Objective:To explore the potential of brain age gap (BAG) as a biomarker of brain health and analyze its causal relationship with common brain diseases.Methods:Brain structural magnetic resonance imaging (sMRI) data from public databases (UK Biobank, ADNI, PPMI) were selected and input into a simple fully convolutional network (SFCN) to estimate BAG. The disease group (with corresponding codes or labels, n=6 796) and healthy control group (without corresponding codes or labels, n=9 660) were defined according to the presence or absence of ICD-10 codes and corresponding brain disease labels. The two-sample t-test was used to compare the BAG differences between the disease and healthy control group; genome-wide association study (GWAS) was used to find genomic regions significantly associated with BAG in 31 520 people in the UK Biobank. The causal effects between BAG and 14 brain diseases were analyzed by Mendelian randomization (MR). Results:The mean absolute error (MAE) between the subject′s chronological age and estimated brain age for the 1 932 subjects in the healthy control group used for model testing was 2.364 years. Compared with the healthy control group, Alzheimer′s disease ( t=33.42), anxiety disorders ( t=2.38), bipolar disorder ( t=3.76), stroke ( t=2.75), demyelinating disease ( t=7.45), major depressive disorder ( t=3.49), Parkinson′s disease ( t=17.69), and post-traumatic stress disorder ( t=2.34) BAG was significantly increased ( PFDR<0.05). There were 8 independent genome-wide risk regions associated with BAG in the GWAS ( P<5×10 -8), 4 of which were novel(related genes: PICK1, TBC1D9, SIAH3, and TMEM98). In MR analysis, a strong causal association between Alzheimer′s disease and BAG was observed (β=0.23,95% CI=0.08-0.38, PFDR=0.030). Conclusion:BAG can be used as a biomarker that reflects brain health information. The occurrence of Alzheimer′s disease will lead to an increase in BAG.

Objective:To explore the potential of brain age gap (BAG) as a biomarker of brain health and analyze its causal relationship with common brain diseases.Methods:Brain structural magnetic resonance imaging (sMRI) data from public databases (UK Biobank, ADNI, PPMI) were selected and input into a simple fully convolutional network (SFCN) to estimate BAG. The disease group (with corresponding codes or labels, n=6 796) and healthy control group (without corresponding codes or labels, n=9 660) were defined according to the presence or absence of ICD-10 codes and corresponding brain disease labels. The two-sample t-test was used to compare the BAG differences between the disease and healthy control group; genome-wide association study (GWAS) was used to find genomic regions significantly associated with BAG in 31 520 people in the UK Biobank. The causal effects between BAG and 14 brain diseases were analyzed by Mendelian randomization (MR). Results:The mean absolute error (MAE) between the subject′s chronological age and estimated brain age for the 1 932 subjects in the healthy control group used for model testing was 2.364 years. Compared with the healthy control group, Alzheimer′s disease ( t=33.42), anxiety disorders ( t=2.38), bipolar disorder ( t=3.76), stroke ( t=2.75), demyelinating disease ( t=7.45), major depressive disorder ( t=3.49), Parkinson′s disease ( t=17.69), and post-traumatic stress disorder ( t=2.34) BAG was significantly increased ( PFDR<0.05). There were 8 independent genome-wide risk regions associated with BAG in the GWAS ( P<5×10 -8), 4 of which were novel(related genes: PICK1, TBC1D9, SIAH3, and TMEM98). In MR analysis, a strong causal association between Alzheimer′s disease and BAG was observed (β=0.23,95% CI=0.08-0.38, PFDR=0.030). Conclusion:BAG can be used as a biomarker that reflects brain health information. The occurrence of Alzheimer′s disease will lead to an increase in BAG.

Cohort study is one of the important research methods in analytical epidemiology because of its clear time sequence relationship, which is better than other observational studies in demonstrating causal association. However, screening diagnosis or other methods are often used to exclude the individuals with outcome events during the enrollment process of the subjects in cohort studies. The accuracy of screening diagnosis and the effectiveness of exclusion will affect the accuracy of the baseline status assessment of the subjects included in the study, which may lead to the causal time sequence reversal of exposure-outcome in the estimation of causal effect. Landmark analysis can be used to control reverse causality by excluding subjects with potentially unknown expose-outcome timing. In this paper, we describe the basic principles and analytical steps of landmark analysis, and use data from the Chinese Longitudinal Healthy Longevity Survey to explore the relationship between physical activity and frailty, and introduce the specific application of landmark analysis for the purpose of facilitating its application and inferring causal effects more accurately in cohort studies.

With completing a baseline survey of a large natural population cohort, conducting regular follow-up has become a key factor in further improving the quality of cohort construction and ensuring its sustainable development. Typical cohort follow-up methods include repeat surveys, routine monitoring, and community-oriented surveillance. However, in practical applications, there are often issues such as high costs, difficulty, and high error rates. Telephone follow-up is an important supplementary method to the methods mentioned above, as it has the characteristics of low cost, fast response, and high quality. However, the with difficult organization, quality control is challenging, response rates are low, and management levels vary widely, which limits its widespread use in large-scale population cohort studies. Given the above problems, this study draws on customer relationship management based on the actual needs of the China Northwest Cohort follow-up. It relies on the REDCap electronic data collection platform to build a telephone follow-up management and quality control system. Targeted solutions are provided for key issues in telephone follow-up implementation, including organizational structure, project management, data collection, and process quality control, to improve the quality control level of telephone follow-up comprehensively and thereby enhance the quality and efficiency of follow-up. We hope to provide standardized follow-up programs and efficient quality control tools for newly established and existing cohort studies.