about hazard value!

[alexxiecan]

hi, everyone:

We hope to calculate the output value of the model, namely the hazard risk. Can we use "def predict_ hazard(self, input, batchsize=8224, numpy=None, eval=True, to_cpu=False,num_workers=0)"？ If possible, is the value it returns about the risk of each patient or what other risks?

3ks

[puckybreg]

Also interested in calculating the hazard beta parameters for the fit model. @alexxiecan I have yet to find anything in the examples or in the code itself that points to the best way to get the covariate effects on the Survival Curve.

For that matter, it seems that PyCox is mostly geared towards prediction and not explanation of the Survival Curve. Am I right in understanding that after fitting the model, there is no clear way to getting at the computed Kaplan-Meir curve beyond running model.predict on the test dataset?

[alexxiecan]

Firstly, I agree with you that the current deep survival model can only be used for prediction, and cannot calculate the hazard ratio or specific h (x) function values of certain variables like R language. But I guess if the data is sent to a trained model, its output should be the value of h (x)? I don't know if my understanding is correct? However, in terms of programming, I have not yet implemented....I hope we can continue to improve pycox and implement certain specific computing functions. 

谢灿 @.***

 

------------------ 原始邮件 ------------------ 发件人: "Jonah @.>; 发送时间: 2023年5月21日(星期天) 凌晨0:50 收件人: @.>; 抄送: @.>; @.>; 主题: Re: [havakv/pycox] about hazard value! (Issue #163)

Also interested in calculating the hazard beta parameters for the fit model. @alexxiecan I have yet to find anything in the examples or in the code itself that points to the best way to get the covariate effects on the Survival Curve.

For that matter, it seems that PyCox is mostly geared towards prediction and not explanation of the Survival Curve. Am I right in understanding that after fitting the model, there is no clear way to getting at the computed Kaplan-Meir curve beyond running model.predict on the test dataset?

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you were mentioned.Message ID: @.***>