An Lab
Xinming An, PhD
The An Lab primarily focuses on developing statistical models and computational techniques/algorithms for complex multimodal high dimensional temporal-spatial data analysis and their applications for studying adverse posttraumatic neuropsychiatric sequelae (APNS), such as pain and PTSD.
These applications include: 1) Development of risk prediction tools, derivation and validation of objective biomarkers; and 2) Identification of subtypes of APNS with homogenous structures/relationships.
Over the past year, the An Lab has made great progress and significant contributions to the field.
1) Heterogeneous Causal Graph:
Working with collaborators from NCSU, we developed an innovative statistical approach to learn and test heterogeneous/individual casual graphs based on the Structural equation model (SEM) framework. The corresponding manuscript has been selected as one of the winners of the 2023 Mental Health Statistics Section Student Paper Award competition and was accepted by one of the top machine learning conferences (ICML 2023).
2) Study multiple APNS outcomes:
Working with collaborators from NCSU, we are developing an innovative statistical approach to investigate risk factors and develop risk prediction tools of multiple APNS outcomes (e.g., pain, somatic symptom, PTSD and depression) based on high dimensional cross-sectional and intensive longitudinal data.
3) Understanding onset, transition and associated risk factors for APNS after trauma exposure:
Based on serial assessments of symptoms from a large cohort study, we identified homogeneous statuses across multiple APNS symptom domains and investigated the dynamic transitions among these statuses during the first two months after trauma exposure. Furthermore, we studied how symptom onset and transitions are affected by equity-relevant characteristics. This study advances our current knowledge about the onset and dynamic transitions of APNS within the first two months after trauma exposure. The observed dynamic prognosis of APNS and associated risk factors can help develop effective preventive interventions for APNS and inform efforts to mitigate health disparities among trauma survivors.
4) Investigating the association of cell-type specific methylation with APNS:
We are working with collaborators on a project to investigate the association of cell-type specific methylation with APNS outcomes (e.g., PTSD, depression and pain). By using an innovative computational technique, this project studies methylation signals at cell-type-specific resolution (individuals by methylation sites by cell types) from tissue-level bulk data (individuals by methylation sites).