Automated deep learning design for medical image classification by health-care professionals with no coding experience: a feasibility study

Deep learning has the potential to transform health care; however, substantial expertise is required to train such models. We sought to evaluate the utility of automated deep learning software to develop medical image diagnostic classifiers by health-care professionals with no coding—and no deep lea...

Full description

Saved in:
Bibliographic Details
Published inThe Lancet. Digital health Vol. 1; no. 5; pp. e232 - e242
Main Authors Faes, Livia, Wagner, Siegfried K, Fu, Dun Jack, Liu, Xiaoxuan, Korot, Edward, Ledsam, Joseph R, Back, Trevor, Chopra, Reena, Pontikos, Nikolas, Kern, Christoph, Moraes, Gabriella, Schmid, Martin K, Sim, Dawn, Balaskas, Konstantinos, Bachmann, Lucas M, Denniston, Alastair K, Keane, Pearse A
Format Journal Article
LanguageEnglish
Published England Elsevier Ltd 01.09.2019
Elsevier
Subjects
Online AccessGet full text
ISSN2589-7500
2589-7500
DOI10.1016/S2589-7500(19)30108-6

Cover

Abstract Deep learning has the potential to transform health care; however, substantial expertise is required to train such models. We sought to evaluate the utility of automated deep learning software to develop medical image diagnostic classifiers by health-care professionals with no coding—and no deep learning—expertise. We used five publicly available open-source datasets: retinal fundus images (MESSIDOR); optical coherence tomography (OCT) images (Guangzhou Medical University and Shiley Eye Institute, version 3); images of skin lesions (Human Against Machine [HAM] 10000), and both paediatric and adult chest x-ray (CXR) images (Guangzhou Medical University and Shiley Eye Institute, version 3 and the National Institute of Health [NIH] dataset, respectively) to separately feed into a neural architecture search framework, hosted through Google Cloud AutoML, that automatically developed a deep learning architecture to classify common diseases. Sensitivity (recall), specificity, and positive predictive value (precision) were used to evaluate the diagnostic properties of the models. The discriminative performance was assessed using the area under the precision recall curve (AUPRC). In the case of the deep learning model developed on a subset of the HAM10000 dataset, we did external validation using the Edinburgh Dermofit Library dataset. Diagnostic properties and discriminative performance from internal validations were high in the binary classification tasks (sensitivity 73·3–97·0%; specificity 67–100%; AUPRC 0·87–1·00). In the multiple classification tasks, the diagnostic properties ranged from 38% to 100% for sensitivity and from 67% to 100% for specificity. The discriminative performance in terms of AUPRC ranged from 0·57 to 1·00 in the five automated deep learning models. In an external validation using the Edinburgh Dermofit Library dataset, the automated deep learning model showed an AUPRC of 0·47, with a sensitivity of 49% and a positive predictive value of 52%. All models, except the automated deep learning model trained on the multilabel classification task of the NIH CXR14 dataset, showed comparable discriminative performance and diagnostic properties to state-of-the-art performing deep learning algorithms. The performance in the external validation study was low. The quality of the open-access datasets (including insufficient information about patient flow and demographics) and the absence of measurement for precision, such as confidence intervals, constituted the major limitations of this study. The availability of automated deep learning platforms provide an opportunity for the medical community to enhance their understanding in model development and evaluation. Although the derivation of classification models without requiring a deep understanding of the mathematical, statistical, and programming principles is attractive, comparable performance to expertly designed models is limited to more elementary classification tasks. Furthermore, care should be placed in adhering to ethical principles when using these automated models to avoid discrimination and causing harm. Future studies should compare several application programming interfaces on thoroughly curated datasets. National Institute for Health Research and Moorfields Eye Charity.
AbstractList Deep learning has the potential to transform health care; however, substantial expertise is required to train such models. We sought to evaluate the utility of automated deep learning software to develop medical image diagnostic classifiers by health-care professionals with no coding—and no deep learning—expertise. We used five publicly available open-source datasets: retinal fundus images (MESSIDOR); optical coherence tomography (OCT) images (Guangzhou Medical University and Shiley Eye Institute, version 3); images of skin lesions (Human Against Machine [HAM] 10000), and both paediatric and adult chest x-ray (CXR) images (Guangzhou Medical University and Shiley Eye Institute, version 3 and the National Institute of Health [NIH] dataset, respectively) to separately feed into a neural architecture search framework, hosted through Google Cloud AutoML, that automatically developed a deep learning architecture to classify common diseases. Sensitivity (recall), specificity, and positive predictive value (precision) were used to evaluate the diagnostic properties of the models. The discriminative performance was assessed using the area under the precision recall curve (AUPRC). In the case of the deep learning model developed on a subset of the HAM10000 dataset, we did external validation using the Edinburgh Dermofit Library dataset. Diagnostic properties and discriminative performance from internal validations were high in the binary classification tasks (sensitivity 73·3–97·0%; specificity 67–100%; AUPRC 0·87–1·00). In the multiple classification tasks, the diagnostic properties ranged from 38% to 100% for sensitivity and from 67% to 100% for specificity. The discriminative performance in terms of AUPRC ranged from 0·57 to 1·00 in the five automated deep learning models. In an external validation using the Edinburgh Dermofit Library dataset, the automated deep learning model showed an AUPRC of 0·47, with a sensitivity of 49% and a positive predictive value of 52%. All models, except the automated deep learning model trained on the multilabel classification task of the NIH CXR14 dataset, showed comparable discriminative performance and diagnostic properties to state-of-the-art performing deep learning algorithms. The performance in the external validation study was low. The quality of the open-access datasets (including insufficient information about patient flow and demographics) and the absence of measurement for precision, such as confidence intervals, constituted the major limitations of this study. The availability of automated deep learning platforms provide an opportunity for the medical community to enhance their understanding in model development and evaluation. Although the derivation of classification models without requiring a deep understanding of the mathematical, statistical, and programming principles is attractive, comparable performance to expertly designed models is limited to more elementary classification tasks. Furthermore, care should be placed in adhering to ethical principles when using these automated models to avoid discrimination and causing harm. Future studies should compare several application programming interfaces on thoroughly curated datasets. National Institute for Health Research and Moorfields Eye Charity.
Background: Deep learning has the potential to transform health care; however, substantial expertise is required to train such models. We sought to evaluate the utility of automated deep learning software to develop medical image diagnostic classifiers by health-care professionals with no coding—and no deep learning—expertise. Methods: We used five publicly available open-source datasets: retinal fundus images (MESSIDOR); optical coherence tomography (OCT) images (Guangzhou Medical University and Shiley Eye Institute, version 3); images of skin lesions (Human Against Machine [HAM] 10000), and both paediatric and adult chest x-ray (CXR) images (Guangzhou Medical University and Shiley Eye Institute, version 3 and the National Institute of Health [NIH] dataset, respectively) to separately feed into a neural architecture search framework, hosted through Google Cloud AutoML, that automatically developed a deep learning architecture to classify common diseases. Sensitivity (recall), specificity, and positive predictive value (precision) were used to evaluate the diagnostic properties of the models. The discriminative performance was assessed using the area under the precision recall curve (AUPRC). In the case of the deep learning model developed on a subset of the HAM10000 dataset, we did external validation using the Edinburgh Dermofit Library dataset. Findings: Diagnostic properties and discriminative performance from internal validations were high in the binary classification tasks (sensitivity 73·3–97·0%; specificity 67–100%; AUPRC 0·87–1·00). In the multiple classification tasks, the diagnostic properties ranged from 38% to 100% for sensitivity and from 67% to 100% for specificity. The discriminative performance in terms of AUPRC ranged from 0·57 to 1·00 in the five automated deep learning models. In an external validation using the Edinburgh Dermofit Library dataset, the automated deep learning model showed an AUPRC of 0·47, with a sensitivity of 49% and a positive predictive value of 52%. Interpretation: All models, except the automated deep learning model trained on the multilabel classification task of the NIH CXR14 dataset, showed comparable discriminative performance and diagnostic properties to state-of-the-art performing deep learning algorithms. The performance in the external validation study was low. The quality of the open-access datasets (including insufficient information about patient flow and demographics) and the absence of measurement for precision, such as confidence intervals, constituted the major limitations of this study. The availability of automated deep learning platforms provide an opportunity for the medical community to enhance their understanding in model development and evaluation. Although the derivation of classification models without requiring a deep understanding of the mathematical, statistical, and programming principles is attractive, comparable performance to expertly designed models is limited to more elementary classification tasks. Furthermore, care should be placed in adhering to ethical principles when using these automated models to avoid discrimination and causing harm. Future studies should compare several application programming interfaces on thoroughly curated datasets. Funding: National Institute for Health Research and Moorfields Eye Charity.
Deep learning has the potential to transform health care; however, substantial expertise is required to train such models. We sought to evaluate the utility of automated deep learning software to develop medical image diagnostic classifiers by health-care professionals with no coding-and no deep learning-expertise.BACKGROUNDDeep learning has the potential to transform health care; however, substantial expertise is required to train such models. We sought to evaluate the utility of automated deep learning software to develop medical image diagnostic classifiers by health-care professionals with no coding-and no deep learning-expertise.We used five publicly available open-source datasets: retinal fundus images (MESSIDOR); optical coherence tomography (OCT) images (Guangzhou Medical University and Shiley Eye Institute, version 3); images of skin lesions (Human Against Machine [HAM] 10000), and both paediatric and adult chest x-ray (CXR) images (Guangzhou Medical University and Shiley Eye Institute, version 3 and the National Institute of Health [NIH] dataset, respectively) to separately feed into a neural architecture search framework, hosted through Google Cloud AutoML, that automatically developed a deep learning architecture to classify common diseases. Sensitivity (recall), specificity, and positive predictive value (precision) were used to evaluate the diagnostic properties of the models. The discriminative performance was assessed using the area under the precision recall curve (AUPRC). In the case of the deep learning model developed on a subset of the HAM10000 dataset, we did external validation using the Edinburgh Dermofit Library dataset.METHODSWe used five publicly available open-source datasets: retinal fundus images (MESSIDOR); optical coherence tomography (OCT) images (Guangzhou Medical University and Shiley Eye Institute, version 3); images of skin lesions (Human Against Machine [HAM] 10000), and both paediatric and adult chest x-ray (CXR) images (Guangzhou Medical University and Shiley Eye Institute, version 3 and the National Institute of Health [NIH] dataset, respectively) to separately feed into a neural architecture search framework, hosted through Google Cloud AutoML, that automatically developed a deep learning architecture to classify common diseases. Sensitivity (recall), specificity, and positive predictive value (precision) were used to evaluate the diagnostic properties of the models. The discriminative performance was assessed using the area under the precision recall curve (AUPRC). In the case of the deep learning model developed on a subset of the HAM10000 dataset, we did external validation using the Edinburgh Dermofit Library dataset.Diagnostic properties and discriminative performance from internal validations were high in the binary classification tasks (sensitivity 73·3-97·0%; specificity 67-100%; AUPRC 0·87-1·00). In the multiple classification tasks, the diagnostic properties ranged from 38% to 100% for sensitivity and from 67% to 100% for specificity. The discriminative performance in terms of AUPRC ranged from 0·57 to 1·00 in the five automated deep learning models. In an external validation using the Edinburgh Dermofit Library dataset, the automated deep learning model showed an AUPRC of 0·47, with a sensitivity of 49% and a positive predictive value of 52%.FINDINGSDiagnostic properties and discriminative performance from internal validations were high in the binary classification tasks (sensitivity 73·3-97·0%; specificity 67-100%; AUPRC 0·87-1·00). In the multiple classification tasks, the diagnostic properties ranged from 38% to 100% for sensitivity and from 67% to 100% for specificity. The discriminative performance in terms of AUPRC ranged from 0·57 to 1·00 in the five automated deep learning models. In an external validation using the Edinburgh Dermofit Library dataset, the automated deep learning model showed an AUPRC of 0·47, with a sensitivity of 49% and a positive predictive value of 52%.All models, except the automated deep learning model trained on the multilabel classification task of the NIH CXR14 dataset, showed comparable discriminative performance and diagnostic properties to state-of-the-art performing deep learning algorithms. The performance in the external validation study was low. The quality of the open-access datasets (including insufficient information about patient flow and demographics) and the absence of measurement for precision, such as confidence intervals, constituted the major limitations of this study. The availability of automated deep learning platforms provide an opportunity for the medical community to enhance their understanding in model development and evaluation. Although the derivation of classification models without requiring a deep understanding of the mathematical, statistical, and programming principles is attractive, comparable performance to expertly designed models is limited to more elementary classification tasks. Furthermore, care should be placed in adhering to ethical principles when using these automated models to avoid discrimination and causing harm. Future studies should compare several application programming interfaces on thoroughly curated datasets.INTERPRETATIONAll models, except the automated deep learning model trained on the multilabel classification task of the NIH CXR14 dataset, showed comparable discriminative performance and diagnostic properties to state-of-the-art performing deep learning algorithms. The performance in the external validation study was low. The quality of the open-access datasets (including insufficient information about patient flow and demographics) and the absence of measurement for precision, such as confidence intervals, constituted the major limitations of this study. The availability of automated deep learning platforms provide an opportunity for the medical community to enhance their understanding in model development and evaluation. Although the derivation of classification models without requiring a deep understanding of the mathematical, statistical, and programming principles is attractive, comparable performance to expertly designed models is limited to more elementary classification tasks. Furthermore, care should be placed in adhering to ethical principles when using these automated models to avoid discrimination and causing harm. Future studies should compare several application programming interfaces on thoroughly curated datasets.National Institute for Health Research and Moorfields Eye Charity.FUNDINGNational Institute for Health Research and Moorfields Eye Charity.
SummaryBackgroundDeep learning has the potential to transform health care; however, substantial expertise is required to train such models. We sought to evaluate the utility of automated deep learning software to develop medical image diagnostic classifiers by health-care professionals with no coding—and no deep learning—expertise. MethodsWe used five publicly available open-source datasets: retinal fundus images (MESSIDOR); optical coherence tomography (OCT) images (Guangzhou Medical University and Shiley Eye Institute, version 3); images of skin lesions (Human Against Machine [HAM] 10000), and both paediatric and adult chest x-ray (CXR) images (Guangzhou Medical University and Shiley Eye Institute, version 3 and the National Institute of Health [NIH] dataset, respectively) to separately feed into a neural architecture search framework, hosted through Google Cloud AutoML, that automatically developed a deep learning architecture to classify common diseases. Sensitivity (recall), specificity, and positive predictive value (precision) were used to evaluate the diagnostic properties of the models. The discriminative performance was assessed using the area under the precision recall curve (AUPRC). In the case of the deep learning model developed on a subset of the HAM10000 dataset, we did external validation using the Edinburgh Dermofit Library dataset. FindingsDiagnostic properties and discriminative performance from internal validations were high in the binary classification tasks (sensitivity 73·3–97·0%; specificity 67–100%; AUPRC 0·87–1·00). In the multiple classification tasks, the diagnostic properties ranged from 38% to 100% for sensitivity and from 67% to 100% for specificity. The discriminative performance in terms of AUPRC ranged from 0·57 to 1·00 in the five automated deep learning models. In an external validation using the Edinburgh Dermofit Library dataset, the automated deep learning model showed an AUPRC of 0·47, with a sensitivity of 49% and a positive predictive value of 52%. InterpretationAll models, except the automated deep learning model trained on the multilabel classification task of the NIH CXR14 dataset, showed comparable discriminative performance and diagnostic properties to state-of-the-art performing deep learning algorithms. The performance in the external validation study was low. The quality of the open-access datasets (including insufficient information about patient flow and demographics) and the absence of measurement for precision, such as confidence intervals, constituted the major limitations of this study. The availability of automated deep learning platforms provide an opportunity for the medical community to enhance their understanding in model development and evaluation. Although the derivation of classification models without requiring a deep understanding of the mathematical, statistical, and programming principles is attractive, comparable performance to expertly designed models is limited to more elementary classification tasks. Furthermore, care should be placed in adhering to ethical principles when using these automated models to avoid discrimination and causing harm. Future studies should compare several application programming interfaces on thoroughly curated datasets. FundingNational Institute for Health Research and Moorfields Eye Charity.
Author Faes, Livia
Wagner, Siegfried K
Balaskas, Konstantinos
Moraes, Gabriella
Denniston, Alastair K
Keane, Pearse A
Fu, Dun Jack
Korot, Edward
Ledsam, Joseph R
Back, Trevor
Chopra, Reena
Liu, Xiaoxuan
Pontikos, Nikolas
Kern, Christoph
Sim, Dawn
Schmid, Martin K
Bachmann, Lucas M
Author_xml – sequence: 1
  givenname: Livia
  surname: Faes
  fullname: Faes, Livia
  organization: Department of Ophthalmology, Cantonal Hospital Lucerne, Lucerne, Switzerland
– sequence: 2
  givenname: Siegfried K
  surname: Wagner
  fullname: Wagner, Siegfried K
  organization: National Institute of Health Research Biomedical Research Center, Moorfields Eye Hospital National Health Service Foundation Trust, and University College London Institute of Ophthalmology, London, UK
– sequence: 3
  givenname: Dun Jack
  surname: Fu
  fullname: Fu, Dun Jack
  organization: Medical Retina Department, Moorfields Eye Hospital National Health Service Foundation Trust, London, UK
– sequence: 4
  givenname: Xiaoxuan
  surname: Liu
  fullname: Liu, Xiaoxuan
  organization: National Institute of Health Research Biomedical Research Center, Moorfields Eye Hospital National Health Service Foundation Trust, and University College London Institute of Ophthalmology, London, UK
– sequence: 5
  givenname: Edward
  surname: Korot
  fullname: Korot, Edward
  organization: Medical Retina Department, Moorfields Eye Hospital National Health Service Foundation Trust, London, UK
– sequence: 6
  givenname: Joseph R
  surname: Ledsam
  fullname: Ledsam, Joseph R
  organization: DeepMind, London, UK
– sequence: 7
  givenname: Trevor
  surname: Back
  fullname: Back, Trevor
  organization: DeepMind, London, UK
– sequence: 8
  givenname: Reena
  surname: Chopra
  fullname: Chopra, Reena
  organization: National Institute of Health Research Biomedical Research Center, Moorfields Eye Hospital National Health Service Foundation Trust, and University College London Institute of Ophthalmology, London, UK
– sequence: 9
  givenname: Nikolas
  surname: Pontikos
  fullname: Pontikos, Nikolas
  organization: National Institute of Health Research Biomedical Research Center, Moorfields Eye Hospital National Health Service Foundation Trust, and University College London Institute of Ophthalmology, London, UK
– sequence: 10
  givenname: Christoph
  surname: Kern
  fullname: Kern, Christoph
  organization: Medical Retina Department, Moorfields Eye Hospital National Health Service Foundation Trust, London, UK
– sequence: 11
  givenname: Gabriella
  surname: Moraes
  fullname: Moraes, Gabriella
  organization: Medical Retina Department, Moorfields Eye Hospital National Health Service Foundation Trust, London, UK
– sequence: 12
  givenname: Martin K
  surname: Schmid
  fullname: Schmid, Martin K
  organization: Department of Ophthalmology, Cantonal Hospital Lucerne, Lucerne, Switzerland
– sequence: 13
  givenname: Dawn
  surname: Sim
  fullname: Sim, Dawn
  organization: National Institute of Health Research Biomedical Research Center, Moorfields Eye Hospital National Health Service Foundation Trust, and University College London Institute of Ophthalmology, London, UK
– sequence: 14
  givenname: Konstantinos
  surname: Balaskas
  fullname: Balaskas, Konstantinos
  organization: National Institute of Health Research Biomedical Research Center, Moorfields Eye Hospital National Health Service Foundation Trust, and University College London Institute of Ophthalmology, London, UK
– sequence: 15
  givenname: Lucas M
  surname: Bachmann
  fullname: Bachmann, Lucas M
  organization: Medigntion, Zurich, Switzerland
– sequence: 16
  givenname: Alastair K
  surname: Denniston
  fullname: Denniston, Alastair K
  organization: National Institute of Health Research Biomedical Research Center, Moorfields Eye Hospital National Health Service Foundation Trust, and University College London Institute of Ophthalmology, London, UK
– sequence: 17
  givenname: Pearse A
  surname: Keane
  fullname: Keane, Pearse A
  email: pearse.keane1@nhs.net
  organization: National Institute of Health Research Biomedical Research Center, Moorfields Eye Hospital National Health Service Foundation Trust, and University College London Institute of Ophthalmology, London, UK
BackLink https://www.ncbi.nlm.nih.gov/pubmed/33323271$$D View this record in MEDLINE/PubMed
BookMark eNqFks1u1DAUhSNUREvpI4C8LIuAf5I4AYFUVfxUqsSisLYc53rGxWMPtkPJM_DSOJMyQpVGrOzcnPP5Xh8_LY6cd1AUzwl-RTBpXt_Quu1KXmN8TrqXDBPcls2j4mRfPvpnf1ycxXiLMaaUMM75k-KYMUYZ5eSk-H0xJr-RCQY0AGyRBRmccav8Fc3KIe0D2sBglLTIbOQKkLIyRqNzJRnvUD-hNUib1qWSAdA2eA35v3fSRnRn0ho5j5QfZib82kIw4BS8QRJpkNH0xpo0oZjGYXpWPNbZBWf362nx7eOHr5efy-svn64uL65LVTOSyrZllWp7qhXXHLdNV9EG80bBQKHpJPQDZpWWSnGKO056rPpuaBnNGkoIVuy0uFq4g5e3YhvyXGESXhqxK_iwEjIkoyyIXvOhogPva6UqJqt57XWdt_m8VrWZ1Sys0W3ldCet3QMJFnNYIs5JiDkJQTqxC0s02Xi-GPON_RghJrExUYG10oEfo6AVxw0jHeNZ-uJeOvY5i_0Bf1PMgreLQAUfYwAtlEm7eFKQxu47uTnQSf3A_XCCQ773iw9yVj8NBKGscfNL-Q4TxFs_hvkRCJJvQOAFMjPyTDNhBrw7DMhZmP808AdrCO1N
CitedBy_id crossref_primary_10_1148_ryai_220062
crossref_primary_10_1016_j_apjo_2024_100089
crossref_primary_10_1542_peds_2020_034546
crossref_primary_10_1109_ACCESS_2020_3004766
crossref_primary_10_25056_JCM_2023_7_2_60
crossref_primary_10_1016_j_xops_2024_100470
crossref_primary_10_1016_S2589_7500_19_30112_8
crossref_primary_10_7759_cureus_46454
crossref_primary_10_1016_j_imu_2022_100853
crossref_primary_10_1007_s42835_024_01919_3
crossref_primary_10_3390_bdcc8110157
crossref_primary_10_1016_j_ceramint_2024_01_298
crossref_primary_10_1038_s41598_022_06127_5
crossref_primary_10_1016_S0140_6736_20_30813_8
crossref_primary_10_21923_jesd_1121792
crossref_primary_10_1038_s41591_023_02293_9
crossref_primary_10_1186_s40942_024_00555_3
crossref_primary_10_1016_j_media_2021_102306
crossref_primary_10_1038_s41598_024_60429_4
crossref_primary_10_1136_bjophthalmol_2021_319030
crossref_primary_10_3390_diagnostics10110910
crossref_primary_10_1007_s12194_019_00552_4
crossref_primary_10_3390_biomedicines10071544
crossref_primary_10_1016_j_jns_2022_120454
crossref_primary_10_1001_jamaophthalmol_2023_6318
crossref_primary_10_1109_ACCESS_2024_3441469
crossref_primary_10_1007_s11547_024_01770_6
crossref_primary_10_1016_j_compbiomed_2023_106649
crossref_primary_10_32604_cmc_2022_024965
crossref_primary_10_5772_dmht_20
crossref_primary_10_1007_s10916_023_01928_1
crossref_primary_10_1111_exsy_12690
crossref_primary_10_1016_j_csbj_2020_08_003
crossref_primary_10_3389_frobt_2022_896028
crossref_primary_10_1007_s00521_021_05943_6
crossref_primary_10_1097_BRS_0000000000003844
crossref_primary_10_4274_dir_2024_242972
crossref_primary_10_1002_wsbm_1501
crossref_primary_10_1167_tvst_13_4_4
crossref_primary_10_32604_csse_2023_035900
crossref_primary_10_1093_pcmedi_pbaa029
crossref_primary_10_32604_cmc_2022_028560
crossref_primary_10_1007_s12350_020_02119_y
crossref_primary_10_3390_ijerph191912200
crossref_primary_10_1371_journal_pone_0273508
crossref_primary_10_1016_j_acra_2022_07_011
crossref_primary_10_1016_j_xops_2024_100495
crossref_primary_10_1002_aesr_202300004
crossref_primary_10_1002_jum_16194
crossref_primary_10_3390_ai3030043
crossref_primary_10_1016_j_patcog_2021_107825
crossref_primary_10_1186_s12880_023_01017_2
crossref_primary_10_1016_j_ejrad_2022_110369
crossref_primary_10_1016_j_ophtha_2022_01_002
crossref_primary_10_1016_j_engappai_2023_107164
crossref_primary_10_2196_40167
crossref_primary_10_3390_jcm13144141
crossref_primary_10_1097_ICU_0000000000000779
crossref_primary_10_1038_s41598_020_76665_3
crossref_primary_10_1016_j_ajoms_2022_02_004
crossref_primary_10_1016_j_apenergy_2021_118049
crossref_primary_10_1080_09273948_2024_2319281
crossref_primary_10_1145_3506695
crossref_primary_10_1016_j_artmed_2023_102547
crossref_primary_10_1093_jnen_nlab005
crossref_primary_10_1097_MOU_0000000000000813
crossref_primary_10_1038_s41598_021_89369_z
crossref_primary_10_1016_S2214_109X_23_00323_6
crossref_primary_10_1136_rmdopen_2023_003105
crossref_primary_10_1186_s12911_025_02950_8
crossref_primary_10_1093_comjnl_bxaa145
crossref_primary_10_1093_jamiaopen_ooac094
crossref_primary_10_1097_01_APO_0000769904_75814_b5
crossref_primary_10_3390_app142411926
crossref_primary_10_1080_10106049_2023_2236576
crossref_primary_10_1097_RUQ_0000000000000683
crossref_primary_10_1016_j_oret_2023_03_003
crossref_primary_10_1016_S2589_7500_23_00050_X
crossref_primary_10_1097_ICU_0000000000000785
crossref_primary_10_1001_jamaophthalmol_2023_4508
crossref_primary_10_2196_49949
crossref_primary_10_1145_3533378
crossref_primary_10_3171_2022_1_FOCUS21652
crossref_primary_10_3390_computers12090174
crossref_primary_10_1016_j_imr_2022_100888
crossref_primary_10_32604_cmc_2020_013125
crossref_primary_10_1016_j_xops_2021_100036
crossref_primary_10_3390_su12093612
crossref_primary_10_1016_j_cels_2023_05_007
crossref_primary_10_1007_s10462_024_11080_y
crossref_primary_10_2174_1875036202114010093
crossref_primary_10_1016_j_bspc_2020_102329
crossref_primary_10_1109_JSTARS_2022_3232583
crossref_primary_10_3390_jcm11030614
crossref_primary_10_1016_j_bspc_2024_107108
crossref_primary_10_3389_fradi_2025_1503625
crossref_primary_10_1109_RBME_2020_3013489
crossref_primary_10_32604_cmc_2023_039518
crossref_primary_10_1097_ICU_0000000000000677
crossref_primary_10_1007_s40820_024_01536_9
crossref_primary_10_1097_ICU_0000000000000678
crossref_primary_10_1002_jcu_23143
crossref_primary_10_1007_s12652_022_03835_8
crossref_primary_10_1038_s42256_021_00305_2
crossref_primary_10_1016_j_ebiom_2024_105463
crossref_primary_10_1038_s41746_023_00759_1
crossref_primary_10_1038_s41598_023_32118_1
crossref_primary_10_1097_APO_0000000000000398
crossref_primary_10_1007_s00330_023_09566_4
crossref_primary_10_1038_s41598_021_89743_x
crossref_primary_10_1146_annurev_bioeng_110220_012203
crossref_primary_10_1007_s10278_022_00724_6
crossref_primary_10_1051_0004_6361_202142998
crossref_primary_10_1007_s11082_023_06076_x
crossref_primary_10_2329_perio_63_119
crossref_primary_10_1080_08820538_2023_2168486
crossref_primary_10_3390_vision8030048
crossref_primary_10_1016_j_knosys_2020_106622
crossref_primary_10_1016_j_preteyeres_2025_101350
crossref_primary_10_1016_j_compbiomed_2023_107777
crossref_primary_10_1093_bioinformatics_btab380
crossref_primary_10_1093_psyrad_kkab009
crossref_primary_10_1016_j_gastha_2022_02_025
crossref_primary_10_1016_j_eswa_2023_121245
crossref_primary_10_1292_jvms_23_0299
crossref_primary_10_2139_ssrn_4074672
crossref_primary_10_3389_fneur_2021_735142
crossref_primary_10_1007_s11082_023_06018_7
crossref_primary_10_1002_cpe_6751
crossref_primary_10_3390_diagnostics11020233
crossref_primary_10_1007_s00417_021_05544_y
crossref_primary_10_2196_43638
crossref_primary_10_1109_MCSE_2020_3009765
crossref_primary_10_1186_s12880_024_01543_7
crossref_primary_10_1159_000525929
crossref_primary_10_1007_s11082_023_06168_8
crossref_primary_10_1016_j_slast_2024_100192
crossref_primary_10_1097_ICU_0000000000000693
crossref_primary_10_1136_bmjophth_2022_000992
crossref_primary_10_1088_1742_6596_2096_1_012028
crossref_primary_10_1167_tvst_10_7_14
crossref_primary_10_32604_cmc_2022_019529
crossref_primary_10_1002_acr2_11665
crossref_primary_10_1016_j_mechmachtheory_2022_104742
crossref_primary_10_1002_int_22449
crossref_primary_10_1016_j_measurement_2020_107703
crossref_primary_10_1371_journal_pdig_0000058
crossref_primary_10_47164_ijngc_v13i3_663
crossref_primary_10_1007_s00417_022_05741_3
crossref_primary_10_3390_app13095472
crossref_primary_10_1590_2318_0889202436e2410917
crossref_primary_10_1136_bjophthalmol_2022_321141
crossref_primary_10_3390_pr8020224
crossref_primary_10_1016_j_neucom_2024_129182
crossref_primary_10_3390_su122310124
crossref_primary_10_1016_j_compmedimag_2024_102441
crossref_primary_10_3390_computers10020024
crossref_primary_10_1038_s41598_024_72889_9
crossref_primary_10_1007_s12652_021_02948_w
crossref_primary_10_4103_sjopt_sjopt_106_22
crossref_primary_10_1038_s41598_024_60807_y
crossref_primary_10_5051_jpis_2104080204
crossref_primary_10_1136_bmjophth_2024_001873
crossref_primary_10_1093_jnen_nlac131
crossref_primary_10_1093_bjd_ljae040
crossref_primary_10_1177_03000605231200371
crossref_primary_10_32604_cmc_2023_041722
crossref_primary_10_3390_healthcare10101940
crossref_primary_10_1007_s11042_025_20633_4
crossref_primary_10_1016_S0140_6736_21_00722_4
Cites_doi 10.1038/nature14539
10.1002/rob.20276
10.1016/j.cell.2018.02.010
10.1109/IROS.2008.4651217
10.5566/ias.1155
10.1109/MSP.2012.2205597
10.1038/nrc1550
10.1056/NEJMp1714229
10.1038/s41591-018-0107-6
10.1016/j.media.2017.12.002
10.1111/acem.12255
10.1038/s41746-019-0079-z
10.1038/nature14236
10.1038/sdata.2018.161
10.1136/bmj.g7594
10.1038/s41591-018-0268-3
ContentType Journal Article
Copyright 2019 The Author(s). Published by Elsevier Ltd. This is an Open Access article under the CC BY 4.0 license.
The Author(s). Published by Elsevier Ltd. This is an Open Access article under the CC BY 4.0 license.
Copyright © 2019 The Author(s). Published by Elsevier Ltd. This is an Open Access article under the CC BY 4.0 license. Published by Elsevier Ltd.. All rights reserved.
Copyright_xml – notice: 2019 The Author(s). Published by Elsevier Ltd. This is an Open Access article under the CC BY 4.0 license.
– notice: The Author(s). Published by Elsevier Ltd. This is an Open Access article under the CC BY 4.0 license.
– notice: Copyright © 2019 The Author(s). Published by Elsevier Ltd. This is an Open Access article under the CC BY 4.0 license. Published by Elsevier Ltd.. All rights reserved.
DBID AAYXX
CITATION
CGR
CUY
CVF
ECM
EIF
NPM
7X8
ADTOC
UNPAY
DOA
DOI 10.1016/S2589-7500(19)30108-6
DatabaseName CrossRef
Medline
MEDLINE
MEDLINE (Ovid)
MEDLINE
MEDLINE
PubMed
MEDLINE - Academic
Unpaywall for CDI: Periodical Content
Unpaywall
DOAJ Directory of Open Access Journals
DatabaseTitle CrossRef
MEDLINE
Medline Complete
MEDLINE with Full Text
PubMed
MEDLINE (Ovid)
MEDLINE - Academic
DatabaseTitleList

MEDLINE
MEDLINE - Academic

Database_xml – sequence: 1
  dbid: DOA
  name: DOAJ Directory of Open Access Journals (ODIN)
  url: https://www.doaj.org/
  sourceTypes: Open Website
– sequence: 2
  dbid: NPM
  name: PubMed
  url: https://proxy.k.utb.cz/login?url=http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed
  sourceTypes: Index Database
– sequence: 3
  dbid: EIF
  name: MEDLINE
  url: https://proxy.k.utb.cz/login?url=https://www.webofscience.com/wos/medline/basic-search
  sourceTypes: Index Database
– sequence: 4
  dbid: UNPAY
  name: Unpaywall
  url: https://proxy.k.utb.cz/login?url=https://unpaywall.org/
  sourceTypes: Open Access Repository
DeliveryMethod fulltext_linktorsrc
Discipline Public Health
EISSN 2589-7500
EndPage e242
ExternalDocumentID oai_doaj_org_article_bf7d42d7b5cc43a4b5ccbf543a6ce8c8
10.1016/s2589-7500(19)30108-6
33323271
10_1016_S2589_7500_19_30108_6
1_s2_0_S2589750019301086
S2589750019301086
Genre Research Support, Non-U.S. Gov't
Journal Article
GrantInformation National Institute for Health Research and Moorfields Eye Charity.
GrantInformation_xml – fundername: Department of Health
  grantid: NIHR-CS-2014-14-023
– fundername: Medical Research Council
  grantid: MC_PC_19005
– fundername: Department of Health
  grantid: CS-2014-14-023
GroupedDBID .1-
.FO
0R~
53G
AAEDW
AALRI
AAMRU
AAXUO
ACLIJ
ACVFH
ADCNI
AEUPX
AEXQZ
AFPUW
AFRHN
AIGII
AITUG
AJUYK
AKBMS
AKYEP
ALMA_UNASSIGNED_HOLDINGS
APXCP
EBS
EJD
FDB
GROUPED_DOAJ
M41
M~E
OK1
ROL
W2D
Z5R
NCXOZ
AAYXX
CITATION
CGR
CUY
CVF
ECM
EIF
NPM
7X8
ADTOC
UNPAY
ID FETCH-LOGICAL-c531t-8834c8b2fc7f70869426076ced2e69aebd034facc720971b0cb9d8320762110c3
IEDL.DBID DOA
ISSN 2589-7500
IngestDate Fri Oct 03 12:52:00 EDT 2025
Tue Aug 19 17:25:59 EDT 2025
Wed Oct 01 12:49:55 EDT 2025
Mon Jul 21 05:34:59 EDT 2025
Tue Jul 01 02:13:49 EDT 2025
Thu Apr 24 22:51:35 EDT 2025
Sun Feb 23 10:19:02 EST 2025
Tue Aug 26 20:17:41 EDT 2025
IsDoiOpenAccess true
IsOpenAccess true
IsPeerReviewed true
IsScholarly true
Issue 5
Language English
License This is an open access article under the CC BY license.
Copyright © 2019 The Author(s). Published by Elsevier Ltd. This is an Open Access article under the CC BY 4.0 license. Published by Elsevier Ltd.. All rights reserved.
cc-by
LinkModel DirectLink
MergedId FETCHMERGED-LOGICAL-c531t-8834c8b2fc7f70869426076ced2e69aebd034facc720971b0cb9d8320762110c3
Notes ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 23
OpenAccessLink https://doaj.org/article/bf7d42d7b5cc43a4b5ccbf543a6ce8c8
PMID 33323271
PQID 2470631937
PQPubID 23479
ParticipantIDs doaj_primary_oai_doaj_org_article_bf7d42d7b5cc43a4b5ccbf543a6ce8c8
unpaywall_primary_10_1016_s2589_7500_19_30108_6
proquest_miscellaneous_2470631937
pubmed_primary_33323271
crossref_citationtrail_10_1016_S2589_7500_19_30108_6
crossref_primary_10_1016_S2589_7500_19_30108_6
elsevier_clinicalkeyesjournals_1_s2_0_S2589750019301086
elsevier_clinicalkey_doi_10_1016_S2589_7500_19_30108_6
ProviderPackageCode CITATION
AAYXX
PublicationCentury 2000
PublicationDate 2019-09-01
PublicationDateYYYYMMDD 2019-09-01
PublicationDate_xml – month: 09
  year: 2019
  text: 2019-09-01
  day: 01
PublicationDecade 2010
PublicationPlace England
PublicationPlace_xml – name: England
PublicationTitle The Lancet. Digital health
PublicationTitleAlternate Lancet Digit Health
PublicationYear 2019
Publisher Elsevier Ltd
Elsevier
Publisher_xml – name: Elsevier Ltd
– name: Elsevier
References Nicholson Price (bib35) 2017
Char, Shah, Magnus (bib43) 2018; 378
Luo, Phung, Tran (bib31) 2016; 18
Jouppi (bib11) May 18, 2016
Vinyals, Toshev, Bengio, Erhan (bib4) 2014
Stupple, Singerman, Celi (bib36) 2019; 2
(bib15) Jan 1, 2019
Hannun, Rajpurkar, Haghpanahi (bib12) 2019; 25
LeCun, Bengio, Hinton (bib1) 2015; 521
Guan, Huang, Zhong, Zheng, Zheng, Yang (bib28) 2018
Li, Wang, Liu, Latecki (bib41) 2018; 45
Tschandl, Rosendahl, Kittler (bib22) 2018; 5
Oakden-Rayner (bib39)
Le, Zoph (bib26) May 17, 2017
Mnih, Kavukcuoglu, Silver (bib3) 2015; 518
Kermany, Goldbaum, Cai (bib20) 2018; 172
(bib24) Feb 19, 2018
Fogel, Kvedar (bib2) 2018; 1
Kohn, Carpenter, Newman (bib32) 2013; 20
Hinton, Deng, Yu (bib5) 2012; 29
Hadsell, Sermanet, Ben (bib8) 2009; 26
Li, Pang, Xiong, Liu, Liang, Wang (bib29)
Tonekaboni, Joshi, McCradden, Goldenberg (bib37) 2019
(bib10) Dec 2, 2015
Thomas (bib17)
Lipton (bib38) 2016
Hadsell R, Erkan A, Sermanet P, Scoffier M, Muller U, LeCun Y. Deep belief net learning in a long-range vision system for autonomous off-road driving. 2008 IEEE/RSJ International Conference on Intelligent Robot Systems, IROS; Nice, France; Sept 22–26, 2018 (4651217).
Collobert, Weston, Bottou, Karlen, Kavukcuoglu, Kuksa (bib6) 2011; 12
Zoph, Le (bib18) 2016
bib40
(bib13) April 2, 2019
Ransohoff (bib42) 2005; 5
Oakden-Rayner (bib33)
Collins, Reitsma, Altman, Moons (bib23) 2015; 350
Wang, Peng, Lu, Lu, Bagheri, Summers (bib21)
Elsken, Metzen, Hutter (bib25) 2018
Krizhevsky, Sutskever, Hinton (bib27) 2012
Krizhevsky, Sutskever, Hinton (bib9) 2012; 1
De Fauw, Ledsam, Romera-Paredes (bib34) 2018; 24
Decencière, Zhang, Cazuguel (bib19) 2014; 33
Zoph, Vasudevan, Shlens, Le (bib16) Nov 2, 2017
Codella NC, Gutman D, Celebi ME, et al. Skin lesion analysis toward melanoma detection: a challenge at the 2017 international symposium on biomedical imaging (ISBI), hosted by the international skin imaging collaboration (ISIC). 2018 IEEE 15th International Symposium on Biomedical Imaging; Washington DC, USA; April 4–7, 2018: 168–72.
MSV (bib14) April 15, 2018
Jouppi (10.1016/S2589-7500(19)30108-6_bib11) 2016
Hadsell (10.1016/S2589-7500(19)30108-6_bib8) 2009; 26
De Fauw (10.1016/S2589-7500(19)30108-6_bib34) 2018; 24
Lipton (10.1016/S2589-7500(19)30108-6_bib38) 2016
10.1016/S2589-7500(19)30108-6_bib7
Hannun (10.1016/S2589-7500(19)30108-6_bib12) 2019; 25
Thomas (10.1016/S2589-7500(19)30108-6_bib17)
Stupple (10.1016/S2589-7500(19)30108-6_bib36) 2019; 2
Vinyals (10.1016/S2589-7500(19)30108-6_bib4) 2014
Ransohoff (10.1016/S2589-7500(19)30108-6_bib42) 2005; 5
Le (10.1016/S2589-7500(19)30108-6_bib26) 2017
Decencière (10.1016/S2589-7500(19)30108-6_bib19) 2014; 33
(10.1016/S2589-7500(19)30108-6_bib24) 2018
MSV (10.1016/S2589-7500(19)30108-6_bib14) 2018
Luo (10.1016/S2589-7500(19)30108-6_bib31) 2016; 18
(10.1016/S2589-7500(19)30108-6_bib13) 2019
Oakden-Rayner (10.1016/S2589-7500(19)30108-6_bib39)
10.1016/S2589-7500(19)30108-6_bib30
Fogel (10.1016/S2589-7500(19)30108-6_bib2) 2018; 1
Collins (10.1016/S2589-7500(19)30108-6_bib23) 2015; 350
Guan (10.1016/S2589-7500(19)30108-6_bib28) 2018
Li (10.1016/S2589-7500(19)30108-6_bib29)
Mnih (10.1016/S2589-7500(19)30108-6_bib3) 2015; 518
Char (10.1016/S2589-7500(19)30108-6_bib43) 2018; 378
Elsken (10.1016/S2589-7500(19)30108-6_bib25) 2018
LeCun (10.1016/S2589-7500(19)30108-6_bib1) 2015; 521
(10.1016/S2589-7500(19)30108-6_bib10) 2015
Nicholson Price (10.1016/S2589-7500(19)30108-6_bib35) 2017
Zoph (10.1016/S2589-7500(19)30108-6_bib16) 2017
Oakden-Rayner (10.1016/S2589-7500(19)30108-6_bib33)
Krizhevsky (10.1016/S2589-7500(19)30108-6_bib27) 2012
Tschandl (10.1016/S2589-7500(19)30108-6_bib22) 2018; 5
Collobert (10.1016/S2589-7500(19)30108-6_bib6) 2011; 12
Zoph (10.1016/S2589-7500(19)30108-6_bib18) 2016
Li (10.1016/S2589-7500(19)30108-6_bib41) 2018; 45
Hinton (10.1016/S2589-7500(19)30108-6_bib5) 2012; 29
Kermany (10.1016/S2589-7500(19)30108-6_bib20) 2018; 172
Wang (10.1016/S2589-7500(19)30108-6_bib21)
Tonekaboni (10.1016/S2589-7500(19)30108-6_bib37) 2019
(10.1016/S2589-7500(19)30108-6_bib15) 2019
Kohn (10.1016/S2589-7500(19)30108-6_bib32) 2013; 20
Krizhevsky (10.1016/S2589-7500(19)30108-6_bib9) 2012; 1
33323266 - Lancet Digit Health. 2019 Sep;1(5):e198-e199. doi: 10.1016/S2589-7500(19)30112-8.
References_xml – year: Dec 2, 2015
  ident: bib10
  article-title: World changing ideas of 2015
  publication-title: Scientific American
– ident: bib33
  article-title: Explain yourself, machine. Producing simple text descriptions for AI interpretability
– volume: 1
  start-page: 5
  year: 2018
  ident: bib2
  article-title: Artificial intelligence powers digital medicine
  publication-title: Dig Med
– year: 2014
  ident: bib4
  article-title: Show and tell: a neural image caption generator
  publication-title: arXiv
– volume: 12
  start-page: 2493
  year: 2011
  end-page: 2537
  ident: bib6
  article-title: Natural language processing (almost) from scratch
  publication-title: J Mach Learning Res
– ident: bib39
  article-title: Exploring the ChestXray14 dataset: problems
– volume: 518
  start-page: 529
  year: 2015
  end-page: 533
  ident: bib3
  article-title: Human-level control through deep reinforcement learning
  publication-title: Nature
– year: April 2, 2019
  ident: bib13
  article-title: 2019 Global AI talent report
  publication-title: ElementAI
– volume: 2
  start-page: 2
  year: 2019
  ident: bib36
  article-title: The reproducibility crisis in the age of digital medicine
  publication-title: Digit Med
– year: 2019
  ident: bib37
  article-title: What clinicians want: contextualizing explainable machine learning for clinical end use
  publication-title: ArXiv
– volume: 25
  start-page: 65
  year: 2019
  end-page: 69
  ident: bib12
  article-title: Cardiologist-level arrhythmia detection and classification in ambulatory electrocardiograms using a deep neural network
  publication-title: Nat Med
– volume: 350
  year: 2015
  ident: bib23
  article-title: Transparent reporting of a multivariable prediction model for individual prognosis or diagnosis (TRIPOD): the TRIPOD statement
  publication-title: BMJ
– ident: bib40
  article-title: Open source research data repository software
– year: 2018
  ident: bib28
  article-title: Diagnose like a radiologist: attention guided convolutional neural network for thorax disease classification
  publication-title: arXiv
– volume: 33
  start-page: 231
  year: 2014
  end-page: 234
  ident: bib19
  article-title: Feedback on a publicly distributed image database: the Messidor database
  publication-title: Image Anal Stereol
– year: Nov 2, 2017
  ident: bib16
  article-title: AutoML for large scale image classification and object detection
  publication-title: Google AI Blog
– volume: 20
  start-page: 1194
  year: 2013
  end-page: 1206
  ident: bib32
  article-title: Understanding the direction of bias in studies of diagnostic test accuracy
  publication-title: Acad Emerge Med
– year: April 15, 2018
  ident: bib14
  article-title: Why AutoML is set to become the future of artificial intelligence
  publication-title: Forbes
– year: 2012
  ident: bib27
  article-title: ImageNet classification with deep convolutional neural networks
  publication-title: Advances in Neural Information Processing Systems
– reference: Hadsell R, Erkan A, Sermanet P, Scoffier M, Muller U, LeCun Y. Deep belief net learning in a long-range vision system for autonomous off-road driving. 2008 IEEE/RSJ International Conference on Intelligent Robot Systems, IROS; Nice, France; Sept 22–26, 2018 (4651217).
– ident: bib21
  article-title: Chestx-ray8: hospital-scale chest x-ray database and benchmarks on weakly-supervised classification and localization of common thorax diseases
– volume: 24
  start-page: 1342
  year: 2018
  ident: bib34
  article-title: Clinically applicable deep learning for diagnosis and referral in retinal disease
  publication-title: Nat Med
– volume: 378
  start-page: 981
  year: 2018
  end-page: 983
  ident: bib43
  article-title: Implementing machine learning in health care–addressing ethical challenges
  publication-title: N Engl J Med
– year: Jan 1, 2019
  ident: bib15
  article-title: Auto machine learning software/tools in 2019: in-depth guide
  publication-title: AI Multiple
– volume: 172
  start-page: 1122
  year: 2018
  end-page: 1131
  ident: bib20
  article-title: Identifying medical diagnoses and treatable diseases by image-based deep learning
  publication-title: Cell
– ident: bib29
  article-title: Convolutional neural networks based transfer learning for diabetic retinopathy fundus image classification. 2017 10th International Congress on Image and Signal Processing, BioMedical Engineering and Informatics; Shanghai, China; Oct 14–16, 2017 (abstract)
– volume: 45
  start-page: 121
  year: 2018
  end-page: 133
  ident: bib41
  article-title: DeepMitosis: Mitosis detection via deep detection, verification and segmentation networks
  publication-title: Med Image Anal
– year: 2016
  ident: bib18
  article-title: Neural architecture search with reinforcement learning
  publication-title: arXiv
– year: May 17, 2017
  ident: bib26
  article-title: Using Machine Learning to Explore Neural Network Architecture
  publication-title: Google AI Blog
– volume: 18
  start-page: e323
  year: 2016
  ident: bib31
  article-title: Guidelines for developing and reporting machine learning predictive models in biomedical research: a multidisciplinary view
  publication-title: JIMR
– ident: bib17
  article-title: fast.ai, making neural nets uncool again
– year: May 18, 2016
  ident: bib11
  article-title: Google supercharges machine learning tasks with TPU custom chip
  publication-title: Google Cloud
– year: 2016
  ident: bib38
  article-title: The Mythos of model interpretability
  publication-title: ArXiv
– volume: 29
  start-page: 82
  year: 2012
  end-page: 97
  ident: bib5
  article-title: Deep neural networks for acoustic modeling in speech recognition
  publication-title: IEEE Signal Process Mag
– volume: 521
  start-page: 436
  year: 2015
  end-page: 444
  ident: bib1
  article-title: Deep learning
  publication-title: Nature
– volume: 26
  start-page: 120
  year: 2009
  end-page: 144
  ident: bib8
  article-title: Learning long-range vision for autonomous off-road driving
  publication-title: J Field Rob
– year: Feb 19, 2018
  ident: bib24
  article-title: What covered entities should know about cloud computing and HIPAA compliance
  publication-title: HIPAA J
– reference: Codella NC, Gutman D, Celebi ME, et al. Skin lesion analysis toward melanoma detection: a challenge at the 2017 international symposium on biomedical imaging (ISBI), hosted by the international skin imaging collaboration (ISIC). 2018 IEEE 15th International Symposium on Biomedical Imaging; Washington DC, USA; April 4–7, 2018: 168–72.
– volume: 5
  start-page: 142
  year: 2005
  end-page: 149
  ident: bib42
  article-title: Bias as a threat to the validity of cancer molecular-marker research
  publication-title: Nat Rev Cancer
– start-page: 116
  year: 2017
  ident: bib35
  article-title: Regulating Black-Box Medicine
  publication-title: Mich L Rev
– year: 2018
  ident: bib25
  article-title: Neural architecture search: a survey
  publication-title: arXiv
– volume: 1
  start-page: 1097
  year: 2012
  end-page: 1105
  ident: bib9
  article-title: ImageNet classification with deep convolutional neural networks
  publication-title: Proceedings of the 25th International Conference on Neural Information Processing Systems
– volume: 5
  year: 2018
  ident: bib22
  article-title: The HAM10000 dataset: a large collection of multi-source dermatoscopic images of common pigmented skin lesions
  publication-title: Sci Data
– year: 2016
  ident: 10.1016/S2589-7500(19)30108-6_bib38
  article-title: The Mythos of model interpretability
  publication-title: ArXiv
– volume: 1
  start-page: 1097
  year: 2012
  ident: 10.1016/S2589-7500(19)30108-6_bib9
  article-title: ImageNet classification with deep convolutional neural networks
  publication-title: Proceedings of the 25th International Conference on Neural Information Processing Systems
– start-page: 116
  year: 2017
  ident: 10.1016/S2589-7500(19)30108-6_bib35
  article-title: Regulating Black-Box Medicine
  publication-title: Mich L Rev
– volume: 18
  start-page: e323
  year: 2016
  ident: 10.1016/S2589-7500(19)30108-6_bib31
  article-title: Guidelines for developing and reporting machine learning predictive models in biomedical research: a multidisciplinary view
  publication-title: JIMR
– volume: 521
  start-page: 436
  year: 2015
  ident: 10.1016/S2589-7500(19)30108-6_bib1
  article-title: Deep learning
  publication-title: Nature
  doi: 10.1038/nature14539
– volume: 1
  start-page: 5
  year: 2018
  ident: 10.1016/S2589-7500(19)30108-6_bib2
  article-title: Artificial intelligence powers digital medicine
  publication-title: Dig Med
– ident: 10.1016/S2589-7500(19)30108-6_bib30
– volume: 26
  start-page: 120
  year: 2009
  ident: 10.1016/S2589-7500(19)30108-6_bib8
  article-title: Learning long-range vision for autonomous off-road driving
  publication-title: J Field Rob
  doi: 10.1002/rob.20276
– volume: 12
  start-page: 2493
  year: 2011
  ident: 10.1016/S2589-7500(19)30108-6_bib6
  article-title: Natural language processing (almost) from scratch
  publication-title: J Mach Learning Res
– volume: 172
  start-page: 1122
  year: 2018
  ident: 10.1016/S2589-7500(19)30108-6_bib20
  article-title: Identifying medical diagnoses and treatable diseases by image-based deep learning
  publication-title: Cell
  doi: 10.1016/j.cell.2018.02.010
– year: 2018
  ident: 10.1016/S2589-7500(19)30108-6_bib25
  article-title: Neural architecture search: a survey
  publication-title: arXiv
– ident: 10.1016/S2589-7500(19)30108-6_bib21
– year: 2019
  ident: 10.1016/S2589-7500(19)30108-6_bib37
  article-title: What clinicians want: contextualizing explainable machine learning for clinical end use
  publication-title: ArXiv
– ident: 10.1016/S2589-7500(19)30108-6_bib7
  doi: 10.1109/IROS.2008.4651217
– year: 2019
  ident: 10.1016/S2589-7500(19)30108-6_bib15
  article-title: Auto machine learning software/tools in 2019: in-depth guide
  publication-title: AI Multiple
– volume: 33
  start-page: 231
  year: 2014
  ident: 10.1016/S2589-7500(19)30108-6_bib19
  article-title: Feedback on a publicly distributed image database: the Messidor database
  publication-title: Image Anal Stereol
  doi: 10.5566/ias.1155
– year: 2018
  ident: 10.1016/S2589-7500(19)30108-6_bib28
  article-title: Diagnose like a radiologist: attention guided convolutional neural network for thorax disease classification
  publication-title: arXiv
– volume: 29
  start-page: 82
  year: 2012
  ident: 10.1016/S2589-7500(19)30108-6_bib5
  article-title: Deep neural networks for acoustic modeling in speech recognition
  publication-title: IEEE Signal Process Mag
  doi: 10.1109/MSP.2012.2205597
– year: 2019
  ident: 10.1016/S2589-7500(19)30108-6_bib13
  article-title: 2019 Global AI talent report
  publication-title: ElementAI
– volume: 5
  start-page: 142
  year: 2005
  ident: 10.1016/S2589-7500(19)30108-6_bib42
  article-title: Bias as a threat to the validity of cancer molecular-marker research
  publication-title: Nat Rev Cancer
  doi: 10.1038/nrc1550
– year: 2014
  ident: 10.1016/S2589-7500(19)30108-6_bib4
  article-title: Show and tell: a neural image caption generator
  publication-title: arXiv
– volume: 378
  start-page: 981
  year: 2018
  ident: 10.1016/S2589-7500(19)30108-6_bib43
  article-title: Implementing machine learning in health care–addressing ethical challenges
  publication-title: N Engl J Med
  doi: 10.1056/NEJMp1714229
– year: 2016
  ident: 10.1016/S2589-7500(19)30108-6_bib11
  article-title: Google supercharges machine learning tasks with TPU custom chip
  publication-title: Google Cloud
– ident: 10.1016/S2589-7500(19)30108-6_bib17
– ident: 10.1016/S2589-7500(19)30108-6_bib39
– year: 2012
  ident: 10.1016/S2589-7500(19)30108-6_bib27
  article-title: ImageNet classification with deep convolutional neural networks
  publication-title: Advances in Neural Information Processing Systems
– volume: 24
  start-page: 1342
  year: 2018
  ident: 10.1016/S2589-7500(19)30108-6_bib34
  article-title: Clinically applicable deep learning for diagnosis and referral in retinal disease
  publication-title: Nat Med
  doi: 10.1038/s41591-018-0107-6
– year: 2016
  ident: 10.1016/S2589-7500(19)30108-6_bib18
  article-title: Neural architecture search with reinforcement learning
  publication-title: arXiv
– year: 2017
  ident: 10.1016/S2589-7500(19)30108-6_bib26
  article-title: Using Machine Learning to Explore Neural Network Architecture
  publication-title: Google AI Blog
– year: 2018
  ident: 10.1016/S2589-7500(19)30108-6_bib14
  article-title: Why AutoML is set to become the future of artificial intelligence
  publication-title: Forbes
– volume: 45
  start-page: 121
  year: 2018
  ident: 10.1016/S2589-7500(19)30108-6_bib41
  article-title: DeepMitosis: Mitosis detection via deep detection, verification and segmentation networks
  publication-title: Med Image Anal
  doi: 10.1016/j.media.2017.12.002
– year: 2015
  ident: 10.1016/S2589-7500(19)30108-6_bib10
  article-title: World changing ideas of 2015
  publication-title: Scientific American
– year: 2018
  ident: 10.1016/S2589-7500(19)30108-6_bib24
  article-title: What covered entities should know about cloud computing and HIPAA compliance
  publication-title: HIPAA J
– volume: 20
  start-page: 1194
  year: 2013
  ident: 10.1016/S2589-7500(19)30108-6_bib32
  article-title: Understanding the direction of bias in studies of diagnostic test accuracy
  publication-title: Acad Emerge Med
  doi: 10.1111/acem.12255
– year: 2017
  ident: 10.1016/S2589-7500(19)30108-6_bib16
  article-title: AutoML for large scale image classification and object detection
  publication-title: Google AI Blog
– volume: 2
  start-page: 2
  year: 2019
  ident: 10.1016/S2589-7500(19)30108-6_bib36
  article-title: The reproducibility crisis in the age of digital medicine
  publication-title: Digit Med
  doi: 10.1038/s41746-019-0079-z
– volume: 518
  start-page: 529
  year: 2015
  ident: 10.1016/S2589-7500(19)30108-6_bib3
  article-title: Human-level control through deep reinforcement learning
  publication-title: Nature
  doi: 10.1038/nature14236
– volume: 5
  year: 2018
  ident: 10.1016/S2589-7500(19)30108-6_bib22
  article-title: The HAM10000 dataset: a large collection of multi-source dermatoscopic images of common pigmented skin lesions
  publication-title: Sci Data
  doi: 10.1038/sdata.2018.161
– ident: 10.1016/S2589-7500(19)30108-6_bib29
– ident: 10.1016/S2589-7500(19)30108-6_bib33
– volume: 350
  year: 2015
  ident: 10.1016/S2589-7500(19)30108-6_bib23
  article-title: Transparent reporting of a multivariable prediction model for individual prognosis or diagnosis (TRIPOD): the TRIPOD statement
  publication-title: BMJ
  doi: 10.1136/bmj.g7594
– volume: 25
  start-page: 65
  year: 2019
  ident: 10.1016/S2589-7500(19)30108-6_bib12
  article-title: Cardiologist-level arrhythmia detection and classification in ambulatory electrocardiograms using a deep neural network
  publication-title: Nat Med
  doi: 10.1038/s41591-018-0268-3
– reference: 33323266 - Lancet Digit Health. 2019 Sep;1(5):e198-e199. doi: 10.1016/S2589-7500(19)30112-8.
SSID ssj0002213777
Score 2.5566797
Snippet Deep learning has the potential to transform health care; however, substantial expertise is required to train such models. We sought to evaluate the utility of...
SummaryBackgroundDeep learning has the potential to transform health care; however, substantial expertise is required to train such models. We sought to...
Background: Deep learning has the potential to transform health care; however, substantial expertise is required to train such models. We sought to evaluate...
SourceID doaj
unpaywall
proquest
pubmed
crossref
elsevier
SourceType Open Website
Open Access Repository
Aggregation Database
Index Database
Enrichment Source
Publisher
StartPage e232
SubjectTerms Adult
Algorithms
Data Interpretation, Statistical
Deep Learning
Feasibility Studies
Fundus Oculi
Humans
Informatics
Internal Medicine
Public Health
Skin Neoplasms - diagnosis
Software
Tomography, Optical Coherence - statistics & numerical data
SummonAdditionalLinks – databaseName: Unpaywall
  dbid: UNPAY
  link: http://utb.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwlV1Lj9MwELagewAOPMQrvGQkDnBINw_HD24FsVohUXGg0nKybMdZregmFUmEyl_gTzOTF1u20sKpaeVJ4vF45nM985mQV0qqIubchFyqJEQjCY2UJoyll4USTtkum_DTkh-v2MeT7GQoVsdamJ39-5gf1kkmVQhhLXodqzdgjZEM-XVywDOA3jNysFp-XnzFA-TGZn-qdPbL7sSfjqZ_Jwxdhpm3yI223JjtD7NeXwg9R3fIcnzpPuPk27xt7Nz9_IvP8Z97dZfcHkAoXfRWc49c8-V98mvRNhXgV5_T3PsNHc6TOIVvmORBAd3S835bh56dgxuiDpE3php1o0vtlvZVlSGmk9HNBdKPmuL_vbSsqKswWFI_MSy_pYYW3gxZulva8d0-IKujD1_eH4fDUQ2hg0nchFKmzEmbFE4UAlZJConvBXc-TzxXxts8SllhnBMJklbZyFmVgzOBNrgCdelDMiur0j8mlOWFTAHmcW8Es7FReeycLQysmGEZL3xA2DiA2g085nicxlpPCWuoXY3a1bHSnXY1D8h8Etv0RB5XCbxD65gaIw939wMMpB6mtbaFyFmSC5s5x1LD8NMWGVxC36WTAeGjbemx1BWcM9zo7Kqni32Cvh5cTK1jXSc66qVRGJA4ioLky9GGNbgK3P8xpa_aWidMACCFdiIgj3rjnnqXgsrTRMQBOZys_ZKe6n1v-uS_JZ6SmwA7h0y9Z2TWfG_9c4B2jX0xTOjf2SNAVg
  priority: 102
  providerName: Unpaywall
Title Automated deep learning design for medical image classification by health-care professionals with no coding experience: a feasibility study
URI https://www.clinicalkey.com/#!/content/1-s2.0-S2589750019301086
https://www.clinicalkey.es/playcontent/1-s2.0-S2589750019301086
https://www.ncbi.nlm.nih.gov/pubmed/33323271
https://www.proquest.com/docview/2470631937
https://doi.org/10.1016/s2589-7500(19)30108-6
https://doaj.org/article/bf7d42d7b5cc43a4b5ccbf543a6ce8c8
UnpaywallVersion publishedVersion
Volume 1
hasFullText 1
inHoldings 1
isFullTextHit
isPrint
journalDatabaseRights – providerCode: PRVAON
  databaseName: DOAJ Directory of Open Access Journals (ODIN)
  customDbUrl:
  eissn: 2589-7500
  dateEnd: 99991231
  omitProxy: true
  ssIdentifier: ssj0002213777
  issn: 2589-7500
  databaseCode: DOA
  dateStart: 20190101
  isFulltext: true
  titleUrlDefault: https://www.doaj.org/
  providerName: Directory of Open Access Journals
– providerCode: PRVHPJ
  databaseName: ROAD: Directory of Open Access Scholarly Resources (ISSN International Center)
  customDbUrl:
  eissn: 2589-7500
  dateEnd: 99991231
  omitProxy: true
  ssIdentifier: ssj0002213777
  issn: 2589-7500
  databaseCode: M~E
  dateStart: 20190101
  isFulltext: true
  titleUrlDefault: https://road.issn.org
  providerName: ISSN International Centre
link http://utb.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwrV1Lb9NAEF6h9gAcKhAvA60WiQMclvix2UdvKWpVIbVCgkjltNqXUavUjnCiKhf-AH-aGdsxqVopFy5xYu0k9szszjfxzLeEvNdKl5kQlgmlc4ZOwqxSlmUqqlJLr11bTXh2Lk6n_MvF-GJjqy-sCevogTvFjVwpA8-DdGPveWE5Hl05hrfCR-XbNt9U6Y1k6qoldUEmPeyVzsdKMwiL6b_2ndG34eSHTH8EF08VE7cCU8vffys-3cWfj8nDZTW3qxs7m23EpJMnZK8Hk3TS3cRT8iBWz8ifyXJRAw6NgYYY57TfF-InfMJiDQoolV53j2fo5TUsJ9QjgsaSodZK1K1o1x3JsCyMzjfIOxqK_9vSqqa-xqBH48CUfEgtLaPtq21XtOWtfU6mJ8ffP5-yfssF5mEyLphSBffK5aWXpYRsRyOBvQRNhzwKbaMLacFL673MkXzKpd7pAIsCjMFM0hcvyE5VV_EVoTyUqgC4JqKV3GVWhwwtZyHzhXRcxoTwtb6N7_nIcVuMmRkKz9BMBs1kMm1aMxmRkE-D2Lwj5NgmcITGHAYjn3Z7ArzM9F5mtnlZQsTaFcy6ZRUWWfiiy22_Lu8TjE2_VDQmM01u0k4ahQFRoyhIvlu7nIEpj89xbBXrZWNyLgFYwjiZkJedLw53V4DKi1xmCRkNznlHT819V_r6f-jpDXkEiLIvwntLdha_lnEfUNvCHbQTFF7Pfh8fkN3p-dfJj789rTzT
linkProvider Directory of Open Access Journals
linkToUnpaywall http://utb.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwlV1Lj9MwELagewAOPMQrvGQkDnBINw_HD24FsVohUXGg0nKybMdZregmFUmEyl_gTzOTF1u20sKpaeVJ4vF45nM985mQV0qqIubchFyqJEQjCY2UJoyll4USTtkum_DTkh-v2MeT7GQoVsdamJ39-5gf1kkmVQhhLXodqzdgjZEM-XVywDOA3jNysFp-XnzFA-TGZn-qdPbL7sSfjqZ_Jwxdhpm3yI223JjtD7NeXwg9R3fIcnzpPuPk27xt7Nz9_IvP8Z97dZfcHkAoXfRWc49c8-V98mvRNhXgV5_T3PsNHc6TOIVvmORBAd3S835bh56dgxuiDpE3php1o0vtlvZVlSGmk9HNBdKPmuL_vbSsqKswWFI_MSy_pYYW3gxZulva8d0-IKujD1_eH4fDUQ2hg0nchFKmzEmbFE4UAlZJConvBXc-TzxXxts8SllhnBMJklbZyFmVgzOBNrgCdelDMiur0j8mlOWFTAHmcW8Es7FReeycLQysmGEZL3xA2DiA2g085nicxlpPCWuoXY3a1bHSnXY1D8h8Etv0RB5XCbxD65gaIw939wMMpB6mtbaFyFmSC5s5x1LD8NMWGVxC36WTAeGjbemx1BWcM9zo7Kqni32Cvh5cTK1jXSc66qVRGJA4ioLky9GGNbgK3P8xpa_aWidMACCFdiIgj3rjnnqXgsrTRMQBOZys_ZKe6n1v-uS_JZ6SmwA7h0y9Z2TWfG_9c4B2jX0xTOjf2SNAVg
openUrl ctx_ver=Z39.88-2004&ctx_enc=info%3Aofi%2Fenc%3AUTF-8&rfr_id=info%3Asid%2Fsummon.serialssolutions.com&rft_val_fmt=info%3Aofi%2Ffmt%3Akev%3Amtx%3Ajournal&rft.genre=article&rft.atitle=Automated+deep+learning+design+for+medical+image+classification+by+health-care+professionals+with+no+coding+experience%3A+a+feasibility+study&rft.jtitle=The+Lancet.+Digital+health&rft.au=Faes%2C+Livia%2C+MD&rft.au=Wagner%2C+Siegfried+K%2C+BMBCh&rft.au=Fu%2C+Dun+Jack%2C+PhD&rft.au=Liu%2C+Xiaoxuan%2C+MBChB&rft.date=2019-09-01&rft.issn=2589-7500&rft.volume=1&rft.issue=5&rft.spage=e232&rft.epage=e242&rft_id=info:doi/10.1016%2FS2589-7500%2819%2930108-6&rft.externalDBID=ECK1-s2.0-S2589750019301086&rft.externalDocID=1_s2_0_S2589750019301086
thumbnail_m http://utb.summon.serialssolutions.com/2.0.0/image/custom?url=https%3A%2F%2Fcdn.clinicalkey.com%2Fck-thumbnails%2F25897500%2FS2589750019X00060%2Fcov150h.gif