Machine Learning Team

National Institute Of Mental Health

Investigators

Linked publications & trials

Abstract

This is the sixth annual report for the Machine Learning Team. The team comprises research scientists Francisco Pereira, Charles Zheng, Dylan Nielson, Ka Chun Lam, Yuan Zhao,Juan Antonio Lossio-Ventura, and Gabriel Loewinger. Postbac IRTA Al Xin joined the team in June. We work very closely with the Data Science and Sharing Team (DSST), both on specific joint projects and informally across many others. Research activity: We continued IRP research projects initiated in the previous year. These were the projects that led to the main published articles: 1) Investigators: Javier Gonzalez-Castillo and Peter Bandettini. The first goal of this project was to introduce different dimensionality reduction methods to researchers analyzing time-varying functional connectivity data. The second goal was to show how different choices of parameters in these methods could lead to enlightening (or artifactual) apparent structure, and provide guidelines on how researchers should approach setting them. 2) Investigators: Dave Jangraw and Argyris Stringaris. The goal of this project was to analyze imaging and behavioral data from many different experiments, to show that the mood of participants drops consistently during rest or performance of simple tasks. The relevance of this is as a potential confound in experiments measuring participant mood. 3) Investigators: Lucrezia Liuzzi, Hanna Keren, and Argyris Stringaris. Goal: To develop models to predict mood during a gambling task, based on participant characteristics, trial parameters, and experiment history. This project was extended to magnetoencephalography (MEG) data. 4) Investigators: MLT + DSST. We developed a new method for extracting brain maps from a deep neural network trained to decode which task is being performed by a subject. The maps show where in the brain the the information used by the network is being processed . The article introduces two different quantitative evaluation procedures for gauging the quality of the maps and compares our method with existing ones 5) Investigators: Martin Hebart (now an external collaborator) and Lukas Muttenthaler (Martins student) discovered interpretable mental representations of objects from a large database of behavioral judgments, which can be used to predict human behavior on a variety of other tasks. We developed an improved version of the method, using approximate Bayesian inference, which should perform better in smaller datasets (such as those collected by the NIMH IRPs Christopher Baker and Alex Martin, who are using the method). 6) Investigators: Lauren Atlas and Joyce Chung. This is the first in a series of papers analyzing data from the longitudinal COVID-19 survey of mental well-being and psychological and behavioral responses during lockdown. The goal of this paper was to analyze the free text answers provided by participants, including plots of emotional valence over time and summaries of the main topics present in such text. 7) Investigators: Neda Sadeghi and Argyris Stringaris. The goal of this project was to investigate whether, compared to pre-pandemic levels, depressive and anxiety symptoms in adolescents with depression increased during the pandemic. 8) Investigators: Martin Hebart and Chris Baker. The goal of this paper was to introduce THINGS-data, a multimodal collection of large-scale neuroimaging and behavioral datasets in humans, comprising densely sampled functional MRI and magnetoencephalographic recordings, as well as 4.70 million similarity judgments in response to thousands of photographic images for up to 1,854 object concepts. In addition, we had publications with external collaborators, or publications by team members that were prepared or completed after their joining the team 1) Sensory and Choice Responses in MT Distinct from Motion Encoding Aaron J. Levi, Yuan Zhao, Il Memming Park and Alexander C. Huk Journal of Neuroscience 22 March 2023, 43 (12) 2090-2103 2) Real-time variational method for learning neural trajectory and its dynamics The Eleventh International Conference on Learning Representations (ICLR), 2023 3) Linear time GPs for inferring latent trajectories from neural spike trains The Fortieth International Conference on Machine Learning (ICML), 2023 4) Clinical concept recognition: Evaluation of existing systems on EHRs Frontiers in Artificial Intelligence 5, 1051724. Several other projects led to article submissions currently under review: 1) PI: Lauren Atlas Sentiment Analysis of COVID-19 Survey Data: A Comparison of ChatGPT and Fine-tuned OPT Against Widely Used Sentiment Analysis Tools 2) PI: Joyce Chung Prediction of mental well-being from individual characteristics and circumstances during the COVID-19 pandemic 3) PI: Soohyun Lee Distinct brain-wide presynaptic networks underlie the functional identity of individual cortical neurons 4) PI: MLT A Statistical Framework for Analysis of Trial-Level Temporal Dynamics in Fiber Photometry Experiments 5) PI: Melissa Brotman Multivariate prediction of temper outbursts in youth enriched for irritability using Ecological Momentary Assessment data 6) PI: Daniel Pine Validation of CBCL depression scores of adolescents in three independent datasets 7) PI: Daniel Pine / Argyris Stringaris Subjective Affective Experience under threat is shaped by environmental affordances 8) PI: Argyris Stringaris On the misery of cognitive effort 9) PI: Argyris Stringaris Test-retest reliability of functional connectivity in depressed adolescents 10) PI: Argyris Stringaris An experimental approach to training mood for resilience 11) PI: Armin Raznahan Interpretable factorization of clinical questionnaires to identify latent factors of psychopathology" We have articles in preparation for submission on the following projects: 1) PI: Armin Raznahan (NIMH). Interpretable meta-factorization of clinical questionnaires to identify general dimensions of psychopathology 2) PI: Bruce Hope (NIDA). Analysis of neural data collected during an extinction paradigm. 6) PI: Bruno Averbeck (NIMH). Analysis of neural data collected during a bandit task paradigm Ongoing Research In addition to all of the above, we have ongoing research or service projects with researchers in the groups of Drs. Atlas, Averbeck, Baker, Bandettini, Brotman, Histed, Lee, Lisanby, Lopez-Guzman, Martin, , Penzo, Pine, Taylor, Tejeda, White, Zarate at NIMH, Schoenbaum, Hope, and Curtis at NIDA, and Lovinger at NIAAA, none of which have yet led to articles. We try to mix high-risk/high-impact projects with those satisfying specific needs of the PIs. These usually consume weeks to months of person-effort, depending on whether there exists a tool to satisfy the request, and the degree to which we have to handle the analysis process (versus advising a researcher in the group). Methods in Development We have a number of internal projects for developing new methods which, if shown to be effective, we will bring to the attention of PIs who might have an application for them. The two methods we are preparing for publication, with code release, are: 1) non-negative factorization method for questionnaire data, with additional constraints to further interpretability 2) statistical method for testing differences in time courses of photometry signals, as a function of covariates of interest (e.g. condition), accounting for confounds and individual variability.Consultation/Service activity We carried out ad-hoc consultations on machine learning methods (which can take hours to days, if they require reading articles or finding/testing code), or co-advising of postbac and postdoc trainees (days-weeks). In addition to the PIs mentioned above, we provided consulting to people in NIMH, NIDA, NINDS, NICHD, and NIAAA. Francisco Pereira participated in the IRP seminar commi