review of Developmental Psychology, Volume 5 is December 2023. Please see http://www.annualreviews.org/page/journal/pubdates for revised estimates.

show abstract

Reliable and generalizable brain-based predictions of cognitive functioning across common psychiatric illness

Cited by 2 publications

References 100 publications

Multilayer meta-matching: translating phenotypic prediction models from multiple datasets to small data

Multilayer meta-matching: translating phenotypic prediction models from multiple datasets to small data

Poverty, Brain Development, and Mental Health: Progress, Challenges, and Paths Forward

Contact Info

Product

Resources

About