Machine Learning Evaluation of Multiple Sclerosis Patients During the Time of COVID-19
Sharon Stoll1, Noah Rubin2, Tess Litchman3, john keaney4, lawrence rubin2, Timothy Vartanian5, Charisse Litchman1
1Yale school of medicine, 2Becare MS Link, 3Beth Israel, 4Yale New Haven Hospital, 5Weill Cornell Medical College
Objective:

To establish correlation of a clinical EDSS with BeCare’s app derived EDSS.

Background:

The COVID-19 pandemic has highlighted the need for remote assessments of neurologic function in MS patients. Quantitative assessments of neurologic function, such as the Expanded Disability Status Scale (EDSS) routinely inform clinical decision-making in MS clinics. The BeCare app was developed to serve not only as a digital equivalent for the EDSS, but also as a tool to monitor disability progression remotely.

Design/Methods:

Subjects were categorized by EDSS of Control (0), Mild (1.0- 3.0), Moderate (3.5 – 6.0) or Severe [6.5 – 9.0]. EDSS was established at baseline and at 3 month followup. Clinical scoring was compared to machine learning algorithm (MLA)  obtained using Mobile App-derived data collected during the office visit.  Of the 25 enrolled patients, 17 patients completed all app functions at baseline and 14 at followup. 

Results:
The Chi-Squared test, used for the statistical analysis, assumes that observed frequencies (MLA) for a categorical variable match the expected frequencies (clinical). For a Chi-square test, a P-value that is greater than the significance level indicates there is sufficient evidence to conclude that the observed distribution is the same as the expected distribution based on the clinical EDSS scores. At the first study collection using a confidence interval of 95% resulted in a Xi-Squared value of 6.251 which is less than the critical value of 7.815 with a P-value of 0.08. At the second study collection, using a confidence interval of 95% resulted in a Xi-Squared value of 1.905 which is less than the critical value of 7.815 along with a P-value of 0.592
Conclusions:

The BeCare App closely replicates clinical assessment of disease severity (categorized as mild, moderate, or severe). Remote monitoring for disease progression will serve to advance treatment of MS patients during and after the Pandemic.