Table 4.

The advantages and disadvantages of the four methods and possible effective improvements.

Methods	Advantages	Disadvantages	Improvements
MLR	Simple and easy to operate	(1) Biomarkers paradox (2) co-linearity (3) Regression equation edge distortion	(1) Z-score correction edge distortion (14) (2) Co-linearity diagnosis and removal of redundant variables
PCA	(1) Avoid co-linearity (54) (2) Further screen aging biomarkers (3)	(1) Biomarkers paradox (2) Regression equation edge distortion	Z-score correction edge distortion (14)
KDM	(1) Resolving the biomarker paradox (2) Avoiding distortion at the edges of the regression equation (76) (3) Suitable for non-linear biomarkers (17)	(1) Complicated calculation (54) (2) CA as a marker of aging is controversial (34)	(1) Cho et al. improved the calculation process (54) (2) Calculating individual △age is more practical than BA (34)
Deep learning	(1) Good at handling high-dimensional dataset (67) (2) The machine extracts features autonomously by learning (67)	(1) Difficulty in building large data (8) (2) The existence of a “black box” and uncontrollable results (3) Excellent programming skills and computer hardware and software support required	(1) Suitable methods can be explored to clarify the weighting of each aging biomarker (2) Multidisciplinary Cooperation

MLR, multiple linear regression; PCA, principal component analysis; KDM, Klemera and Doubal’s method.