Teague Henry

A Monte Carlo evaluation of weighted community detection algorithms

The past decade has been marked with a proliferation of community detection algorithms that aim to organize nodes (e.g., individuals, brain regions, variables) into modular structures that indicate subgroups, clusters, or communities. Motivated by the emergence of big data across many fields of inquiry, these methodological developments have primarily focused on the detection of communities of nodes from matrices that are very large. However, it remains unknown if the algorithms can reliably detect communities in smaller graph sizes (i.e., 1000 nodes and fewer) which are commonly used in brain research. More importantly, these algorithms have predominantly been tested only on binary or sparse count matrices and it remains unclear the degree to which the algorithms can recover community structure for different types of matrices, such as the often used cross-correlation matrices representing functional connectivity across predefined brain regions. Of the publicly available approaches for weighted graphs that can detect communities in graph sizes of at least 1000, prior research has demonstrated that Newmans spectral approach (i.e., Leading Eigenvalue), Walktrap, Fast Modularity, the Louvain method (i.e., multilevel community method), Label Propagation, and Infomap all recover communities exceptionally well in certain circumstances. The purpose of the present Monte Carlo simulation study is to test these methods across a large number of conditions, including varied graph sizes and types of matrix (sparse count, correlation, and reflected Euclidean distance), to identify which algorithm is optimal for specific types of data matrices. The results indicate that when the data are in the form of sparse count networks (such as those seen in diffusion tensor imaging), Label Propagation and Walktrap surfaced as the most reliable methods for community detection. For dense, weighted networks such as correlation matrices capturing functional connectivity, Walktrap consistently outperformed the other approaches for recovering communities.

An Exploratory Analysis of Personality, Attitudes, and Study Skills on the Learning Curve within a Team-Based Learning Environment

Objective. To examine factors that determine the interindividual variability of learning within a team-based learning environment. Methods. Students in a pharmacokinetics course were given 4 interim, low-stakes cumulative assessments throughout the semester and a cumulative final examination. Students’ Myers-Briggs personality type was assessed, as well as their study skills, motivations, and attitudes towards team-learning. A latent curve model (LCM) was applied and various covariates were assessed to improve the regression model. Results. A quadratic LCM was applied for the first 4 assessments to predict final examination performance. None of the covariates examined significantly impacted the regression model fit except metacognitive self-regulation, which explained some of the variability in the rate of learning. There were some correlations between personality type and attitudes towards team learning, with introverts having a lower opinion of team-learning than extroverts. Conclusion. The LCM could readily describe the learning curve. Extroverted and introverted personality types had the same learning performance even though preference for team-learning was lower in introverts. Other personality traits, study skills, or practice did not significantly contribute to the learning variability in this course.

Concordance networks and application to clustering cancer symptomology

Symptoms of complex illnesses such as cancer often present with a high degree of heterogeneity between patients. At the same time, there are often core symptoms that act as common drivers for other symptoms, such as fatigue leading to depression and cognitive dysfunction. These symptoms are termed bridge symptoms and when combined with heterogeneity in symptom presentation, are difficult to detect using traditional unsupervised clustering techniques. This article develops a method for identifying patient communities based on bridge symptoms termed concordance network clustering. An empirical study of breast cancer symptomatology is presented, and demonstrates the applicability of this method for identifying bridge symptoms.

Social Position Predicting Physical Activity Level in Youth: An Application of Hidden Markov Modeling on Network Statistics

Social positioning has been shown to have impacts on physical activity in youth. In this study Hidden Markov Modeling is used to infer latent social positions from a set of computed network statistics in two network of youth over time. The association between physical activity and social position is analyzed. Youth in less centrally located social roles tended to have less physical activity than youth with more centrally located social positions.