Abstract Biomedical research typically involves longitudinal study designs where samples from individuals are measured repeatedly over time and the goal is to identify risk factors (covariates) that associated with an outcome value. General linear mixed effect models standard workhorse for statistical analysis of data. However, data can be complicated reasons such as difficulties in modelling correlated values, functional (time-varying) covariates, nonlinear non-stationary effects, model...

Conditional variational autoencoders (CVAEs) are versatile deep latent variable models that extend the standard VAE framework by conditioning generative model with auxiliary covariates. The original CVAE assumes data samples independent, whereas more recent conditional models, such as Gaussian process (GP) prior VAEs, can account for complex correlation structures across all samples. While several methods have been proposed to learn VAEs from partially observed datasets, these fall short...

Conditional variational autoencoders (CVAEs) are versatile deep generative models that extend the standard VAE framework by conditioning model with auxiliary covariates. The original CVAE assumes data samples independent, whereas more recent conditional models, such as Gaussian process (GP) prior VAEs, can account for complex correlation structures across all samples. While several methods have been proposed to learn VAEs from partially observed datasets, these fall short VAEs. In this work,...

Abstract Motivation Biomedical research typically involves longitudinal study designs where samples from individuals are measured repeatedly over time and the goal is to identify risk factors (covariates) that associated with an outcome value. General linear mixed effect models have become standard workhorse for statistical analysis of data designs. However, can be complicated both practical theoretical reasons, including difficulties in modelling, correlated values, functional...

Abstract Numerous time-course gene expression datasets have been generated for studying the biological dynamics that drive disease progression; and nearly as many methods proposed to analyse them. However, barely any method exists can appropriately model data while accounting heterogeneity entails complex diseases. Most manage fulfil either one of those qualities, but not both. The lack appropriate hinders our capability understanding process pursuing preventive treatments. We present a...

Longitudinal datasets measured repeatedly over time from individual subjects, arise in many biomedical, psychological, social, and other studies. A common approach to analyse high-dimensional data that contains missing values is learn a low-dimensional representation using variational autoencoders (VAEs). However, standard VAEs assume the learnt representations are i.i.d., fail capture correlations between samples. We propose VAE (L-VAE), uses multi-output additive Gaussian process (GP)...

Abstract Numerous time-course gene expression datasets have been curated for studying the biological dynamics that drive disease progression; and nearly as many methods proposed to analyse them. However, barely any method exists can appropriately model data at same time account heterogeneity entails complex diseases. Most manage fulfil either one of those qualities, but not both. The lack appropriate hinders our capability understanding process pursuing preventive or curative treatments....

Clinical patient records are an example of high-dimensional data that is typically collected from disparate sources and comprises multiple likelihoods with noisy as well missing values. In this work, we propose unsupervised generative model can learn a low-dimensional representation among the observations in latent space, while making use all available heterogeneous setting We improve upon existing Gaussian process variable (GPLVM) by incorporating deep neural network parameterised...

The variational autoencoder (VAE) is a popular deep latent variable model used to analyse high-dimensional datasets by learning low-dimensional representation of the data. It simultaneously learns generative and an inference network perform approximate posterior inference. Recently proposed extensions VAEs that can handle temporal longitudinal data have applications in healthcare, behavioural modelling, predictive maintenance. However, these do not account for heterogeneous (i.e., comprising...

The variational autoencoder (VAE) is a popular deep latent variable model used to analyse high-dimensional datasets by learning low-dimensional representation of the data. It simultaneously learns generative and an inference network perform approximate posterior inference. Recently proposed extensions VAEs that can handle temporal longitudinal data have applications in healthcare, behavioural modelling, predictive maintenance. However, these do not account for heterogeneous data, i.e.,...