Case Study: Predictive Modeling for Treatment of Relapsing-Remitting Multiple Sclerosis (RRMS)

The Situation

Early recognition of the progression from relapsing-remitting multiple sclerosis (RRMS) to secondary progressive MS (SPMS) remains challenging. EVERSANA has conducted a study utilizing data from administrative claims to employ predictive models that determine disease progression and appropriate disease modifying therapies (DMTs) for patients based on a multitude of factors, including demographics, disease characteristics and clinical patterns.

A midsize pharmaceutical company needed to recognize the disease progression of RRMS and identify therapeutic interventions to slow its progression. EVERSANA partnered with this client and used machine learning (ML) to build predictive models using an observational study leveraging a Kuwaiti MS clinical registry (N=2,265 patients). These ML-driven predictive models needed to identify risk factors associated with progression from RRMS to SPMS, as well as the role of specific DMTs in delaying the progression.

The Solution

The project faced several challenges, including the complexity of MS as a disease with diverse manifestations and progression rates among patients. Additionally, compiling comprehensive and accurate patient data posed significant hurdles, requiring the validation and integration of information from various sources. The variability in treatment response factors also added layers of complexity to model development and required sophisticated analytical techniques.

EVERSANA’s approach involved a multiphase strategy starting with the collection and integration of detailed patient data from Kuwait’s national MS registry. Following data integration, predictive models were developed using advanced ML techniques, including gradient boosting and SHAP values, to determine the impact of different variables on treatment efficacy and understand the rationale for a patient progressing to a second-line treatment. The models were then rigorously tested and validated to ensure their reliability in clinical decision-making processes.

The Results

Based on the output of these models, key predictors of progression were identified. These included the duration of symptoms, age, a high frequency of an expanded disability status scale (EDSS) score greater than 3 and more than three lesions on a recent MRI.

A high-efficacy DMT used as first-line therapy played a key role in delaying progression. Additionally, patients older than 30 were helped the most by switching to a highly effective DMT as a second-line therapy.

The client was able to utilize these predictive models developed by EVERSANA to recognize MS patients who were potentially transitioning from RRMS to SPMS and inform treatment modifications to mitigate disease progression. These models can predict treatment needs with high accuracy, thus ensuring patients receive the best therapy sooner. Our findings provided the client with a robust foundation for the future of personalized medicine in managing patients with MS, potentially affecting clinical practice.