Opportunities and obstacles for deep learning in biology and medicine

被引：1176

作者：

Ching, Travers ^{[1
]}

Himmelstein, Daniel S. ^{[2
]}

Beaulieu-Jones, Brett K. ^{[3
]}

Kalinin, Alexandr A. ^{[4
]}

Do, Brian T. ^{[5
]}

Way, Gregory P. ^{[2
]}

Ferrero, Enrico ^{[6
]}

Agapow, Paul-Michael ^{[7
]}

Zietz, Michael

Hoffman, Michael M. ^{[8
,9
,10
]}

Xie, Wei ^{[11
]}

Rosen, Gail L. ^{[12
]}

Lengerich, Benjamin J. ^{[13
]}

Israeli, Johnny ^{[14
]}

Lanchantin, Jack ^{[17
]}

Woloszynek, Stephen

Carpenter, Anne E. ^{[18
,19
]}

Shrikumar, Avanti ^{[15
]}

Xu, Jinbo ^{[20
]}

Cofer, Evan M. ^{[21
,22
]}

Lavender, Christopher A. ^{[23
]}

Turaga, Srinivas C. ^{[24
]}

Alexandari, Amr M.

Lu, Zhiyong ^{[25
,26
]}

Harris, David J. ^{[27
]}

DeCaprio, Dave ^{[28
]}

Qi, Yanjun

Kundaje, Anshul ^{[15
,16
]}

Peng, Yifan ^{[25
,26
]}

Wiley, Laura K. ^{[29
]}

Segler, Marwin H. S. ^{[30
]}

Boca, Simina M. ^{[31
]}

Swamidass, S. Joshua ^{[32
]}

Huang, Austin ^{[33
]}

Gitter, Anthony ^{[34
,35
]}

Greene, Casey S. ^{[2
]}

机构：

[1] Univ Hawaii Manoa, Mol Biosci & Bioengn Grad Prog, Honolulu, HI USA

[2] Univ Penn, Perelman Sch Med, Dept Syst Pharmacol & Translat Therapeut, Philadelphia, PA 19104 USA

[3] Univ Penn, Perelman Sch Med, Genom & Computat Biol Grad Grp, Philadelphia, PA 19104 USA

[4] Univ Michigan, Sch Med, Dept Computat Med & Bioinformat, Ann Arbor, MI USA

[5] Harvard Med Sch, Boston, MA USA

[6] GlaxoSmithKline, Computat Biol & Stats, Target Sci, Stevenage, Herts, England

[7] Imperial Coll London, Data Sci Inst, London, England

[8] Princess Margaret Canc Ctr, Toronto, ON, Canada

[9] Univ Toronto, Dept Med Biophys, Toronto, ON, Canada

[10] Univ Toronto, Dept Comp Sci, Toronto, ON, Canada

[11] Vanderbilt Univ, Elect Engn & Comp Sci, 221 Kirkland Hall, Nashville, TN 37235 USA

[12] Drexel Univ, Dept Elect & Comp Engn, Ecol & Evolutionary Signal Proc & Informat Lab, Philadelphia, PA 19104 USA

[13] Carnegie Mellon Univ, Sch Comp Sci, Computat Biol Dept, Pittsburgh, PA USA

[14] Stanford Univ, Biophys Program, Stanford, CA 94305 USA

[15] Stanford Univ, Dept Comp Sci, Stanford, CA 94305 USA

[16] Stanford Univ, Dept Genet, Stanford, CA 94305 USA

[17] Univ Virginia, Dept Comp Sci, Charlottesville, VA 22903 USA

[18] Broad Inst Harvard, Imaging Platform, Cambridge, MA USA

[19] MIT, 77 Massachusetts Ave, Cambridge, MA 02139 USA

[20] Toyota Technol Inst, Chicago, IL USA

[21] Trinity Univ, Dept Comp Sci, San Antonio, TX USA

[22] Princeton Univ, Lewis Sigler Inst Integrat Genom, Princeton, NJ 08544 USA

[23] NIEHS, NIH, Integrat Bioinformat, POB 12233, Res Triangle Pk, NC 27709 USA

[24] Howard Hughes Med Inst, Janelia Res Campus, Ashbum, VA USA

[25] NIH, Natl Ctr Biotechnol Informat, Bethesda, MD USA

[26] NIH, Natl Lib Med, Bethesda, MD USA

[27] Univ Florida, Dept Wildlife Ecol & Conservat, Gainesville, FL USA

[28] ClosedLoop Ai, Austin, TX USA

[29] Univ Colorado, Sch Med, Div Biomed Informat & Personalized Med, Aurora, CO USA

[30] Westfal Wilhelms Univ Munster, Inst Organ Chem, Munster, Germany

[31] Georgetown Univ, Med Ctr, Innovat Ctr Biomed Informat, Washington, DC 20007 USA

[32] Washington Univ, Dept Pathol & Immunol, St Louis, MO USA

[33] Brown Univ, Dept Med, Providence, RI 02912 USA

[34] Univ Wisconsin, Dept Biostat & Med Informat, Madison, WI 53706 USA

[35] Morgridge Inst Res, Madison, WI 53715 USA

来源：

JOURNAL OF THE ROYAL SOCIETY INTERFACE | 2018年 / 15卷 / 141期

基金：

美国国家科学基金会; 美国国家卫生研究院; 加拿大自然科学与工程研究理事会;

关键词：

deep learning; genomics; precision medicine; machine learning; CONVOLUTIONAL NEURAL-NETWORKS; PROTEIN-PROTEIN INTERACTION; ELECTRONIC HEALTH RECORDS; COMPUTER-AIDED DETECTION; HIGH-THROUGHPUT; GENE-EXPRESSION; SECONDARY STRUCTURE; DRUG DISCOVERY; SINGLE-CELL; DECISION-SUPPORT;

D O I：

10.1098/rsif.2017.0387

中图分类号：

O [数理科学和化学]; P [天文学、地球科学]; Q [生物科学]; N [自然科学总论];

学科分类号：

07 ; 0710 ; 09 ;

摘要：

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.

引用

页数：47

共 546 条

[11] Classification of Breast MRI Lesions using Small-Size Training Sets: Comparison of Deep Learning Approaches [J].