Usability in mobile health applications: Eye tracking in cognitive load evaluation

Objective: This study examines the usability of the mobile health application e-Nabız with eye tracking, traditional and model-based methods, and the relationship between these methods was investigated. As a result of the research, it is aimed to contribute to a more effective investigation of the cognitive load imposed on users by mobile applications by obtaining important information about the effect of cognitive load on user performance. Method: The cognitive load that occurs during the use of the mobile health application e-Nabız was estimated with the GOMS model and data on perceived cognitive load were obtained through eye tracking, NASA-RTLX, SUS and retrospective interviews in user tests. The obtained cognitive load data were compared in terms of demographic information (user experience, e-Nabız usage frequency, educational status and gender), while the relationship between the methods used was examined. Results: A strong relationship was found between performance and GOMS and NASA-RTLX in estimating cognitive load. While a negative relationship was found between performance and fixation rate and a positive relationship between performance and fixation duration, no significant relationship was found between performance and blinks per minute. A positive strong relationship was found between the GOMS model and perceived cognitive load, and a negative strong relationship was found between GOMS and blinks per minute, but no significant relationship was found between GOMS and fixation rate and fixation duration. While there was a strong negative relationship between perceived cognitive load and the number of blinks per minute, no significant relationship was observed between perceived cognitive load and fixation duration and fixation rate. Retrospective interviews with users, SUS results and researcher observation support the cognitive load data obtained in this study. Conclusion: This study revealed that there is a strong relationship between estimating cognitive load by modeling method and measuring it by user tests, and one or more of these methods can be used depending on their suitability for the system to be evaluated.