Classify Criteria

When completing the overall classification framework, we manually refined and classified all the lifestyles included in PCaLiStDB. We qualitatively analyzed them to determine the attributes of the lifestyles based on statistical analysis and set the threshold as p-value <0.05. So far, the effects of lifestyles in PCaLiStDB on PCa were classified as protective, risk, no influencing, unclear and contradictory. In addition, according to the p-value and 95% CI, the levels of effects on PCa were scored as weak, moderate, strong and extremely strong correlation. The specific criteria are as follows:

Effect type	P value	Effect index	Impact levels	Impact degree
Protective factor	P<0.05	0.9≤a<1.0 0.7≤a<0.9 a<0.7	Weak Medium Strong	☆ ☆☆ ☆☆☆
No influencing factor	P<0.05	1.0≤a<1.2
Risk factor	P<0.05	1.2≤a<1.5 1.5≤a<3.0 3.0≤a<10.0 a≥10.0	Weak Medium Strong Super strong	☆ ☆☆ ☆☆☆ ☆☆☆☆
Unclear	P>0.05	No statistical significance factor
Contradictory factor	P<0.05	There are more than two conclusions in different literatures about a certain lifestyle

*a refers to the value of effect index, i.e the a in (a,b,c).
‘☆’ indicate the correlation extent of this lifestyle with the incidence of PCa,the more the number of ‘☆’, the stronger the effect of this lifestyle on the incidence of PCa.