P4 Medicine for Heterogeneity of Dry Eye: A Mobile Health-based Digital Cohort Study

Abstract

During the 5^th Science, Technology, and Innovation Basic Plan, the Japanese government proposed a novel societal concept -Society 5.0- that promoted a healthcare system characterized by its capability to provide unintrusive, predictive, longitudinal care through the integration of cyber and physical space. The role of Society 5.0 in managing our quality of vision will become more important in the modern digitalized and aging society, both of which are known risk factors for developing dry eye. Dry eye is the most common ocular surface disease encountered in Japan with symptoms including increased dryness, eye discomfort, and decreased visual acuity. Owing to its complexity, implementation of P4 (predictive, preventive, personalized, participatory) medicine in managing dry eye requires a comprehensive understanding of its pathology, as well as a strategy to visualize and stratify its risk factors. Using DryEyeRhythm^®, a mobile health (mHealth) smartphone software (app), we established a route to collect holistic medical big data on dry eye, such as the subjective symptoms and lifestyle data for each individual. The studies to date aided in determining the risk factors for severe dry eye, the association between major depressive disorder and dry eye exacerbation, eye drop treatment adherence, app-based stratification algorithms based on symptomology, blink detection biosensoring as a dry eye-related digital phenotype, and effectiveness of app-based dry eye diagnosis support compared to traditional methods. These results contribute to elucidating disease pathophysiology and promoting preventive and effective measures to counteract dry eye through mHealth.

Keywords: P4 medicine; big data; dry eye; mobile health; smartphone application.

Figure 1

Description of user experience for DryEyeRhythm^Ⓡ The figure is used from Inomata T. et al. with permission.

Figure 2

Screenshots of DryEyeRhythm^Ⓡ (a)Welcome screen, (b)eConsent, (c) screen for entering participant characteristics, (d)Ocular Surface Disease Index questionnaire, (e) lifestyle information questionnaire, and (f)depressive symptoms questionnaire. The figure is used from Eguchi A. et al. with permission.

Figure 3

Characteristic visualization of symptomatic dry eye using collected comprehensive dry eye related health data and biosensoring data by DryEyeRhytm^Ⓡ (a)The heatmap of the correlation between each item of the Ocular Surface Disease Index (OSDI) and Self-rating Depression Scale (SDS) questionnaires. (b)A bubble chart of representative combinations of types of eye drops used by symptomatic dry eye individuals based on data from the subscales of the OSDI. Among the total 51 combinations, the top 20 eye drop combinations are shown. The X-axis represents the ocular symptoms score, the Y-axis represents the vision related-function score, and the Z-axis represents the environmental triggers score based on the OSDI questionnaire. The bubbles represent the proportion of the combinations of eye drops used. (c)Dimension reduction of individuals with symptomatic dry eye─via Uniform Manifold Approximation and Projection with spectral clustering identified by unsupervised clustering analysis (n=2,619 individuals collected by DryEyeRhythm^Ⓡ)─depicted seven clusters when stratified for subjective symptoms based on the 12 items of the Japanese version of OSDI. (d)Fraction of individuals within each cluster visualized on the left most panel, along with a corresponding heat map of subjective symptoms from individuals within the identified clusters. (e)Risk factors for each cluster in symptomatic dry eye compared with other clusters visualized in a circular layout. The figures are used from a; Inomata T et al., b; Eguchi A. et al., and c-e; Inomata T. et al., with permission.

Figure 4

Blink sensoring by DryEyeRhythm^Ⓡ (a)The duration of the participant's maximum blink interval (MBI) was recorded by DryEyeRhythm^Ⓡ. (b)MBI was significantly shortened in symptomatic dry eye vs. non-symptomatic dry eye (8.6 s vs. 11.0 s, ***P<0.001) (c)MBI each cluster (Kruskal-Wallis test, n=3,593, *P=0.016, ***P<0.001). DE, dry eye. The figure is used from Inomata T. et al. with permission.