Evaluating the Reliability of Neuro-QWERTY Index, a Computer-Based Method for Early Diagnostic of Parkinson's Disease

Early diagnose of Parkinson's disease has been possible by using neuro-QWERTY index, a computational algorithm for analyzing users’ interactions with a computer keyboard. This research was devoted to study the reliability of neuro-QWERTY index; as detecting the reliability is necessary to clear the usefulness of diagnostic and screening tests.
Early Parkinson's disease database, which was used in aggregated format by Giancardo et al. in 2016 to introduce neuro-QWERTY index, and to assess its validity, was used in this study in original format to assess the test-retest reliability of neuro-QWERTY index. Intraclass correlation coefficient (ICC), Bland-Altman plot, and coefficient of repeatability (CR) were used for statistical analyses. Moreover, using DeLong’s statistical test, area under curves (AUCs) of receiver operating curves (ROCs) was compared for test and retest scores, and their average scores.
Findings: Intraclass correlation coefficient was 0.94, with a 95% confidence interval (CI) of 0.89 to 0.97. Bland-Altman plot showed that differences between test and retest scores were small. Coefficient of repeatability was 0.04, with a 95% CI of 0.03 to 0.05. The area under curves were 0.89, 0.90, and 0.92 for receiver operating curves of test score, retest score, and their average score, respectively. Differences between the areas under curves of receiver operating curves were not significant (P ≥ 0.25).
Neuro-QWERTY index have a good reliability. Therefore, it could act as a suitable candidate for screening of Parkinson's disease, considering simplicity and inexpensiveness of its measurements, no side effects, good validity in terms of sensitivity and specificity, and as shown in this study, its acceptable reliability.