Development of the phenome-wide biological age prediction model with application of artificial intelligence
Principal Investigator: Dr Eun Kyung Choe
Approved Research ID: 56892
Approval date: June 1st 2020
.In medical field, we use two concept of age, chronological age vs. biological age. Chronological age is the number of years a person has been alive, while biological age refers to how old a person functionally, medically, namely biologically. Biological age could be a powerful predictor for chronic disease (such as diabetes, hypertension), malignant disease, and mortality. In previous studies, biological age has been predicted using DNA methylation, telomere, laboratory test and exercise ability. But there are no golden standard to calculate biological age, nor there has not been a comprehensive analysis using various type of clinical features systematically. In our project, we would use all the clinical features available in UK biobank such as laboratory test results, ICD code and disease categories, environmental factor (including diet, exercise, smoking, job, residence), genetic factor (SNP array results), ethnical factor, image feature (DEXA, fundus images using CNN and brain image phenotypes from previous studies) to design a prediction model for biological age by artificial intelligence, including machine learning and deep learning. With the each phenotype association results, we would perform a network analysis to integrate the biological age predictors. The project duration is expected to be 3 years. By our analysis, we would be able to provide the most optimal model to predict biological age, as well as elucidate the interaction among the predictors by phenome wide network analysis. In the era of aging society, the mechanism of the aging would be the most important framework to enhance the healthy aging in healthcare. We believe that with our research, we could provide the comprehensive nature of biological age and illuminate new paradigm to approach healthcare in future medicine.