Smartphone Keystroke-based Cognitive Impairment Diagnostic Methodology

With the global population rapidly aging, dementia has emerged as a major public health concern. Significant efforts are being made to address the health and other problems associated with dementia. Although several advances have been made in the diagnosis and treatment of dementia, many challenges remain unresolved. Existing dementia diagnostic methods are often not performed in a timely manner due to poor accessibility, leading to delays in initiation of treatment. In this study, we propose a smartphone keystroke-based diagnostic method as a solution to overcome the limitations associated with the early diagnosis and treatment of dementia. As compared to other existing methods, the proposed diagnostic method is easier to develop and maintain, promoting commercialization and widespread use. The diagnostic method has been designed to extract language-agnostic keystroke data features from smartphone keyboard input logs. Rather than simply extracting features assessing motor skills from keystroke data or features demonstrating language usage patterns from text, this study focused on features that can assess cognitive abilities without using linguistic characteristics. Clinical trials were conducted in patients with mild cognitive impairment and early Alzheimer's dementia were conducted, and a series of experiments and validation tests were performed using the trial data. The results demonstrated that the proposed smartphone keystroke-based diagnostic method is effective in diagnosing cognitive impairment. The proposed method does not require the use of any special equipment except smartphones, which facilitates low-cost commercialization. This study presents a diagnostic method that addresses the problem of people who avoid tests for the diagnosis of dementia due to economic and psychological burdens.Clinical RelevanceThis study provides an approach for early detection of dementia using ordinary smartphone keystroke logs. The proposed method has the potential to improve the quality of life of patients with dementia. Following large-scale clinical research and the integration of more digital biomarkers, the methodology proposed in this study can potentially facilitate the development of an early diagnostic platform for dementia.