Smartphone-Based Symbol-Digit Modalities Test Reliably Measures Cognitive Function in Multiple Sclerosis Patients

Limited time for patient encounters prevents reliable evaluation of all neurological functions in routine clinical practice. Quantifying neurological disability in a patient-autonomous manner via smartphones may remedy this problem, if such tests provide reliable, disease-relevant information. We developed a smartphone version of the cognitive processing speed test, the Symbol-Digit Modalities Test (SDMT), and assessed its clinical utility. The traditional SDMT uses identical symbol-number codes, allowing memorization after repeated trials. In the phone app, the symbol-number codes are randomly generated. In 154 multiple sclerosis (MS) patients and 39 healthy volunteers (HV), traditional and smartphone SDMT have good agreement (Lin's coefficient of concordance [CCC] = 0.84) and comparable test-retest variance. In subjects with available volumetric MRI and digitalized neurological examinations (112 MS, 12 HV), the SDMT scores were highly associated with T2 lesion load and brain parenchymal fraction, when controlled for relevant clinical characteristics. The smartphone SDMT association with clinical/imaging features was stronger (R2 = 0.75, p < 0.0001) than traditional SDMT (R2 = 0.65, p < 0.0001). In the longitudinal subcohort, improvements from testing repetition (learning effects), were identifiable using non-linear regression in 14/16 subjects and, on average, peaked after 8 trials. Averaging several post-learning SDMT results significantly lowers the threshold for detecting true decline in test performance. In conclusion, smartphone, self-administered SDMT is a reliable substitute of the traditional SDMT for measuring processing speed in MS patients. Granular measurements at home increase sensitivity to detect true performance decline in comparison to sporadic assessments in the clinic.