ML in MS Research via Administrative Data

Predictive Research question

Machine learning in prediction modelling: How can machine learning methods be used in the development of a comorbidity index?

Comorbidities play an important role in a person’s disease course and prognosis, often contributing to worse outcomes. A comorbidity index, which summarizes a person’s total comorbidity burden, represents an important tool for confounder adjustment in observational studies. Machine learning methods can be useful in the development of comorbidity indices that are predictive of relevant outcomes. Through the example of a comorbidity index tailored to people with multiple sclerosis (Frank et al. 2023), we will showcase how machine learning methods can be used in the development of such an index. We will also highlight how the process using machine learning compares to that using traditional regression approaches and discuss the benefits and drawbacks to each approach.

References

Frank, Hanna A, Melissa Chao, Helen Tremlett, Ruth Ann Marrie, Lisa M Lix, and Mohammad Ehsanul Karim. 2023. “The Impact of Comorbidities on Outcomes in the Multiple Sclerosis Population: A Rapid Review Protocol.” https://doi.org/10.21203/rs.3.pex-2438/v1.