Kim TH, et al. Psychiatry Investig. 2010 Jun;7(2):102-108. http://dx.doi.org/10.4306/pi.2010.7.2.102

The classification of Alzheimer’s disease (AD) using deep learning methods has shown promising results, but successful application in clinical settings requires a combination of high accuracy, short processing time, and generalizability to various populations. In this study, we developed a convolutional neural network (CNN)-based AD classification algorithm using magnetic resonance imaging (MRI) scans from AD patients and age/gender-matched cognitively normal controls from two populations that differ in ethnicity and education level. These populations come from the Seoul National University Bundang Hospital (SNUBH) and Alzheimer’s Disease Neuroimaging Initiative (ADNI). For each population, we trained CNNs on five subsets using coronal slices of T1-weighted images that cover the medial temporal lobe. We evaluated the models on validation subsets from both the same population (within-dataset validation) and other population (between-dataset validation). Our models achieved average areas under the curves of 0.91–0.94 for within-dataset validation and 0.88–0.89 for between-dataset validation. The mean processing time per person was 23–24 s. The within-dataset and between-dataset performances were comparable between the ADNI-derived and SNUBH-derived models. These results demonstrate the generalizability of our models to different patients with different ethnicities and education levels, as well as their potential for deployment as fast and accurate diagnostic support tools for AD.

Dry eye disease (DED) is one of the most common ocular diseases. Ocular symptoms of DED vary from mild discomfort to severe ocular fatigue and pain that may affect daily activities.^1-5 Several population-based studies^2-4,6-15 have reported the prevalence of DED to range from 4.3% to 73.5%. The large disparity in the reported prevalence is attributed to the lack of a standardized diagnostic definition and differences in study populations.⁴ Although there is much inconsistency in the prevalence of DED in the literature, risk factors, such as old age, female sex, and hazardous environment, including air pollution and indoor air problems at work, are assumed to be associated with the prevalence of DED.⁴ The results of previous studies^16-19 have also suggested ethnic differences in the prevalence of DED between white and Asian populations, although a direct comparison of such studies involves certain limitations, such as discrepancy in the definition of DED and a difference in the age group.

With no effective treatments for cognitive decline or dementia, improving the evidence base for modifiable risk factors is a research priority. This study investigated associations between risk factors and late-life cognitive decline on a global scale, including comparisons between ethno-regional groups. We harmonized longitudinal data from 20 population-based cohorts from 15 countries over 5 continents, including 48,522 individuals (58.4% women) aged 54–105 (mean = 72.7) years and without dementia at baseline. Studies had 2–15 years of follow-up. The risk factors investigated were age, sex, education, alcohol consumption, anxiety, apolipoprotein E ε4 allele (APOE*4) status, atrial fibrillation, blood pressure and pulse pressure, body mass index, cardiovascular disease, depression, diabetes, self-rated health, high cholesterol, hypertension, peripheral vascular disease, physical activity, smoking, and history of stroke. Associations with risk factors were determined for a global cognitive composite outcome (memory, language, processing speed, and executive functioning tests) and Mini-Mental State Examination score. Individual participant data meta-analyses of multivariable linear mixed model results pooled across cohorts revealed that for at least 1 cognitive outcome, age (B = −0.1, SE = 0.01), APOE*4 carriage (B = −0.31, SE = 0.11), depression (B = −0.11, SE = 0.06), diabetes (B = −0.23, SE = 0.10), current smoking (B = −0.20, SE = 0.08), and history of stroke (B = −0.22, SE = 0.09) were independently associated with poorer cognitive performance (p < 0.05 for all), and higher levels of education (B = 0.12, SE = 0.02) and vigorous physical activity (B = 0.17, SE = 0.06) were associated with better performance (p < 0.01 for both). Age (B = −0.07, SE = 0.01), APOE*4 carriage (B = −0.41, SE = 0.18), and diabetes (B = −0.18, SE = 0.10) were independently associated with faster cognitive decline (p < 0.05 for all). Different effects between Asian people and white people included stronger associations for Asian people between ever smoking and poorer cognition (group by risk factor interaction: B = −0.24, SE = 0.12), and between diabetes and cognitive decline (B = −0.66, SE = 0.27; p < 0.05 for both). Limitations of our study include a loss or distortion of risk factor data with harmonization, and not investigating factors at midlife. These results suggest that education, smoking, physical activity, diabetes, and stroke are all modifiable factors associated with cognitive decline. If these factors are determined to be causal, controlling them could minimize worldwide levels of cognitive decline. However, any global prevention strategy may need to consider ethno-regional differences.

Due to an unprecedented rate of population aging, South Korea is facing a dementia epidemic. For this reason, the Korean Longitudinal Study on Cognitive Aging and Dementia (KLOSCAD) was launched in 2009 with support from the Korean Health Industry Development Institute to investigate the epidemiology, biopsychosocial risk factors, and outcomes of dementia and dementia-related conditions. The KLOSCAD is the first nationwide multi-center population-based prospective cohort study. In October 2010, 12,694 individuals were randomly sampled from residents aged 60 years or older who lived in 13 districts across South Korea. In the baseline assessment, which was conducted from November 2010 through October 2012, 6,818 (53.7%) individuals participated. Follow-up assessments have been conducted every two years, with the first follow-up assessment conducted between November 2012 and October 2014, and the second between November 2014 and October 2016. The third is now in progress, and will span from November 2016 to October 2018. Diagnosis of cognitive disorders, neuropsychological battery, behavioral and psychological symptoms of dementia, activities of daily living, physical and neurologic examination and laboratory tests, life styles, quality of life, and identification of death were evaluated in each assessment. The cumulative drop-out rate at the second follow-up assessment was 38.7%. Dementia and mild cognitive impairment were 5.0% and 27.0%, respectively. The KLOSCAD may provide strong scientific evidence for advancing the fight against dementia both in Korea and globally.