Natural language processing analysis of the theories of people with multiple sclerosis about causes of their disease

While potential risk factors for multiple sclerosis (MS) have been extensively researched, it remains unclear how persons with MS theorize about their MS. Such theories may affect mental health and treatment adherence. Using natural language processing techniques, we investigated large-scale text data about theories that persons with MS have about the causes of their disease. We examined the topics into which their theories could be grouped and the prevalence of each theory topic.

A total of 486 participants of the Swiss MS Registry longitudinal citizen science project provided text data on their theories about the etiology of MS. We used the transformer-based BERTopic Python library for topic modeling to identify underlying topics. We then conducted an in-depth characterization of the topics and assessed their prevalence.

The topic modeling analysis identifies 19 distinct topics that participants theorize as causal for their MS. The topics most frequently cited are Mental Distress (31.5%), Stress (Exhaustion, Work) (29.8%), Heredity/Familial Aggregation (27.4%), and Diet, Obesity (16.0%). The 19 theory topics can be grouped into four high-level categories: physical health (mentioned by 56.2% of all participants), mental health (mentioned by 53.7%), risk factors established in the scientific literature (genetics, Epstein-Barr virus, smoking, vitamin D deficiency/low sunlight exposure; mentioned by 47.7%), and fate/coincidence (mentioned by 3.1%). Our study highlights the importance of mental health issues for theories participants have about the causes of their MS.

Our findings emphasize the importance of communication between healthcare professionals and persons with MS about the pathogenesis of MS, the scientific evidence base and mental health.