JMIR Publications

ABSTRACT

Background:

As the world’s population is aging, there are growing needs for prevention and prediction of dementia among the general population. The availability of national time-series health examination data in South Korea invites an opportunity to utilise deep learning, an artificial intelligence technology, to expedite the analysis of mass and sequential data.

Objective:

We aimed to compare the discriminative accuracy between a time-series deep learning algorithm and conventional statistical methods for predicting all-cause dementia and Alzheimer’s dementia using periodic health examination data.

Methods:

We used a South Korean national health examination cohort database to identify dementia and Alzheimer’s dementia based on diagnostic codes in medical claims data over a 10-year period. Our analyses of all-cause dementia and Alzheimer’s dementia included 479,845 and 465,081 individuals, respectively, who were 40–79 years of age and without dementia at baseline. We compared the predictive performance of three models: Cox proportional hazards model using only baseline data (HR-B), Cox proportional hazards model using repeated measurements (HR-R), and deep learning model using repeated measurements (DL-R).

Results:

The discrimination indices (95% confidence interval) for HR-B, HR-R, and DL-R models predicting all-cause dementia were 0.84 (0.83–0.85), 0.87 (0.86–0.88), and 0.90 (0.90–0.90), respectively. The discrimination indices for HR-B, HR-R, and DL-R models predicting Alzheimer’s dementia were 0.87 (0.86–0.88), 0.90 (0.88–0.91), and 0.91 (0.91–0.91), respectively. Thus, the DL-R model showed the best performance, followed by the HR-R model, in predicting both all-cause dementia and Alzheimer’s dementia. The DL-R model was superior to the HR-R model in all validation groups tested.

Conclusions:

A deep learning algorithm using time-series data could be an accurate, cost-effective method of predicting dementia. A combination of deep learning and proportional hazards models could help enhance prevention strategies for dementia.