From 15b8931712fb2ae0dc2338650cc11d9501c5c612 Mon Sep 17 00:00:00 2001 From: alexpron <45215023+alexpron@users.noreply.github.com> Date: Thu, 1 Aug 2024 09:27:34 +0000 Subject: [PATCH] content: import publications from Bibtex --- .../publication/deyoung-beyond-2024/cite.bib | 14 ++++ .../publication/deyoung-beyond-2024/index.md | 56 +++++++++++++ .../publication/feldman-value-2024/cite.bib | 14 ++++ .../publication/feldman-value-2024/index.md | 16 ++++ .../publication/grady-influence-2020/cite.bib | 18 ++++ .../publication/grady-influence-2020/index.md | 48 +++++++++++ .../isager-exploring-2024/cite.bib | 16 ++++ .../isager-exploring-2024/index.md | 18 ++++ .../marek-reproducible-2022/cite.bib | 20 +++++ .../marek-reproducible-2022/index.md | 83 +++++++++++++++++++ .../rosenblatt-data-2024-1/cite.bib | 18 ++++ .../rosenblatt-data-2024-1/index.md | 33 ++++++++ .../publication/soskic-garden-nodate/cite.bib | 13 +++ .../publication/soskic-garden-nodate/index.md | 38 +++++++++ .../spisak-multivariate-2023/cite.bib | 19 +++++ .../spisak-multivariate-2023/index.md | 20 +++++ content/publication/tendler-why-2023/cite.bib | 17 ++++ content/publication/tendler-why-2023/index.md | 26 ++++++ 18 files changed, 487 insertions(+) create mode 100644 content/publication/deyoung-beyond-2024/cite.bib create mode 100644 content/publication/deyoung-beyond-2024/index.md create mode 100644 content/publication/feldman-value-2024/cite.bib create mode 100644 content/publication/feldman-value-2024/index.md create mode 100644 content/publication/grady-influence-2020/cite.bib create mode 100644 content/publication/grady-influence-2020/index.md create mode 100644 content/publication/isager-exploring-2024/cite.bib create mode 100644 content/publication/isager-exploring-2024/index.md create mode 100644 content/publication/marek-reproducible-2022/cite.bib create mode 100644 content/publication/marek-reproducible-2022/index.md create mode 100644 content/publication/rosenblatt-data-2024-1/cite.bib create mode 100644 content/publication/rosenblatt-data-2024-1/index.md create mode 100644 content/publication/soskic-garden-nodate/cite.bib create mode 100644 content/publication/soskic-garden-nodate/index.md create mode 100644 content/publication/spisak-multivariate-2023/cite.bib create mode 100644 content/publication/spisak-multivariate-2023/index.md create mode 100644 content/publication/tendler-why-2023/cite.bib create mode 100644 content/publication/tendler-why-2023/index.md diff --git a/content/publication/deyoung-beyond-2024/cite.bib b/content/publication/deyoung-beyond-2024/cite.bib new file mode 100644 index 0000000..4b2247a --- /dev/null +++ b/content/publication/deyoung-beyond-2024/cite.bib @@ -0,0 +1,14 @@ +@misc{deyoung_beyond_2024, + abstract = {Linking neurobiology to relatively stable individual differences in cognition, emotion, motivation, and behavior can require large sample sizes to yield replicable results. Given the nature of between-person research, sample sizes at least in the hundreds are likely to be necessary in most neuroimaging studies of individual differences, regardless of whether they are investigating the whole brain or more focal hypotheses. However, the appropriate sample size depends on the expected effect size. Therefore, we propose four strategies to increase effect sizes in neuroimaging research, which may help to enable the detection of replicable between-person effects in samples in the hundreds rather than the thousands: (1) theoretical matching between neuroimaging tasks and behavioral constructs of interest; (2) increasing the reliability of both neural and psychological measurement; (3) individualization of measures for each participant; and (4) using multivariate approaches with cross-validation instead of univariate approaches. We discuss challenges associated with these methods and highlight strategies for improvements that will help the field to move toward a more robust and accessible neuroscience of individual differences.}, + author = {DeYoung, Colin G. and Hilger, Kirsten and Hanson, Jamie L. and Abend, Rany and Allen, Timothy and Beaty, Roger and Blain, Scott D. and Chavez, Robert and Engel, Stephen A. and Ma, Feilong and Fornito, Alex and Genç, Erhan and Goghari, Vina and Grazioplene, Rachael G. and Homan, Philipp and Joyner, Keenan and Kaczkurkin, Antonia N. and Latzman, Robert D and Martin, Elizabeth A and Nikolaidis, Aki and Pickering, Alan and Safron, Adam and Sassenberg, Tyler and Servaas, Michelle and Smillie, Luke D. and Spreng, R. Nathan and Viding, Essi and Wacker, Jan}, + copyright = {https://creativecommons.org/licenses/by/4.0/legalcode}, + doi = {10.31219/osf.io/bjn62}, + file = {DeYoung et al. - 2024 - Beyond Increasing Sample Sizes Optimizing Effect .pdf:/home/alpron/Zotero/storage/ZS9NMZRN/DeYoung et al. - 2024 - Beyond Increasing Sample Sizes Optimizing Effect .pdf:application/pdf}, + language = {en}, + month = {July}, + shorttitle = {Beyond Increasing Sample Sizes}, + title = {Beyond Increasing Sample Sizes: Optimizing Effect Sizes in Neuroimaging Research on Individual Differences}, + url = {https://osf.io/bjn62}, + urldate = {2024-07-30}, + year = {2024} +} diff --git a/content/publication/deyoung-beyond-2024/index.md b/content/publication/deyoung-beyond-2024/index.md new file mode 100644 index 0000000..fbbe8a6 --- /dev/null +++ b/content/publication/deyoung-beyond-2024/index.md @@ -0,0 +1,56 @@ +--- +title: 'Beyond Increasing Sample Sizes: Optimizing Effect Sizes in Neuroimaging Research + on Individual Differences' +authors: +- Colin G. DeYoung +- Kirsten Hilger +- Jamie L. Hanson +- Rany Abend +- Timothy Allen +- Roger Beaty +- Scott D. Blain +- Robert Chavez +- Stephen A. Engel +- Feilong Ma +- Alex Fornito +- Erhan Genç +- Vina Goghari +- Rachael G. Grazioplene +- Philipp Homan +- Keenan Joyner +- Antonia N. Kaczkurkin +- Robert D Latzman +- Elizabeth A Martin +- Aki Nikolaidis +- Alan Pickering +- Adam Safron +- Tyler Sassenberg +- Michelle Servaas +- Luke D. Smillie +- R. Nathan Spreng +- Essi Viding +- Jan Wacker +date: '2024-07-01' +publishDate: '2024-08-01T09:27:33.839571Z' +publication_types: +- manuscript +doi: 10.31219/osf.io/bjn62 +abstract: 'Linking neurobiology to relatively stable individual differences in cognition, + emotion, motivation, and behavior can require large sample sizes to yield replicable + results. Given the nature of between-person research, sample sizes at least in the + hundreds are likely to be necessary in most neuroimaging studies of individual differences, + regardless of whether they are investigating the whole brain or more focal hypotheses. + However, the appropriate sample size depends on the expected effect size. Therefore, + we propose four strategies to increase effect sizes in neuroimaging research, which + may help to enable the detection of replicable between-person effects in samples + in the hundreds rather than the thousands: (1) theoretical matching between neuroimaging + tasks and behavioral constructs of interest; (2) increasing the reliability of both + neural and psychological measurement; (3) individualization of measures for each + participant; and (4) using multivariate approaches with cross-validation instead + of univariate approaches. We discuss challenges associated with these methods and + highlight strategies for improvements that will help the field to move toward a + more robust and accessible neuroscience of individual differences.' +links: +- name: URL + url: https://osf.io/bjn62 +--- diff --git a/content/publication/feldman-value-2024/cite.bib b/content/publication/feldman-value-2024/cite.bib new file mode 100644 index 0000000..f87840a --- /dev/null +++ b/content/publication/feldman-value-2024/cite.bib @@ -0,0 +1,14 @@ +@article{feldman_value_2024, + author = {Feldman, Gilad}, + collaborator = {Center For Open Science}, + copyright = {Creative Commons Attribution 4.0 International}, + doi = {10.17605/OSF.IO/BTNUJ}, + file = {Feldman - 2024 - The value of replications goes beyond replicabilit.pdf:/home/alpron/Zotero/storage/MAJXU58Z/Feldman - 2024 - The value of replications goes beyond replicabilit.pdf:application/pdf}, + keywords = {to read}, + note = {Publisher: OSF}, + shorttitle = {The value of replications goes beyond replicability and is tied to the value of the research it replicates}, + title = {The value of replications goes beyond replicability and is tied to the value of the research it replicates: Commentary on Isager et al. (2024)}, + url = {https://osf.io/btnuj/}, + urldate = {2024-08-01}, + year = {2024} +} diff --git a/content/publication/feldman-value-2024/index.md b/content/publication/feldman-value-2024/index.md new file mode 100644 index 0000000..a44027b --- /dev/null +++ b/content/publication/feldman-value-2024/index.md @@ -0,0 +1,16 @@ +--- +title: 'The value of replications goes beyond replicability and is tied to the value + of the research it replicates: Commentary on Isager et al. (2024)' +authors: +- Gilad Feldman +date: '2024-01-01' +publishDate: '2024-08-01T09:27:33.854306Z' +publication_types: +- article-journal +doi: 10.17605/OSF.IO/BTNUJ +tags: +- to read +links: +- name: URL + url: https://osf.io/btnuj/ +--- diff --git a/content/publication/grady-influence-2020/cite.bib b/content/publication/grady-influence-2020/cite.bib new file mode 100644 index 0000000..cd5fddd --- /dev/null +++ b/content/publication/grady-influence-2020/cite.bib @@ -0,0 +1,18 @@ +@article{grady_influence_2020, + abstract = {Limited statistical power due to small sample sizes is a problem in fMRI research. Most of the work to date has examined the impact of sample size on task‐related activation, with less attention paid to the influence of sample size on brain‐behavior correlations, especially in actual experimental fMRI data. We addressed this issue using two large data sets (a working memory task, N = 171, and a relational processing task, N = 865) and both univariate and multivariate approaches to voxel‐wise correlations. We created subsamples of different sizes and calculated correlations between task‐related activity at each voxel and task performance. Across both data sets the magnitude of the brain‐behavior correlations decreased and similarity across spatial maps increased with larger sample sizes. The multivariate technique identified more extensive correlated areas and more similarity across spatial maps, suggesting that a multivariate approach would provide a consistent advantage over univariate approaches in the stability of brain‐behavior correlations. In addition, the multivariate analyses showed that a sample size of roughly 80 or more participants would be needed for stable estimates of correlation magnitude in these data sets. Importantly, a number of additional factors would likely influence the choice of sample size for assessing such correlations in any given experiment, including the cognitive task of interest and the amount of data collected per participant. Our results provide novel experimental evidence in two independent data sets that the sample size commonly used in fMRI studies of 20–30 participants is very unlikely to be sufficient for obtaining reproducible brain‐behavior correlations, regardless of analytic approach., Limited statistical power due to small sample sizes is a problem in fMRI research. Most of the work to date has examined the impact of sample size on task‐related activation, with less attention paid to the influence of sample size on brain‐behavior correlations, especially in actual experimental fMRI data. Our results provide novel experimental evidence in two independent data sets that the sample size commonly used in fMRI studies of 20–30 participants is very unlikely to be sufficient for obtaining reproducible brain‐behavior correlations, regardless of whether a univariate or multivariate approach is used.}, + author = {Grady, Cheryl L. and Rieck, Jenny R. and Nichol, Daniel and Rodrigue, Karen M. and Kennedy, Kristen M.}, + doi = {10.1002/hbm.25217}, + file = {PubMed Central Full Text PDF:/home/alpron/Zotero/storage/W742CY9G/Grady et al. - 2020 - Influence of sample size and analytic approach on .pdf:application/pdf}, + issn = {1065-9471}, + journal = {Human Brain Mapping}, + month = {September}, + number = {1}, + pages = {204--219}, + pmcid = {PMC7721240}, + pmid = {32996635}, + title = {Influence of sample size and analytic approach on stability and interpretation of brain‐behavior correlations in task‐related fMRI data}, + url = {https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7721240/}, + urldate = {2024-07-31}, + volume = {42}, + year = {2020} +} diff --git a/content/publication/grady-influence-2020/index.md b/content/publication/grady-influence-2020/index.md new file mode 100644 index 0000000..94fc37f --- /dev/null +++ b/content/publication/grady-influence-2020/index.md @@ -0,0 +1,48 @@ +--- +title: Influence of sample size and analytic approach on stability and interpretation + of brain‐behavior correlations in task‐related fMRI data +authors: +- Cheryl L. Grady +- Jenny R. Rieck +- Daniel Nichol +- Karen M. Rodrigue +- Kristen M. Kennedy +date: '2020-09-01' +publishDate: '2024-08-01T09:27:33.831633Z' +publication_types: +- article-journal +publication: '*Human Brain Mapping*' +doi: 10.1002/hbm.25217 +abstract: Limited statistical power due to small sample sizes is a problem in fMRI + research. Most of the work to date has examined the impact of sample size on task‐related + activation, with less attention paid to the influence of sample size on brain‐behavior + correlations, especially in actual experimental fMRI data. We addressed this issue + using two large data sets (a working memory task, N = 171, and a relational processing + task, N = 865) and both univariate and multivariate approaches to voxel‐wise correlations. + We created subsamples of different sizes and calculated correlations between task‐related + activity at each voxel and task performance. Across both data sets the magnitude + of the brain‐behavior correlations decreased and similarity across spatial maps + increased with larger sample sizes. The multivariate technique identified more extensive + correlated areas and more similarity across spatial maps, suggesting that a multivariate + approach would provide a consistent advantage over univariate approaches in the + stability of brain‐behavior correlations. In addition, the multivariate analyses + showed that a sample size of roughly 80 or more participants would be needed for + stable estimates of correlation magnitude in these data sets. Importantly, a number + of additional factors would likely influence the choice of sample size for assessing + such correlations in any given experiment, including the cognitive task of interest + and the amount of data collected per participant. Our results provide novel experimental + evidence in two independent data sets that the sample size commonly used in fMRI + studies of 20–30 participants is very unlikely to be sufficient for obtaining reproducible + brain‐behavior correlations, regardless of analytic approach., Limited statistical + power due to small sample sizes is a problem in fMRI research. Most of the work + to date has examined the impact of sample size on task‐related activation, with + less attention paid to the influence of sample size on brain‐behavior correlations, + especially in actual experimental fMRI data. Our results provide novel experimental + evidence in two independent data sets that the sample size commonly used in fMRI + studies of 20–30 participants is very unlikely to be sufficient for obtaining reproducible + brain‐behavior correlations, regardless of whether a univariate or multivariate + approach is used. +links: +- name: URL + url: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7721240/ +--- diff --git a/content/publication/isager-exploring-2024/cite.bib b/content/publication/isager-exploring-2024/cite.bib new file mode 100644 index 0000000..5a24de2 --- /dev/null +++ b/content/publication/isager-exploring-2024/cite.bib @@ -0,0 +1,16 @@ +@article{isager_exploring_2024, + author = {Isager, Peder M. and Lakens, Daniël and Van Leeuwen, Thed and Van 'T Veer, Anna E.}, + doi = {10.1016/j.cortex.2023.10.012}, + file = {Isager et al. - 2024 - Exploring a formal approach to selecting studies f.pdf:/home/alpron/Zotero/storage/Z2VP3RRD/Isager et al. - 2024 - Exploring a formal approach to selecting studies f.pdf:application/pdf}, + issn = {00109452}, + journal = {Cortex}, + language = {en}, + month = {February}, + pages = {330--346}, + shorttitle = {Exploring a formal approach to selecting studies for replication}, + title = {Exploring a formal approach to selecting studies for replication: A feasibility study in social neuroscience}, + url = {https://linkinghub.elsevier.com/retrieve/pii/S0010945223002691}, + urldate = {2024-08-01}, + volume = {171}, + year = {2024} +} diff --git a/content/publication/isager-exploring-2024/index.md b/content/publication/isager-exploring-2024/index.md new file mode 100644 index 0000000..032c2e7 --- /dev/null +++ b/content/publication/isager-exploring-2024/index.md @@ -0,0 +1,18 @@ +--- +title: 'Exploring a formal approach to selecting studies for replication: A feasibility + study in social neuroscience' +authors: +- Peder M. Isager +- Daniël Lakens +- Thed Van Leeuwen +- Anna E. Van 'T Veer +date: '2024-02-01' +publishDate: '2024-08-01T09:27:33.860414Z' +publication_types: +- article-journal +publication: '*Cortex*' +doi: 10.1016/j.cortex.2023.10.012 +links: +- name: URL + url: https://linkinghub.elsevier.com/retrieve/pii/S0010945223002691 +--- diff --git a/content/publication/marek-reproducible-2022/cite.bib b/content/publication/marek-reproducible-2022/cite.bib new file mode 100644 index 0000000..94c5e21 --- /dev/null +++ b/content/publication/marek-reproducible-2022/cite.bib @@ -0,0 +1,20 @@ +@article{marek_reproducible_2022, + abstract = {Magnetic resonance imaging (MRI) has transformed our understanding of the human brain through well-replicated mapping of abilities to specific structures (for example, lesion studies) and functions1–3 (for example, task functional MRI (fMRI)). Mental health research and care have yet to realize similar advances from MRI. A primary challenge has been replicating associations between inter-individual differences in brain structure or function and complex cognitive or mental health phenotypes (brain-wide association studies (BWAS)). Such BWAS have typically relied on sample sizes appropriate for classical brain mapping4 (the median neuroimaging study sample size is about 25), but potentially too small for capturing reproducible brain–behavioural phenotype associations5,6. Here we used three of the largest neuroimaging datasets currently available—with a total sample size of around 50,000 individuals—to quantify BWAS effect sizes and reproducibility as a function of sample size. BWAS associations were smaller than previously thought, resulting in statistically underpowered studies, inflated effect sizes and replication failures at typical sample sizes. As sample sizes grew into the thousands, replication rates began to improve and effect size inflation decreased. More robust BWAS effects were detected for functional MRI (versus structural), cognitive tests (versus mental health questionnaires) and multivariate methods (versus univariate). Smaller than expected brain–phenotype associations and variability across population subsamples can explain widespread BWAS replication failures. In contrast to non-BWAS approaches with larger effects (for example, lesions, interventions and within-person), BWAS reproducibility requires samples with thousands of individuals.}, + author = {Marek, Scott and Tervo-Clemmens, Brenden and Calabro, Finnegan J. and Montez, David F. and Kay, Benjamin P. and Hatoum, Alexander S. and Donohue, Meghan Rose and Foran, William and Miller, Ryland L. and Hendrickson, Timothy J. and Malone, Stephen M. and Kandala, Sridhar and Feczko, Eric and Miranda-Dominguez, Oscar and Graham, Alice M. and Earl, Eric A. and Perrone, Anders J. and Cordova, Michaela and Doyle, Olivia and Moore, Lucille A. and Conan, Gregory M. and Uriarte, Johnny and Snider, Kathy and Lynch, Benjamin J. and Wilgenbusch, James C. and Pengo, Thomas and Tam, Angela and Chen, Jianzhong and Newbold, Dillan J. and Zheng, Annie and Seider, Nicole A. and Van, Andrew N. and Metoki, Athanasia and Chauvin, Roselyne J. and Laumann, Timothy O. and Greene, Deanna J. and Petersen, Steven E. and Garavan, Hugh and Thompson, Wesley K. and Nichols, Thomas E. and Yeo, B. T. Thomas and Barch, Deanna M. and Luna, Beatriz and Fair, Damien A. and Dosenbach, Nico U. F.}, + copyright = {2022 The Author(s), under exclusive licence to Springer Nature Limited}, + doi = {10.1038/s41586-022-04492-9}, + file = {Full Text PDF:/home/alpron/Zotero/storage/RVNDP5RZ/Marek et al. - 2022 - Reproducible brain-wide association studies requir.pdf:application/pdf}, + issn = {1476-4687}, + journal = {Nature}, + keywords = {Cognitive neuroscience, Psychology}, + language = {en}, + month = {March}, + note = {Publisher: Nature Publishing Group}, + number = {7902}, + pages = {654--660}, + title = {Reproducible brain-wide association studies require thousands of individuals}, + url = {https://www.nature.com/articles/s41586-022-04492-9}, + urldate = {2024-07-31}, + volume = {603}, + year = {2022} +} diff --git a/content/publication/marek-reproducible-2022/index.md b/content/publication/marek-reproducible-2022/index.md new file mode 100644 index 0000000..0d0c962 --- /dev/null +++ b/content/publication/marek-reproducible-2022/index.md @@ -0,0 +1,83 @@ +--- +title: Reproducible brain-wide association studies require thousands of individuals +authors: +- Scott Marek +- Brenden Tervo-Clemmens +- Finnegan J. Calabro +- David F. Montez +- Benjamin P. Kay +- Alexander S. Hatoum +- Meghan Rose Donohue +- William Foran +- Ryland L. Miller +- Timothy J. Hendrickson +- Stephen M. Malone +- Sridhar Kandala +- Eric Feczko +- Oscar Miranda-Dominguez +- Alice M. Graham +- Eric A. Earl +- Anders J. Perrone +- Michaela Cordova +- Olivia Doyle +- Lucille A. Moore +- Gregory M. Conan +- Johnny Uriarte +- Kathy Snider +- Benjamin J. Lynch +- James C. Wilgenbusch +- Thomas Pengo +- Angela Tam +- Jianzhong Chen +- Dillan J. Newbold +- Annie Zheng +- Nicole A. Seider +- Andrew N. Van +- Athanasia Metoki +- Roselyne J. Chauvin +- Timothy O. Laumann +- Deanna J. Greene +- Steven E. Petersen +- Hugh Garavan +- Wesley K. Thompson +- Thomas E. Nichols +- B. T. Thomas Yeo +- Deanna M. Barch +- Beatriz Luna +- Damien A. Fair +- Nico U. F. Dosenbach +date: '2022-03-01' +publishDate: '2024-08-01T09:27:33.822524Z' +publication_types: +- article-journal +publication: '*Nature*' +doi: 10.1038/s41586-022-04492-9 +abstract: Magnetic resonance imaging (MRI) has transformed our understanding of the + human brain through well-replicated mapping of abilities to specific structures + (for example, lesion studies) and functions1–3 (for example, task functional MRI + (fMRI)). Mental health research and care have yet to realize similar advances from + MRI. A primary challenge has been replicating associations between inter-individual + differences in brain structure or function and complex cognitive or mental health + phenotypes (brain-wide association studies (BWAS)). Such BWAS have typically relied + on sample sizes appropriate for classical brain mapping4 (the median neuroimaging + study sample size is about 25), but potentially too small for capturing reproducible + brain–behavioural phenotype associations5,6. Here we used three of the largest neuroimaging + datasets currently available—with a total sample size of around 50,000 individuals—to + quantify BWAS effect sizes and reproducibility as a function of sample size. BWAS + associations were smaller than previously thought, resulting in statistically underpowered + studies, inflated effect sizes and replication failures at typical sample sizes. + As sample sizes grew into the thousands, replication rates began to improve and + effect size inflation decreased. More robust BWAS effects were detected for functional + MRI (versus structural), cognitive tests (versus mental health questionnaires) and + multivariate methods (versus univariate). Smaller than expected brain–phenotype + associations and variability across population subsamples can explain widespread + BWAS replication failures. In contrast to non-BWAS approaches with larger effects + (for example, lesions, interventions and within-person), BWAS reproducibility requires + samples with thousands of individuals. +tags: +- Cognitive neuroscience +- Psychology +links: +- name: URL + url: https://www.nature.com/articles/s41586-022-04492-9 +--- diff --git a/content/publication/rosenblatt-data-2024-1/cite.bib b/content/publication/rosenblatt-data-2024-1/cite.bib new file mode 100644 index 0000000..88ba982 --- /dev/null +++ b/content/publication/rosenblatt-data-2024-1/cite.bib @@ -0,0 +1,18 @@ +@article{rosenblatt_data_2024-1, + abstract = {Abstract +Predictive modeling is a central technique in neuroimaging to identify brain-behavior relationships and test their generalizability to unseen data. However, data leakage undermines the validity of predictive models by breaching the separation between training and test data. Leakage is always an incorrect practice but still pervasive in machine learning. Understanding its effects on neuroimaging predictive models can inform how leakage affects existing literature. Here, we investigate the effects of five forms of leakage–involving feature selection, covariate correction, and dependence between subjects–on functional and structural connectome-based machine learning models across four datasets and three phenotypes. Leakage via feature selection and repeated subjects drastically inflates prediction performance, whereas other forms of leakage have minor effects. Furthermore, small datasets exacerbate the effects of leakage. Overall, our results illustrate the variable effects of leakage and underscore the importance of avoiding data leakage to improve the validity and reproducibility of predictive modeling.}, + author = {Rosenblatt, Matthew and Tejavibulya, Link and Jiang, Rongtao and Noble, Stephanie and Scheinost, Dustin}, + doi = {10.1038/s41467-024-46150-w}, + file = {Rosenblatt et al. - 2024 - Data leakage inflates prediction performance in co.pdf:/home/alpron/Zotero/storage/P3NIEWBK/Rosenblatt et al. - 2024 - Data leakage inflates prediction performance in co.pdf:application/pdf}, + issn = {2041-1723}, + journal = {Nature Communications}, + language = {en}, + month = {February}, + number = {1}, + pages = {1829}, + title = {Data leakage inflates prediction performance in connectome-based machine learning models}, + url = {https://www.nature.com/articles/s41467-024-46150-w}, + urldate = {2024-07-15}, + volume = {15}, + year = {2024} +} diff --git a/content/publication/rosenblatt-data-2024-1/index.md b/content/publication/rosenblatt-data-2024-1/index.md new file mode 100644 index 0000000..8b33a20 --- /dev/null +++ b/content/publication/rosenblatt-data-2024-1/index.md @@ -0,0 +1,33 @@ +--- +title: Data leakage inflates prediction performance in connectome-based machine learning + models +authors: +- Matthew Rosenblatt +- Link Tejavibulya +- Rongtao Jiang +- Stephanie Noble +- Dustin Scheinost +date: '2024-02-01' +publishDate: '2024-08-01T09:27:33.794976Z' +publication_types: +- article-journal +publication: '*Nature Communications*' +doi: 10.1038/s41467-024-46150-w +abstract: Abstract Predictive modeling is a central technique in neuroimaging to identify + brain-behavior relationships and test their generalizability to unseen data. However, + data leakage undermines the validity of predictive models by breaching the separation + between training and test data. Leakage is always an incorrect practice but still + pervasive in machine learning. Understanding its effects on neuroimaging predictive + models can inform how leakage affects existing literature. Here, we investigate + the effects of five forms of leakage–involving feature selection, covariate correction, + and dependence between subjects–on functional and structural connectome-based machine + learning models across four datasets and three phenotypes. Leakage via feature selection + and repeated subjects drastically inflates prediction performance, whereas other + forms of leakage have minor effects. Furthermore, small datasets exacerbate the + effects of leakage. Overall, our results illustrate the variable effects of leakage + and underscore the importance of avoiding data leakage to improve the validity and + reproducibility of predictive modeling. +links: +- name: URL + url: https://www.nature.com/articles/s41467-024-46150-w +--- diff --git a/content/publication/soskic-garden-nodate/cite.bib b/content/publication/soskic-garden-nodate/cite.bib new file mode 100644 index 0000000..c61886a --- /dev/null +++ b/content/publication/soskic-garden-nodate/cite.bib @@ -0,0 +1,13 @@ +@article{soskic_garden_nodate, + abstract = {Abstract This study tackles the Garden of Forking Paths, as a challenge for replicability and reproducibility of ERP studies. Here, we applied a multiverse analysis to a sample ERP N400 dataset, donated by an independent research team. We analyzed this dataset using 14 pipelines selected to showcase the full range of methodological variability found in the N400 literature using systematic review approach. The selected pipelines were compared in depth by looking into statistical test outcomes, descriptive statistics, effect size, data quality, and statistical power. In this way we provide a worked example of how analytic flexibility can impact results in research fields with high dimensionality such as ERP, when analyzed using standard null-hypothesis significance testing. Out of the methodological decisions that were varied, high-pass filter cut-off, artifact removal method, baseline duration, reference, measurement latency and locations, and amplitude measure (peak vs. mean) were all shown to affect at least some of the study outcome measures. Low-pass filtering was the only step which did not notably influence any of these measures. This study shows that even some of the seemingly minor procedural deviations can influence the conclusions of an ERP study. We demonstrate the power of multiverse analysis in both identifying the most reliable effects in a given study, and for providing insights into consequences of methodological decisions.}, + author = {Šoškić, Anđela and Styles, Suzy J. and Kappenman, Emily S. and Ković, Vanja}, + doi = {https://doi.org/10.1111/psyp.14628}, + journal = {Psychophysiology}, + keywords = {to read}, + note = {_eprint: https://onlinelibrary.wiley.com/doi/pdf/10.1111/psyp.14628}, + number = {n/a}, + pages = {e14628}, + title = {Garden of forking paths in ERP research – Effects of varying pre-processing and analysis steps in an N400 experiment}, + url = {https://onlinelibrary.wiley.com/doi/abs/10.1111/psyp.14628}, + volume = {n/a} +} diff --git a/content/publication/soskic-garden-nodate/index.md b/content/publication/soskic-garden-nodate/index.md new file mode 100644 index 0000000..eedaa4d --- /dev/null +++ b/content/publication/soskic-garden-nodate/index.md @@ -0,0 +1,38 @@ +--- +title: Garden of forking paths in ERP research – Effects of varying pre-processing + and analysis steps in an N400 experiment +authors: +- Anđela Šoškić +- Suzy J. Styles +- Emily S. Kappenman +- Vanja Ković +date: -01-01 +publishDate: '2024-08-01T09:27:33.807759Z' +publication_types: +- article-journal +publication: '*Psychophysiology*' +doi: https://doi.org/10.1111/psyp.14628 +abstract: Abstract This study tackles the Garden of Forking Paths, as a challenge + for replicability and reproducibility of ERP studies. Here, we applied a multiverse + analysis to a sample ERP N400 dataset, donated by an independent research team. + We analyzed this dataset using 14 pipelines selected to showcase the full range + of methodological variability found in the N400 literature using systematic review + approach. The selected pipelines were compared in depth by looking into statistical + test outcomes, descriptive statistics, effect size, data quality, and statistical + power. In this way we provide a worked example of how analytic flexibility can impact + results in research fields with high dimensionality such as ERP, when analyzed using + standard null-hypothesis significance testing. Out of the methodological decisions + that were varied, high-pass filter cut-off, artifact removal method, baseline duration, + reference, measurement latency and locations, and amplitude measure (peak vs. mean) + were all shown to affect at least some of the study outcome measures. Low-pass filtering + was the only step which did not notably influence any of these measures. This study + shows that even some of the seemingly minor procedural deviations can influence + the conclusions of an ERP study. We demonstrate the power of multiverse analysis + in both identifying the most reliable effects in a given study, and for providing + insights into consequences of methodological decisions. +tags: +- to read +links: +- name: URL + url: https://onlinelibrary.wiley.com/doi/abs/10.1111/psyp.14628 +--- diff --git a/content/publication/spisak-multivariate-2023/cite.bib b/content/publication/spisak-multivariate-2023/cite.bib new file mode 100644 index 0000000..c90c983 --- /dev/null +++ b/content/publication/spisak-multivariate-2023/cite.bib @@ -0,0 +1,19 @@ +@article{spisak_multivariate_2023, + author = {Spisak, Tamas and Bingel, Ulrike and Wager, Tor D.}, + copyright = {2023 The Author(s)}, + doi = {10.1038/s41586-023-05745-x}, + file = {Full Text PDF:/home/alpron/Zotero/storage/PQNM323H/Spisak et al. - 2023 - Multivariate BWAS can be replicable with moderate .pdf:application/pdf}, + issn = {1476-4687}, + journal = {Nature}, + keywords = {Cognitive neuroscience, Learning algorithms, Neuroscience}, + language = {en}, + month = {March}, + note = {Publisher: Nature Publishing Group}, + number = {7951}, + pages = {E4--E7}, + title = {Multivariate BWAS can be replicable with moderate sample sizes}, + url = {https://www.nature.com/articles/s41586-023-05745-x}, + urldate = {2024-07-31}, + volume = {615}, + year = {2023} +} diff --git a/content/publication/spisak-multivariate-2023/index.md b/content/publication/spisak-multivariate-2023/index.md new file mode 100644 index 0000000..34f1cc6 --- /dev/null +++ b/content/publication/spisak-multivariate-2023/index.md @@ -0,0 +1,20 @@ +--- +title: Multivariate BWAS can be replicable with moderate sample sizes +authors: +- Tamas Spisak +- Ulrike Bingel +- Tor D. Wager +date: '2023-03-01' +publishDate: '2024-08-01T09:27:33.815783Z' +publication_types: +- article-journal +publication: '*Nature*' +doi: 10.1038/s41586-023-05745-x +tags: +- Cognitive neuroscience +- Learning algorithms +- Neuroscience +links: +- name: URL + url: https://www.nature.com/articles/s41586-023-05745-x +--- diff --git a/content/publication/tendler-why-2023/cite.bib b/content/publication/tendler-why-2023/cite.bib new file mode 100644 index 0000000..3991ef7 --- /dev/null +++ b/content/publication/tendler-why-2023/cite.bib @@ -0,0 +1,17 @@ +@article{tendler_why_2023, + abstract = {A lab handbook is a flexible document that outlines the ethos of a research lab or group. A good handbook will outline the different roles within the lab, explain what is expected of all lab members, provide an overview of the culture the lab aims to create, and describe how the lab supports its members so that they can develop as researchers. Here we describe how we wrote a lab handbook for a large research group, and provide resources to help other labs write their own handbooks.}, + author = {Tendler, Benjamin C and Welland, Maddie and Miller, Karla L and The WIN Handbook Team}, + doi = {10.7554/eLife.88853}, + file = {Tendler et al. - 2023 - Why every lab needs a handbook.pdf:/home/alpron/Zotero/storage/75L3MYKQ/Tendler et al. - 2023 - Why every lab needs a handbook.pdf:application/pdf}, + issn = {2050-084X}, + journal = {eLife}, + keywords = {best practices, to read}, + language = {en}, + month = {July}, + pages = {e88853}, + title = {Why every lab needs a handbook}, + url = {https://elifesciences.org/articles/88853}, + urldate = {2024-08-01}, + volume = {12}, + year = {2023} +} diff --git a/content/publication/tendler-why-2023/index.md b/content/publication/tendler-why-2023/index.md new file mode 100644 index 0000000..36d1788 --- /dev/null +++ b/content/publication/tendler-why-2023/index.md @@ -0,0 +1,26 @@ +--- +title: Why every lab needs a handbook +authors: +- Benjamin C Tendler +- Maddie Welland +- Karla L Miller +- The WIN Handbook Team +date: '2023-07-01' +publishDate: '2024-08-01T09:27:33.847482Z' +publication_types: +- article-journal +publication: '*eLife*' +doi: 10.7554/eLife.88853 +abstract: A lab handbook is a flexible document that outlines the ethos of a research + lab or group. A good handbook will outline the different roles within the lab, explain + what is expected of all lab members, provide an overview of the culture the lab + aims to create, and describe how the lab supports its members so that they can develop + as researchers. Here we describe how we wrote a lab handbook for a large research + group, and provide resources to help other labs write their own handbooks. +tags: +- best practices +- to read +links: +- name: URL + url: https://elifesciences.org/articles/88853 +---