Opportunities and obstacles for deep learning in biology and medicine

Abstract

Deep learning describes a class of machine learning algorithms that are capable of combining raw inputs into layers of intermediate features. These algorithms have recently shown impressive results across a variety of domains. Biology and medicine are data-rich disciplines, but the data are complex and often ill-understood. Hence, deep learning techniques may be particularly well suited to solve problems of these fields. We examine applications of deep learning to a variety of biomedical problems—patient classification, fundamental biological processes and treatment of patients—and discuss whether deep learning will be able to transform these tasks or if the biomedical sphere poses unique challenges. Following from an extensive literature review, we find that deep learning has yet to revolutionize biomedicine or definitively resolve any of the most pressing challenges in the field, but promising advances have been made on the prior state of the art. Even though improvements over previous baselines have been modest in general, the recent progress indicates that deep learning methods will provide valuable means for speeding up or aiding human investigation. Though progress has been made linking a specific neural network's prediction to input features, understanding how users should interpret these models to make testable hypotheses about the system under study remains an open challenge. Furthermore, the limited amount of labelled data for training presents problems in some domains, as do legal and privacy constraints on work with sensitive health records. Nonetheless, we foresee deep learning enabling changes at both bench and bedside with the potential to transform several areas of biology and medicine.

Precision Oncology: The Road Ahead

“…In all cases, a key challenge is the selection of features from each platform as inputs to the clustering algorithms; for example, it is possible to summarize mutations, gene expression, and DNA methylation events as binary alterations [80], and then treat any missing data as a non-alteration event. We anticipate that recent advances in methods for learning low-dimensional representations of multiple data types such as deep neural nets [83] will soon be applied in molecular classification of tumors, given the amount of molecular cancer data being produced and the successful application of deep neural nets in areas of computer vision, natural language processing, and biology [84]. Initial molecular subtype studies have often focused on clustering samples into subtypes based on gene expression in a single cancer type, which have provided robust biomarkers and subgroups, coherent with patient survival profiles (e.g, in breast cancer [85] or colorectal cancer (CRC) [86]).…”

Extracting chemical–protein relations with ensembles of SVM and deep learning models

“…Deep neural networks have recently achieved promising results in the biomedical relation extraction task (4). When compared with traditional machine-learning methods, they may be able to overcome the feature sparsity and engineering problems.…”

L’intelligence artificielle à l’aide des myologues

“…Pour qu'une « intelligence artificielle », utilisant des algorithmes de deep learning, puisse analyser et interpréter des données associées aux pathologies neuromusculaires, il est fondamental de lui proposer un apprentissage sur des informations médicales pertinentes [31]. Pour cela on peut aborder les choses suivant deux approches diffé-rentes : la première est basée sur la quantité d'informations présentes dans les bases de données (« data-driven » [19]), la seconde sur la qualité de ces données (« goal-driven » [32] …”