Deep Neural Networks for Semiparametric Frailty Models via H-likelihood
This work addresses prediction challenges in clustered survival analysis for fields like healthcare or social sciences, representing an incremental improvement by integrating deep learning with frailty models.
The authors tackled the problem of predicting clustered time-to-event data by proposing a deep neural network based gamma frailty model (DNN-FM), which uses h-likelihood for training and improves prediction performance over existing methods, as shown in experimental studies and real data analysis.
For prediction of clustered time-to-event data, we propose a new deep neural network based gamma frailty model (DNN-FM). An advantage of the proposed model is that the joint maximization of the new h-likelihood provides maximum likelihood estimators for fixed parameters and best unbiased predictors for random frailties. Thus, the proposed DNN-FM is trained by using a negative profiled h-likelihood as a loss function, constructed by profiling out the non-parametric baseline hazard. Experimental studies show that the proposed method enhances the prediction performance of the existing methods. A real data analysis shows that the inclusion of subject-specific frailties helps to improve prediction of the DNN based Cox model (DNN-Cox).