The Trauma and Resilience Informatics Laboratory

Understanding the variables that contribute to an individual’s risk or resilience after trauma is complex. There are a great many variables that might be involved, including characteristics of the traumatic event, the child’s previous exposure to trauma, child developmental factors, demographics, family response to the child: and factors related to genes, gene expression, and neurophysiologic response. The literature on risk and resilience factors after trauma is rather inconsistent and robust risk and protective factors have not been identified: certainly none that can translate to clinical practice. Many investigators believe the inconsistency in the research literature relates to conventional data analytic methods that cannot approach the complexity of the problem of traumatic stress.

The Trauma and Resilience Informatics Laboratory aims to address this significant research bottleneck by developing advanced computational procedures and analytic approaches to explore very large data sets with information that includes a wide diversity of biobehavioral and social variables collected from traumatized individuals. In partnerships with the NYU Langone Center for Health Informatics and Bioinformatics, we have developed and validated novel computational algorithms for this research and we apply our algorithms, and those developed by others, for these research purposes. Our research includes:

  • The development and validation of the Complex Systems-Causal Network method: a web-based set of algorithms that uniquely integrates the methods of causal discovery with the methods of complex systems science to search data sets for a system of variables (of any type) that supports the development of traumatic stress. Our method then seeks to discover the most central variables for the systems functioning. The discovery of a variable that disproportionally contributes to system functioning may reveal a previously unknown target for intervention.
  • The application of advanced machine learning approaches to identify the set of variables within any data set that can distinguish which traumatized individuals acquire traumatic stress reactions and which do not.
  • Software development to enable our computational approaches to be available to a wide community of investigators.