How to build a Risk score formula using gene expression from cox regression for a survival analysis

Question

[Alex]

Dear everyone

I'm confused that how to build Risk score formula for survival analysis. I have been seen many articles involved the Risk score formula building, but I have spent a day to solved it but failed. can someone taken a example to show how to build it?

This is the article details:

Base on the expression level of five lncRNAs, we designed a risk-score formula for ccRCC patients’ survival prediction. The risk score formula is as following:

Risk score= (1.43 × expression level of AC069513.4) +
(0.81 × expression level of AC003092.1) +
(1.64 × expression level of RP11-507K2.3) +
(-6.56 × expression level of CTC-205M6.2) +
(-1.72 × expression level of U91328.21)

but I don't know how to calculate the Risk score formular

Thank you very much!

Alex

[zx8754]

Didn't read the article, but it looks like a simple formula, multiply each expression with some weight then sum all.

Answer 1

[Kevin Blighe]

As to which zx8754 alluded, it is simply about multiplying the expression values by a weight, and then summing these up.

The weights that they use are the beta estimates (coefficients) from the fitted Cox proportional hazards model. The exponential function applied to these coefficients gives the odds ratios.

This, and variations of it, represent a very common way to calculate 'risk scores'.

Hopefully that helps.

Kevin