Longitudinal data analysis with statistical and machine learning methods in neuroscience

Download

SerenayCakar_Master_s_Thesis.pdf

Date

2022-8

Author

Çakar, Serenay

Metadata

Show full item record

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

Item Usage Stats

521
views

468
downloads

Exploration of brain activity under different conditions has been subject to many neuroscience studies. The recent developments in cognitive studies provide the opportunity to work on neural correlates of specific cognitive processes such as working memory, decision making, response inhibition, perception, and sensation. Brain response studies constitute multidimensional, multilevel or nested data sets formed by different parts of the brain of individuals. Hence, it is of significant importance to implement data analysis methods appropriate for the longitudinal structures. However, previous studies on brain response utilized methods that do not consider the dependency, multilevel and nested structure. In this thesis, we propose to apply different statistical and machine learning methods on cognitive data to fill the aforementioned deficiencies. We analyze open-access data, including optical density measures collected from 36 locations of the brain within 26 subjects through functional near-infrared spectroscopy (fNIRS). fNIRS signals are used to measure relative changes in oxyhemoglobin and deoxyhemoglobin concentrations. The nested structure of the data, which is having observations from different brain regions within subjects, is also considered. The content of this thesis provides a comprehensive implementation and comparison of several statistical and machine learning algorithms which are Linear Mixed Model (LMM) and its robustified version, Generalized LMM Tree (GLMM tree), Random Effects Expectation-Maximization Tree (RE-EM tree), Unbiased RE-EM tree, Longitudinal Classification and Regression Tree, and Gaussian Process Boosting. According to one of our findings, the GLMM tree with nested structure shows the best predictive performance as it provides the lowest model performance metrics. However, there is a trade-off between accuracy and speed since the speed of this algorithm is lower compared to other methods except robustified LMM.

Subject Keywords

Cognitive Studies, fNIRS, N-back Data, Linear Mixed Model, Robust Modeling, Machine Learning Algorithms

URI

https://hdl.handle.net/11511/98590

Collections

Graduate School of Natural and Applied Sciences, Thesis

Suggestions

OpenMETU
Core

Hybrid statistical and machine learning modeling of cognitive neuroscience data Çakar, Serenay; Gökalp Yavuz, Fulya (2023-01-01) The nested data structure is prevalent for cognitive measure experiments due to repeatedly taken observations from different brain locations within subjects. The analysis methods used for this data type should consider the dependency structure among the repeated measurements. However, the dependency assumption is mainly ignored in the cognitive neuroscience data analysis literature. We consider both statistical, and machine learning methods extended to repeated data analysis and compare distinct algorithms ...
Mood, Memory, and Motor Performance and the Severity of Tardive Dyskinesia Gilleard, C. J.; Vaddadi, KS (SAGE Publications, 1986-12) <jats:p> This study tested the hypothesis that the features of tardive dyskinesia were associated with motor slowing, memory impairment, and depressive apathy all of which are considered to characterize the so-called subcortical dementias. In a sample of 48 psychiatric patients all fulfilling research criteria for tardive dyskinesia and without other signs of organic illness age-independent correlations were observed between severity of orofacial dyskinesia and measures of memory, motor performance and mood...
Toward a new understanding of virtual research collaborations: Complex adaptive systems framework Aydınoğlu, Arsev Umur (2013-01-01) Virtual research collaborations (VRCs) have become an important method of conducting scientific activity; however, they are often regarded and treated as traditional scientific collaborations. Their success is measured by scholarly productivity and adherence to budget by funding agencies, participating scientists, and scholars. VRCs operate in complex environments interacting with other complex systems. A holistic (or organicist) approach is needed to make sense of this complexity. For that purpose, this st...
Systemizing: A Cross-Cultural Constant for Motivation to Learn Science Zeyer, Albert; Cetin-Dindar, Ayla; Zain, Ahmad Nurulazam Md; Jurisevic, Mojca; Devetak, Iztok; Odermatt, Freia (Wiley, 2013-11-01) The present study is based on the empathizing-systemizing (E-S) theory of cognitive science. It was hypothesized that the influence of students' gender on their motivation to learn science is often overestimated in the research literature and that cognitive style is more important formotivation than students' gender. By using structural equation modeling, and based on previous research, a precise causal model was formulated to test this hypothesis. Then, using multiple group confirmatory analysis, the model...
Parallel implementation of the boundary element method for electromagnetic source imaging of the human brain Ataseven, Yoldaş; Gençer, Nevzat Güneri; Department of Electrical and Electronics Engineering (2005) Human brain functions are based on the electrochemical activity and interaction of the neurons constituting the brain. Some brain diseases are characterized by abnormalities of this activity. Detection of the location and orientation of this electrical activity is called electro-magnetic source imaging (EMSI) and is of signi cant importance since it promises to serve as a powerful tool for neuroscience. Boundary Element Method (BEM) is a method applicable for EMSI on realistic head geometries that generates...

Citation Formats

S. Çakar, “Longitudinal data analysis with statistical and machine learning methods in neuroscience,” M.S. - Master of Science, Middle East Technical University, 2022.