Opportunities and obstacles for deep learning in biology and medicine
By Travers Ching, Daniel S. Himmelstein, Brett K. Beaulieu-Jones, Alexandr A Kalinin, Brian T Do, G. Way, E. Ferrero, P. Agapow, M. Zietz, Michael M. Hoffman and 26 more. 2017
2,030 citations recorded by the source corpus.
Deep learning, which describes a class of machine learning algorithms, has recently showed impressive results across a variety of domains. Biology and medicine are data rich, but the data are complex and often ill-understood. Problems of this nature may be particularly well-suited to deep learning techniques. We examine applications of deep learning to a variety of biomedical problems -- patient classification, fundamental biological processes, and treatment of patients -- to predict whether deep learning will transform these tasks or if the biomedical sphere poses unique challenges. We find that deep learning has yet to revolutionize or definitively resolve any of these problems, but promising advances have been made on the prior state of the art. Even when improvement over a previous baseline has been modest, we have seen signs that deep learning methods may speed or aid human investigation. More work is needed to address concerns related to interpretability and how to best model each problem. Furthermore, the limited amount of labeled data for training presents problems in some domains, as can legal and privacy constraints on work with sensitive health records. Nonetheless, we foresee deep learning powering changes at the bench and bedside with the potential to transform several areas of biology and medicine.
Subfields: Artificial Intelligence, Biophysics, Molecular Biology, Computational Biology & Bioinformatics, Medical & Health AI
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Artificial Intelligence
- Yoshua Bengio
- Quoc V. Le
- Percy Liang
- Sergey Levine
- Mohit Bansal
- Jianfeng Gao
- Ruslan Salakhutdinov
- Richard Socher
In Biophysics
- Jure Leskovec
- Alán Aspuru-Guzik
- Joel T. Dudley
- Brett K. Beaulieu-Jones
- Jascha Narain Sohl-Dickstein
- Hanspeter Pfister
- A. Gholipour
- Alexander V. Buslaev
In Molecular Biology
- Joe G. Greener
- Shaun M. Kandathil
- David T. Jones
- Marinka Zitnik
- Jure Leskovec
- Junzhou Huang
- Sameer Quazi
- Weihua Hu
In Computational Biology & Bioinformatics
- Joe G. Greener
- David T. Jones
- Shaun M. Kandathil
- Lewis Moffat
- Jure Leskovec
- Junzhou Huang
- Brett K. Beaulieu-Jones
- Marinka Zitnik
In Medical & Health AI
- Marinka Zitnik
- Chao Yu
- Jure Leskovec
- Jiming Liu
- Brett K. Beaulieu-Jones
- Joel T. Dudley
- Jascha Narain Sohl-Dickstein
- Lewis Smith
Citations in the corpus
- Privacy-Preserving Generative Deep Neural Networks Support Clinical Data Sharing (2017, methodology): “GANs have become a class of widely used machine learning methods and have recently been used in biology and medicine [9] and have been used to generate biomedical data [10,11].”
- Strategies for Pre-training Graph Neural Networks (2019, background): “Several key studies (Xu et al., 2017; Ching et al., 2018; Wang et al., 2019) matter of increasing the number of labeled pre-training datasets that are from the same domain as the downstream task.”
- Split learning for health: Distributed deep learning without sharing raw patient data (2018, background): “Deep learning methods in general have found a pervasive suite of applications in biology, clinical medicine, genomics and public health as surveyed in [24, 23, 25, 29, 30, 31].”
- Albumentations: fast and flexible image augmentations (2018, background): “For example, very extensive image augmentations are typical in medical image analysis, where datasets are often small and expensive to acquire, and annotations are often sparse, compared to the natural images [7], [8].”
- An overview of deep learning in medical imaging focusing on MRI (2018, background): “For example the comprehensive review [101],24 covering both medicine and biology and spanning from imaging applications in healthcare to protein-protein interaction and uncertainty quantification; key concepts of deep learning for clinical radiologists [102–111], including radiomics and imaging geno”
- Learning to predict the cosmological structure formation (2018, background): “This new paradigm is also significantly impacting a variety of domains in the sciences, from biology (27, 28) to chemistry (29, 30) and physics (31, 32).”
- Reinforcement Learning in Healthcare: A Survey (2019, methodology): “methods and machine learning approaches in a variety of healthcare applications [5], [6], [7], [8], [9].”
- ProteinNet: a standardized data set for machine learning of protein structure (2019, background): “Background Deep learning has revolutionized many areas of computer science including computer vision, natural language processing, and speech recognition [1], and is now being widely applied to bioinformatic problems ranging from clinical image classification [2] to prediction of protein-DNA binding”
- WILDS: A Benchmark of in-the-Wild Distribution Shifts (2020, neutral): “These datasets have powered ML models that aim to learn predictive representations of functional DNA and predict genome-wide biochemical profiles in contexts for which experimental data is unavailable (Ching et al., 2018; Eraslan et al., 2019; Libbrecht and Noble, 2015).”
- Deep learning with noisy labels: exploring techniques and remedies in medical image analysis (2019, background): “impact is only expected to grow [1], [2].”
- Identifying Ethical Considerations for Machine Learning Healthcare Applications (2020, background): “…can be higher than that of individual experts in interpretation of clinical images such as radiologic scans, pathology slides, and photographs of skin lesions (Ching et al. 2018), the estimated accuracy of a ML-HCA is dependent on the clinical context in which the application is being assessed.”
- Multimodal deep learning for biomedical data fusion: a review (2022, background): “Although DL architectures for biomedical applications have been reviewed [16], the different DL-based fusion strategies for heterogeneous data have not.”
- Artificial intelligence in cancer target identification and drug discovery (2022, background): “Although the recent advances in biotechnologies have enabled the fast generation of massive biomedical data, such data often suffer from inconsistency in production and information missing in annotation, resulting in the lack of reliable and consistent data for validating deep learning models 264 .”