A Review on Bayesian Deep Learning in Healthcare: Applications and Challenges

Abstract

In the last decade, Deep Learning (DL) has revolutionized the use of artificial intelligence, and it has been deployed in different fields of healthcare applications such as image processing, natural language processing, and signal processing. DL models have also been intensely used in different tasks of healthcare such as disease diagnostics and treatments. Deep learning techniques have surpassed other machine learning algorithms and proved to be the ultimate tools for many state-of-the-art applications. Despite all that success, classical deep learning has limitations and their models tend to be very confident about their predicted decisions because it does not know when it makes mistake. For the healthcare field, this limitation can have a negative impact on models predictions since almost all decisions regarding patients and diseases are sensitive. Therefore, Bayesian deep learning (BDL) has been developed to overcome these limitations. Unlike classical DL, BDL uses probability distributions for the model parameters, which makes it possible to estimate the whole uncertainties associated with the predicted outputs. In this regard, BDL offers a rigorous framework to quantify all sources of uncertainties in the model. This study reviews popular techniques of using Bayesian deep learning with their benefits and limitations. It also reviewed recent deep learning architecture such as Convolutional Neural Networks and Recurrent Neural Networks. In particular, the applications of Bayesian deep learning in healthcare have been discussed such as its use in medical imaging tasks, clinical signal processing, medical natural language processing, and electronic health records. Furthermore, this paper has covered the deployment of Bayesian deep learning for some of the widespread diseases. This paper has also discussed the fundamental research challenges and highlighted some research gaps in both the Bayesian deep learning and healthcare perspective.

Exploring uncertainty measures in convolutional neural network for semantic segmentation of oral cancer images

“…BDL models provide a framework for estimating uncertainty by modeling the posterior distribution. 20 – 22 …”

“…BDL models provide a framework for estimating uncertainty by modeling the posterior distribution. [20][21][22] Bayesian networks are probabilistic models, not deterministic ones, that learn a distribution over their weights. Given training data X and Y, they aim to learn the posterior distribution of the neural network's weights W. The posterior distribution is often approximated using variational inference methods, such as Dropout variational inference.…”

Bayesian deep learning methods applied to diabetic retinopathy disease: a review

“…BDL denotes to probabilistic DL, which primarily relies on the Bayes theorem. In Bayesian techniques, the likelihood of the data and prior "expert" knowledge are used to create posterior distributions, which can indicate various levels of modal uncertainty [42]. It is noteworthy that there is another phrase used in the literature is Bayesian neural networks (BNN).…”

“…Figure 4 illustrates the difference between classical and Bayesian NNs. [42] There are two types of uncertainty that BDL can handle in general: Aleatoric uncertainty and epistemic uncertainty [44] see Figure 5. Aleatoric uncertainty is the type of complicated uncertainty in data "known as data uncertainty" that causes uncertainty in predictions [45].…”

“…On the other hand, Epistemic uncertainty (as well-known as knowledge uncertainty) results from insufficient knowledge in the model framework. Epistemic uncertainty is manageable, and the Bayesian approach is an effective means of handling this kind of uncertainty without overfitting [42]. Seeking to quantify epistemic uncertainty, one can select different models to respond various questions in model-based prediction.…”