Table 2. Rank of risk factors in deep learning model.

Feature name	Sum of ranks	Feature name	Mean of values
Age	233,322	Age	0.405
Systolic blood pressure	359,881	Systolic blood pressure	0.262
Sex	390,006	Diastolic blood pressure	0.153
Diastolic blood pressure	548,049	Sex	0.116
Fasting plasma glucose	584,936	Fasting plasma glucose	0.111
Gamma-glutamyl transpeptidase	664,941	Current smoking	0.111
Aspartate transaminase	668,470	Exercise	0.105
Hemoglobin	683,408	Aspartate transaminase	0.074
Total cholesterol	757,839	Gamma-glutamyl transpeptidase	0.066
Exercise	776,784	Hemoglobin	0.061
Alcohol intake	814,943	Alcohol intake	0.052
Body mass index	837,221	Total cholesterol	0.045
Urine protein	867,499	Body mass index	0.032
Alanine transaminase	973,370	Urine protein	0.028
Family history of etc (include cancer)	1,150,578	History of Hypertension	0.026
Family history of Stroke	1,187,298	Family history of etc (include cancer)	0.025
Family history of Diabetes mellitus	1,299,493	History of etc (include cancer)	0.022
Family history of Heart disease	1,392,376	Alanine transaminase	0.015
Family history of Hypertension	1,412,000	History of Diabetes mellitus	0.012
Current smoking	1,486,150	Family history of Hypertension	0.004
History of Hypertension	1,546,467	Family history of Diabetes mellitus	0.002
History of Diabetes mellitus	1,567,078	Family history of Stroke	0.001
History of etc (include cancer)	1,585,386	Family history of Heart disease	0.000

Sum of ranks indicate ranking each sample by absolute value of LRP, then ascending order by summing the ranks by variables in all samples. Mean of values indicate calculate the mean for the absolute value of LRP by variable in all samples and sort in descending order.