MSCopilot: New smartphone-based digital biomarkers correlate with Expanded Disability Status Scale scores in people with Multiple Sclerosis

Abstract

Background: A previous clinical study showed the high specificity, sensitivity and reliability of MSCopilot, a software medical device designed by Ad Scientiam for the self-assessment of people with Multiple Sclerosis (PwMS), compared to the traditional Multiple Sclerosis Functional Composite (MSFC). We conducted further analyses to assess MSCopilot's performance with respect to the Expanded Disability Status Scale (EDSS). Methods: The data of 116 PwMS were analysed. We studied the correlations between MSCopilot scores and the EDSS, and their ability to distinguish PwMS with high and low EDSS through a study of the distribution of the digital test scores as well as logistic regression models. The same analyses were performed using the MSFC tests. Results: MSCopilot composite scores were as highly correlated to the EDSS (|r| = 0.65, p < 0.01) as their MSFC counterparts, confirming the known correlation of the MSFC with the EDSS. In a linear regression framework, the Walking digital tests have good explanatory power, especially for PwMS with EDSS > 3.5 (R²adj=0.47). The mean values of each MSCopilot subscore were significantly different between patients with an EDSS > 3.5 and others (p < 0.05), which could not be proved for the MSFC Cognition tests. MSCopilot4 was the best model to predict an EDSS score > 3.5 (AUC = 0.92). Conclusion: These analyses confirm the reliability of MSCopilot and show interesting correlations with the EDSS (similar results obtained with the MSFC). MSCopilot was able to highlight nuances in the different stages of MS the MSFC could not capture.