D Gage Jordan1, E Samuel Winer1, Taban Salem2. 1. Department of Psychology, Mississippi State University, Starkville, Mississippi. 2. Harding Hospital, The Ohio State University Wexner Medical Center, Columbus, Ohio.
Abstract
OBJECTIVE: Network analysis in psychology has ushered in a potentially revolutionary way of analyzing clinical data. One novel methodology is in the construction of temporal networks, models that examine directionality between symptoms over time. This paper provides context for how these models are applied to clinically-relevant longitudinal data. METHODS: We provide a survey of statistical and methodological issues involved in temporal network analysis, providing a description of available estimation tools and applications for conducting such analyses. Further, we provide supplemental R code and discuss simulations examining temporal networks that vary in sample size, number of variables, and number of time points. RESULTS: The following packages and software are reviewed: graphicalVAR, mlVAR, gimme, SparseTSCGM, mgm, psychonetrics, and the Mplus dynamic structural equation modeling module. We discuss the utility each procedure has for specific design considerations. CONCLUSION: We conclude with notes on resources for estimating these models, emphasizing how temporal networks best approximate network theory.
OBJECTIVE: Network analysis in psychology has ushered in a potentially revolutionary way of analyzing clinical data. One novel methodology is in the construction of temporal networks, models that examine directionality between symptoms over time. This paper provides context for how these models are applied to clinically-relevant longitudinal data. METHODS: We provide a survey of statistical and methodological issues involved in temporal network analysis, providing a description of available estimation tools and applications for conducting such analyses. Further, we provide supplemental R code and discuss simulations examining temporal networks that vary in sample size, number of variables, and number of time points. RESULTS: The following packages and software are reviewed: graphicalVAR, mlVAR, gimme, SparseTSCGM, mgm, psychonetrics, and the Mplus dynamic structural equation modeling module. We discuss the utility each procedure has for specific design considerations. CONCLUSION: We conclude with notes on resources for estimating these models, emphasizing how temporal networks best approximate network theory.
Authors: Corrado Sandini; Daniela Zöller; Maude Schneider; Anjali Tarun; Marco Armando; Barnaby Nelson; Paul G Amminger; Hok Pan Yuen; Connie Markulev; Monica R Schäffer; Nilufar Mossaheb; Monika Schlögelhofer; Stefan Smesny; Ian B Hickie; Gregor Emanuel Berger; Eric Yh Chen; Lieuwe de Haan; Dorien H Nieman; Merete Nordentoft; Anita Riecher-Rössler; Swapna Verma; Andrew Thompson; Alison Ruth Yung; Patrick D McGorry; Dimitri Van De Ville; Stephan Eliez Journal: Elife Date: 2021-09-27 Impact factor: 8.140
Authors: Robin N Groen; Oisín Ryan; Johanna T W Wigman; Harriëtte Riese; Brenda W J H Penninx; Erik J Giltay; Marieke Wichers; Catharina A Hartman Journal: BMC Med Date: 2020-09-29 Impact factor: 8.775