Multi-morbidity is common in persons with spinal cord injury (SCI). Network Analysis is a tool used to visualize and estimate complex relationships among variables. Three network models: Gaussian Graphical Model, Ising model, and Mixed Graphical Model were applied to the 2011-2012 Canadian SCI Community Survey dataset, which included individuals with traumatic and non-traumatic SCI. Data utilized included demographic and injury data as we well as 30 secondary health conditions (comorbidities and secondary complication) that are included in the Multi-Morbidity Index (MMI-30). Five health outcomes were included: healthcare utilization (HCU), health status (i.e. Short Form-12 physical and mental component summary (SF-12 PCS & MCS) score), life satisfaction, and quality of life. Using Network Analysis, we reduced the number of items in the Multi-Morbidity Index (MMI-30) by 5 items (MMI-25) and the psychometric properties were comparable.
This interactive workshop will include presentations from a clinician, researcher and person with lived experience (PLEX).
The goals of this workshop are to:
Review Machine Learning methodology (with an emphasis on Network Analysis) in SCI research.
Demonstrate how Network Analysis was used to investigate the relationship between secondary health conditions and health outcomes to create a modified version of the Multi-Morbidity Index.
Discuss future opportunities for using Network Analysis and other Machine Learning methodologies in SCI research and provide examples of how these knowledge products can be used to inform SCI care.
This workshop will demonstrate the benefit of using Network Analysis, a type of Machine Learning, in SCI research. Specifically, the example of how Network Analysis identified key associations among 30 secondary health conditions and five health outcomes which resulted in the MMI-25 will be discussed as well as future opportunities for using Network Analysis and other Machine Learning methodologies in SCI research.
