Comparison of machine learning clustering algorithms for detecting heterogeneity of treatment effect in acute respiratory distress syndrome: A secondary analysis of three randomised controlled trials

Heterogeneity in Acute Respiratory Distress Syndrome (ARDS), as a consequence of its non-specific definition, has led to a multitude of negative randomised controlled trials (RCTs). Investigators have sought to identify heterogeneity of treatment effect (HTE) in RCTs using clustering algorithms. We...

Full description

Saved in:
Bibliographic Details
Published in:EBioMedicine Vol. 74; p. 103697
Main Authors: Sinha, Pratik, Spicer, Alexandra, Delucchi, Kevin L, McAuley, Daniel F, Calfee, Carolyn S, Churpek, Matthew M
Format: Journal Article
Language:English
Published: Netherlands Elsevier B.V 01.12.2021
Elsevier
Subjects:
ISSN:2352-3964, 2352-3964
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Abstract Heterogeneity in Acute Respiratory Distress Syndrome (ARDS), as a consequence of its non-specific definition, has led to a multitude of negative randomised controlled trials (RCTs). Investigators have sought to identify heterogeneity of treatment effect (HTE) in RCTs using clustering algorithms. We evaluated the proficiency of several commonly-used machine-learning algorithms to identify clusters where HTE may be detected. Five unsupervised: Latent class analysis (LCA), K-means, partition around medoids, hierarchical, and spectral clustering; and four supervised algorithms: model-based recursive partitioning, Causal Forest (CF), and X-learner with Random Forest (XL-RF) and Bayesian Additive Regression Trees were individually applied to three prior ARDS RCTs. Clinical data and research protein biomarkers were used as partitioning variables, with the latter excluded for secondary analyses. For a clustering schema, HTE was evaluated based on the interaction term of treatment group and cluster with day-90 mortality as the dependent variable. No single algorithm identified clusters with significant HTE in all three trials. LCA, XL-RF, and CF identified HTE most frequently (2/3 RCTs). Important partitioning variables in the unsupervised approaches were consistent across algorithms and RCTs. In supervised models, important partitioning variables varied between algorithms and across RCTs. In algorithms where clusters demonstrated HTE in the same trial, patients frequently interchanged clusters from treatment-benefit to treatment-harm clusters across algorithms. LCA aside, results from all other algorithms were subject to significant alteration in cluster composition and HTE with random seed change. Removing research biomarkers as partitioning variables greatly reduced the chances of detecting HTE across all algorithms. Machine-learning algorithms were inconsistent in their abilities to identify clusters with significant HTE. Protein biomarkers were essential in identifying clusters with HTE. Investigations using machine-learning approaches to identify clusters to seek HTE require cautious interpretation. NIGMS R35 GM142992 (PS), NHLBI R35 HL140026 (CSC); NIGMS R01 GM123193, Department of Defense W81XWH-21-1-0009, NIA R21 AG068720, NIDA R01 DA051464 (MMC)
AbstractList Background: Heterogeneity in Acute Respiratory Distress Syndrome (ARDS), as a consequence of its non-specific definition, has led to a multitude of negative randomised controlled trials (RCTs). Investigators have sought to identify heterogeneity of treatment effect (HTE) in RCTs using clustering algorithms. We evaluated the proficiency of several commonly-used machine-learning algorithms to identify clusters where HTE may be detected. Methods: Five unsupervised: Latent class analysis (LCA), K-means, partition around medoids, hierarchical, and spectral clustering; and four supervised algorithms: model-based recursive partitioning, Causal Forest (CF), and X-learner with Random Forest (XL-RF) and Bayesian Additive Regression Trees were individually applied to three prior ARDS RCTs. Clinical data and research protein biomarkers were used as partitioning variables, with the latter excluded for secondary analyses. For a clustering schema, HTE was evaluated based on the interaction term of treatment group and cluster with day-90 mortality as the dependent variable. Findings: No single algorithm identified clusters with significant HTE in all three trials. LCA, XL-RF, and CF identified HTE most frequently (2/3 RCTs). Important partitioning variables in the unsupervised approaches were consistent across algorithms and RCTs. In supervised models, important partitioning variables varied between algorithms and across RCTs. In algorithms where clusters demonstrated HTE in the same trial, patients frequently interchanged clusters from treatment-benefit to treatment-harm clusters across algorithms. LCA aside, results from all other algorithms were subject to significant alteration in cluster composition and HTE with random seed change. Removing research biomarkers as partitioning variables greatly reduced the chances of detecting HTE across all algorithms. Interpretation: Machine-learning algorithms were inconsistent in their abilities to identify clusters with significant HTE. Protein biomarkers were essential in identifying clusters with HTE. Investigations using machine-learning approaches to identify clusters to seek HTE require cautious interpretation. Funding: NIGMS R35 GM142992 (PS), NHLBI R35 HL140026 (CSC); NIGMS R01 GM123193, Department of Defense W81XWH-21-1-0009, NIA R21 AG068720, NIDA R01 DA051464 (MMC)
Heterogeneity in Acute Respiratory Distress Syndrome (ARDS), as a consequence of its non-specific definition, has led to a multitude of negative randomised controlled trials (RCTs). Investigators have sought to identify heterogeneity of treatment effect (HTE) in RCTs using clustering algorithms. We evaluated the proficiency of several commonly-used machine-learning algorithms to identify clusters where HTE may be detected. Five unsupervised: Latent class analysis (LCA), K-means, partition around medoids, hierarchical, and spectral clustering; and four supervised algorithms: model-based recursive partitioning, Causal Forest (CF), and X-learner with Random Forest (XL-RF) and Bayesian Additive Regression Trees were individually applied to three prior ARDS RCTs. Clinical data and research protein biomarkers were used as partitioning variables, with the latter excluded for secondary analyses. For a clustering schema, HTE was evaluated based on the interaction term of treatment group and cluster with day-90 mortality as the dependent variable. No single algorithm identified clusters with significant HTE in all three trials. LCA, XL-RF, and CF identified HTE most frequently (2/3 RCTs). Important partitioning variables in the unsupervised approaches were consistent across algorithms and RCTs. In supervised models, important partitioning variables varied between algorithms and across RCTs. In algorithms where clusters demonstrated HTE in the same trial, patients frequently interchanged clusters from treatment-benefit to treatment-harm clusters across algorithms. LCA aside, results from all other algorithms were subject to significant alteration in cluster composition and HTE with random seed change. Removing research biomarkers as partitioning variables greatly reduced the chances of detecting HTE across all algorithms. Machine-learning algorithms were inconsistent in their abilities to identify clusters with significant HTE. Protein biomarkers were essential in identifying clusters with HTE. Investigations using machine-learning approaches to identify clusters to seek HTE require cautious interpretation. NIGMS R35 GM142992 (PS), NHLBI R35 HL140026 (CSC); NIGMS R01 GM123193, Department of Defense W81XWH-21-1-0009, NIA R21 AG068720, NIDA R01 DA051464 (MMC).
AbstractBackgroundHeterogeneity in Acute Respiratory Distress Syndrome (ARDS), as a consequence of its non-specific definition, has led to a multitude of negative randomised controlled trials (RCTs). Investigators have sought to identify heterogeneity of treatment effect (HTE) in RCTs using clustering algorithms. We evaluated the proficiency of several commonly-used machine-learning algorithms to identify clusters where HTE may be detected. MethodsFive unsupervised: Latent class analysis (LCA), K-means, partition around medoids, hierarchical, and spectral clustering; and four supervised algorithms: model-based recursive partitioning, Causal Forest (CF), and X-learner with Random Forest (XL-RF) and Bayesian Additive Regression Trees were individually applied to three prior ARDS RCTs. Clinical data and research protein biomarkers were used as partitioning variables, with the latter excluded for secondary analyses. For a clustering schema, HTE was evaluated based on the interaction term of treatment group and cluster with day-90 mortality as the dependent variable. FindingsNo single algorithm identified clusters with significant HTE in all three trials. LCA, XL-RF, and CF identified HTE most frequently (2/3 RCTs). Important partitioning variables in the unsupervised approaches were consistent across algorithms and RCTs. In supervised models, important partitioning variables varied between algorithms and across RCTs. In algorithms where clusters demonstrated HTE in the same trial, patients frequently interchanged clusters from treatment-benefit to treatment-harm clusters across algorithms. LCA aside, results from all other algorithms were subject to significant alteration in cluster composition and HTE with random seed change. Removing research biomarkers as partitioning variables greatly reduced the chances of detecting HTE across all algorithms. InterpretationMachine-learning algorithms were inconsistent in their abilities to identify clusters with significant HTE. Protein biomarkers were essential in identifying clusters with HTE. Investigations using machine-learning approaches to identify clusters to seek HTE require cautious interpretation. FundingNIGMS R35 GM142992 (PS), NHLBI R35 HL140026 (CSC); NIGMS R01 GM123193, Department of Defense W81XWH-21-1-0009, NIA R21 AG068720, NIDA R01 DA051464 (MMC)
Heterogeneity in Acute Respiratory Distress Syndrome (ARDS), as a consequence of its non-specific definition, has led to a multitude of negative randomised controlled trials (RCTs). Investigators have sought to identify heterogeneity of treatment effect (HTE) in RCTs using clustering algorithms. We evaluated the proficiency of several commonly-used machine-learning algorithms to identify clusters where HTE may be detected.BACKGROUNDHeterogeneity in Acute Respiratory Distress Syndrome (ARDS), as a consequence of its non-specific definition, has led to a multitude of negative randomised controlled trials (RCTs). Investigators have sought to identify heterogeneity of treatment effect (HTE) in RCTs using clustering algorithms. We evaluated the proficiency of several commonly-used machine-learning algorithms to identify clusters where HTE may be detected.Five unsupervised: Latent class analysis (LCA), K-means, partition around medoids, hierarchical, and spectral clustering; and four supervised algorithms: model-based recursive partitioning, Causal Forest (CF), and X-learner with Random Forest (XL-RF) and Bayesian Additive Regression Trees were individually applied to three prior ARDS RCTs. Clinical data and research protein biomarkers were used as partitioning variables, with the latter excluded for secondary analyses. For a clustering schema, HTE was evaluated based on the interaction term of treatment group and cluster with day-90 mortality as the dependent variable.METHODSFive unsupervised: Latent class analysis (LCA), K-means, partition around medoids, hierarchical, and spectral clustering; and four supervised algorithms: model-based recursive partitioning, Causal Forest (CF), and X-learner with Random Forest (XL-RF) and Bayesian Additive Regression Trees were individually applied to three prior ARDS RCTs. Clinical data and research protein biomarkers were used as partitioning variables, with the latter excluded for secondary analyses. For a clustering schema, HTE was evaluated based on the interaction term of treatment group and cluster with day-90 mortality as the dependent variable.No single algorithm identified clusters with significant HTE in all three trials. LCA, XL-RF, and CF identified HTE most frequently (2/3 RCTs). Important partitioning variables in the unsupervised approaches were consistent across algorithms and RCTs. In supervised models, important partitioning variables varied between algorithms and across RCTs. In algorithms where clusters demonstrated HTE in the same trial, patients frequently interchanged clusters from treatment-benefit to treatment-harm clusters across algorithms. LCA aside, results from all other algorithms were subject to significant alteration in cluster composition and HTE with random seed change. Removing research biomarkers as partitioning variables greatly reduced the chances of detecting HTE across all algorithms.FINDINGSNo single algorithm identified clusters with significant HTE in all three trials. LCA, XL-RF, and CF identified HTE most frequently (2/3 RCTs). Important partitioning variables in the unsupervised approaches were consistent across algorithms and RCTs. In supervised models, important partitioning variables varied between algorithms and across RCTs. In algorithms where clusters demonstrated HTE in the same trial, patients frequently interchanged clusters from treatment-benefit to treatment-harm clusters across algorithms. LCA aside, results from all other algorithms were subject to significant alteration in cluster composition and HTE with random seed change. Removing research biomarkers as partitioning variables greatly reduced the chances of detecting HTE across all algorithms.Machine-learning algorithms were inconsistent in their abilities to identify clusters with significant HTE. Protein biomarkers were essential in identifying clusters with HTE. Investigations using machine-learning approaches to identify clusters to seek HTE require cautious interpretation.INTERPRETATIONMachine-learning algorithms were inconsistent in their abilities to identify clusters with significant HTE. Protein biomarkers were essential in identifying clusters with HTE. Investigations using machine-learning approaches to identify clusters to seek HTE require cautious interpretation.NIGMS R35 GM142992 (PS), NHLBI R35 HL140026 (CSC); NIGMS R01 GM123193, Department of Defense W81XWH-21-1-0009, NIA R21 AG068720, NIDA R01 DA051464 (MMC).FUNDINGNIGMS R35 GM142992 (PS), NHLBI R35 HL140026 (CSC); NIGMS R01 GM123193, Department of Defense W81XWH-21-1-0009, NIA R21 AG068720, NIDA R01 DA051464 (MMC).
ArticleNumber 103697
Author McAuley, Daniel F
Delucchi, Kevin L
Spicer, Alexandra
Calfee, Carolyn S
Churpek, Matthew M
Sinha, Pratik
Author_xml – sequence: 1
  givenname: Pratik
  orcidid: 0000-0003-3751-9079
  surname: Sinha
  fullname: Sinha, Pratik
  email: p.sinha@wustl.edu
  organization: Division of Clinical and Translational Research, Division of Critical Care, Department of Anesthesia, Washington University School of Medicine, Saint Louis, MO
– sequence: 2
  givenname: Alexandra
  surname: Spicer
  fullname: Spicer, Alexandra
  organization: Department of Medicine, University of Wisconsin- Madison, Madison, Wisconsin
– sequence: 3
  givenname: Kevin L
  surname: Delucchi
  fullname: Delucchi, Kevin L
  organization: Department of Psychiatry and Behavioral Sciences; University of California, San Francisco; San Francisco, CA
– sequence: 4
  givenname: Daniel F
  surname: McAuley
  fullname: McAuley, Daniel F
  organization: Wellcome-Wolfson Institute for Experimental Medicine, Queen's University Belfast
– sequence: 5
  givenname: Carolyn S
  surname: Calfee
  fullname: Calfee, Carolyn S
  organization: Department of Medicine, Division of Pulmonary, Critical Care, Allergy and Sleep Medicine; University of California, San Francisco; San Francisco, CA
– sequence: 6
  givenname: Matthew M
  surname: Churpek
  fullname: Churpek, Matthew M
  organization: Department of Medicine, University of Wisconsin- Madison, Madison, Wisconsin
BackLink https://www.ncbi.nlm.nih.gov/pubmed/34861492$$D View this record in MEDLINE/PubMed
BookMark eNqFU9tuEzEQXaEiWkK_AAn5kZcUXzbOLohKVcSlUiUegGdrYs8mDl472N5K-UJ-C29SqrYSivzg2ZlzzqzHxy-rEx88VtVrRi8YZfLd5gKXNvQXnHJWMkK282fVGRczPhWtrE8exKfVeUobSimb1SXZvKhORd1IVrf8rPqzCP0Wok3Bk9CRHvTaeiQOIXrrV0S7IWWMYwhuFaLN6z6RLkRiMKPOY2FdohhW6NHm3aiSI0Lu0WeCXVdAxHoCeshIIqatjZBD3BFjUwGmRNLOmxh6fE-uSEIdvIFSBg9ul2zaC64jFjJ4E3qb0JACyjE4V8IcLbj0qnrelQ3P7_ZJ9fPzpx-Lr9Obb1-uF1c3Uy0Fz9O5lGYmO8Y6IdBgLWuuhYBGdjMjcQ7QoQDGmWg1b8s3mxswQjS6E4AIQkyq64OuCbBR22j78q8qgFX7RIgrBTFb7VAZbajmsma6MbWg2CyXWrZohEGo62VbtC4PWtth2aPRZWAR3CPRxxVv12oVblUj61lZReDtnUAMvwdMWZXpaHQOPIYhKS6pbDkX5TyT6s3DXvdN_jmhAMQBoGNIKWJ3D2FUjZZTG7W3nBotpw6WK6z2CUvbDNmO1wPWHeF-PHCx3NetxaiStug1GhuLacpA7RH-5RO-dtZbDe4X7jBtwhCLhZJiKnFF1ffxQYzvgTNK65bJIvDh_wJH2_8FroQj1w
CitedBy_id crossref_primary_10_1080_17476348_2023_2176302
crossref_primary_10_3310_TTND8896
crossref_primary_10_1186_s13054_022_04036_7
crossref_primary_10_1016_j_cct_2024_107778
crossref_primary_10_1007_s15010_025_02607_8
crossref_primary_10_1088_1755_1315_1142_1_012033
crossref_primary_10_3389_fimmu_2023_1209959
crossref_primary_10_1016_j_jclinepi_2024_111538
crossref_primary_10_1001_jamanetworkopen_2025_22390
crossref_primary_10_1016_j_chstcc_2024_100062
crossref_primary_10_1038_s43856_023_00423_5
crossref_primary_10_1097_SHK_0000000000002421
crossref_primary_10_1186_s13054_022_04265_w
crossref_primary_10_1016_j_ccc_2023_03_002
crossref_primary_10_1016_S2213_2600_23_00237_0
crossref_primary_10_1186_s13613_023_01143_8
crossref_primary_10_1513_AnnalsATS_202211_946OC
crossref_primary_10_1097_CCE_0000000000001312
crossref_primary_10_1146_annurev_med_043021_014005
crossref_primary_10_3389_fphys_2025_1649114
crossref_primary_10_1016_j_chstcc_2023_100018
crossref_primary_10_1097_CCM_0000000000006004
crossref_primary_10_3390_jcm12144650
crossref_primary_10_1186_s12931_024_02834_x
crossref_primary_10_1097_CCM_0000000000006362
crossref_primary_10_1097_MCC_0000000000001280
crossref_primary_10_1016_j_ebiom_2021_103770
crossref_primary_10_2147_IJGM_S391969
crossref_primary_10_1016_j_ccm_2024_08_005
crossref_primary_10_1136_thorax_2024_221989
crossref_primary_10_1016_j_jcrc_2024_154792
crossref_primary_10_1007_s00134_023_07239_w
crossref_primary_10_1016_j_ebiom_2021_103809
crossref_primary_10_1001_jama_2024_2933
crossref_primary_10_1007_s10916_024_02085_9
crossref_primary_10_1164_rccm_202209_1799OC
crossref_primary_10_1097_CCM_0000000000006371
Cites_doi 10.1016/S2213-2600(14)70097-9
10.1007/s00134-018-5378-3
10.1080/01621459.2017.1319839
10.1001/jamainternmed.2016.9125
10.7326/M18-3667
10.1136/bmj.h5651
10.3389/fnins.2016.00267
10.1056/NEJMoa032193
10.1056/NEJMoa062200
10.1164/rccm.201603-0645OC
10.1056/NEJMoa1401520
10.1136/bmj.k4245
10.1016/j.molmed.2014.01.007
10.1016/S2213-2600(18)30177-2
10.1186/1745-6215-10-48
10.1001/jama.2019.5791
10.1038/s41572-019-0069-0
10.1093/bib/bbn058
10.1016/j.chest.2018.04.037
10.1001/jama.298.10.1209
10.1093/bib/bbv018
10.1097/CCM.0000000000004710
10.1073/pnas.1510489113
ContentType Journal Article
Copyright 2021 The Authors
The Authors
Copyright © 2021 The Authors. Published by Elsevier B.V. All rights reserved.
2021 The Authors 2021
Copyright_xml – notice: 2021 The Authors
– notice: The Authors
– notice: Copyright © 2021 The Authors. Published by Elsevier B.V. All rights reserved.
– notice: 2021 The Authors 2021
DBID 6I.
AAFTH
AAYXX
CITATION
NPM
7X8
5PM
DOA
DOI 10.1016/j.ebiom.2021.103697
DatabaseName ScienceDirect Open Access Titles
Elsevier:ScienceDirect:Open Access
CrossRef
PubMed
MEDLINE - Academic
PubMed Central (Full Participant titles)
DOAJ Open Access Full Text
DatabaseTitle CrossRef
PubMed
MEDLINE - Academic
DatabaseTitleList
PubMed



MEDLINE - Academic
Database_xml – sequence: 1
  dbid: DOA
  name: DOAJ Directory of Open Access Journals
  url: https://www.doaj.org/
  sourceTypes: Open Website
– sequence: 2
  dbid: NPM
  name: PubMed
  url: http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed
  sourceTypes: Index Database
– sequence: 3
  dbid: 7X8
  name: MEDLINE - Academic
  url: https://search.proquest.com/medline
  sourceTypes: Aggregation Database
DeliveryMethod fulltext_linktorsrc
Discipline Biology
EISSN 2352-3964
EndPage 103697
ExternalDocumentID oai_doaj_org_article_dcd0c2641c8d430e8bbc69ed3dea44b9
PMC8645454
34861492
10_1016_j_ebiom_2021_103697
S2352396421004916
1_s2_0_S2352396421004916
Genre Journal Article
GrantInformation NIGMS R35 GM142992 (PS), NHLBI R35 HL140026 (CSC); NIGMS R01 GM123193, Department of Defense W81XWH-21-1-0009, NIA R21 AG068720, NIDA R01 DA051464 (MMC).
GrantInformation_xml – fundername: NIDA NIH HHS
  grantid: R01 DA051464
– fundername: NHLBI NIH HHS
  grantid: R35 HL140026
– fundername: NIGMS NIH HHS
  grantid: R01 GM123193
– fundername: NIGMS NIH HHS
  grantid: R35 GM142992
– fundername: NIA NIH HHS
  grantid: R21 AG068720
GroupedDBID .1-
.FO
0R~
4.4
457
53G
5VS
AAEDT
AAEDW
AAIKJ
AALRI
AAMRU
AAXUO
AAYWO
ABMAC
ACGFS
ACVFH
ADBBV
ADCNI
ADEZE
ADRAZ
ADVLN
AEUPX
AEXQZ
AFPUW
AFRHN
AFTJW
AGHFR
AIGII
AITUG
AJUYK
AKBMS
AKRWK
AKYEP
ALMA_UNASSIGNED_HOLDINGS
AMRAJ
AOIJS
APXCP
BCNDV
EBS
EJD
FDB
GROUPED_DOAJ
HYE
HZ~
IPNFZ
KQ8
M41
M48
O9-
OK1
RIG
ROL
RPM
SSZ
Z5R
0SF
6I.
AACTN
AAFTH
AFCTW
NCXOZ
AAYXX
CITATION
NPM
7X8
5PM
ID FETCH-LOGICAL-c632t-766d56f11f33ede4642c33a86f5d6e7aafe3a12139c29e7a17dad338cf3aeea33
IEDL.DBID DOA
ISICitedReferencesCount 38
ISICitedReferencesURI http://www.webofscience.com/api/gateway?GWVersion=2&SrcApp=Summon&SrcAuth=ProQuest&DestLinkType=CitingArticles&DestApp=WOS_CPL&KeyUT=000730209800010&url=https%3A%2F%2Fcvtisr.summon.serialssolutions.com%2F%23%21%2Fsearch%3Fho%3Df%26include.ft.matches%3Dt%26l%3Dnull%26q%3D
ISSN 2352-3964
IngestDate Fri Oct 03 12:48:22 EDT 2025
Tue Nov 04 02:01:59 EST 2025
Wed Oct 01 11:09:12 EDT 2025
Thu Jan 02 22:55:33 EST 2025
Thu Nov 13 04:35:03 EST 2025
Tue Nov 18 22:30:40 EST 2025
Tue Jul 25 20:58:01 EDT 2023
Sun Feb 23 10:19:25 EST 2025
Tue Aug 26 16:33:52 EDT 2025
IsDoiOpenAccess true
IsOpenAccess true
IsPeerReviewed true
IsScholarly true
Keywords LCA
Clustering
machine learning
RCTs
ARDS
Heterogeneity of treatment effect
Language English
License This is an open access article under the CC BY license.
Copyright © 2021 The Authors. Published by Elsevier B.V. All rights reserved.
This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
LinkModel DirectLink
MergedId FETCHMERGED-LOGICAL-c632t-766d56f11f33ede4642c33a86f5d6e7aafe3a12139c29e7a17dad338cf3aeea33
Notes ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 23
ObjectType-Undefined-3
ORCID 0000-0003-3751-9079
OpenAccessLink https://doaj.org/article/dcd0c2641c8d430e8bbc69ed3dea44b9
PMID 34861492
PQID 2606922321
PQPubID 23479
PageCount 1
ParticipantIDs doaj_primary_oai_doaj_org_article_dcd0c2641c8d430e8bbc69ed3dea44b9
pubmedcentral_primary_oai_pubmedcentral_nih_gov_8645454
proquest_miscellaneous_2606922321
pubmed_primary_34861492
crossref_primary_10_1016_j_ebiom_2021_103697
crossref_citationtrail_10_1016_j_ebiom_2021_103697
elsevier_sciencedirect_doi_10_1016_j_ebiom_2021_103697
elsevier_clinicalkeyesjournals_1_s2_0_S2352396421004916
elsevier_clinicalkey_doi_10_1016_j_ebiom_2021_103697
PublicationCentury 2000
PublicationDate 2021-12-01
PublicationDateYYYYMMDD 2021-12-01
PublicationDate_xml – month: 12
  year: 2021
  text: 2021-12-01
  day: 01
PublicationDecade 2020
PublicationPlace Netherlands
PublicationPlace_xml – name: Netherlands
PublicationTitle EBioMedicine
PublicationTitleAlternate EBioMedicine
PublicationYear 2021
Publisher Elsevier B.V
Elsevier
Publisher_xml – name: Elsevier B.V
– name: Elsevier
References Greene, Giffin, Greene, Moore (bib0021) 2016; 17
Wiedemann, Wheeler, Bernard, Thompson (bib0017) 2006; 354
Sinha, Delucchi, Thompson, McAuley, Matthay, Calfee (bib0012) 2018; 44
Athey, Imbens (bib0014) 2016; 113
Kent, Paulus, van Klaveren, D'Agostino, Goodman, Hayward (bib0024) 2020; 172
Calfee, Delucchi, Sinha, Matthay, Hackett, Shankar-Hari (bib0010) 2018; 6
Marshall (bib0003) 2014; 20
Calfee, Delucchi, Parsons, Thompson, Ware, Matthay (bib0009) 2014; 2
Andreopoulos, An, Wang, Schroeder (bib0019) 2009; 10
Varadhan, Seeger (bib0005) 2013
Wager, Athey (bib0015) 2018; 113
Burke, Sussman, Kent, Hayward (bib0006) 2015; 351
Kent, Hayward (bib0002) 2007; 298
Lipton ZC. The Doctor Just Won't Accept That!2017 November 01, 2017:[arXiv:1711.08037 p.]. Available from
Brower, Lanken, MacIntyre, Matthay, Morris, Ancukiewicz (bib0016) 2004; 351
Sanchez-Pinto, Luo, Churpek (bib0022) 2018; 154
Famous, Delucchi, Ware, Kangelaris, Liu, Thompson (bib0011) 2017; 195
Feng, Wallace, Grissom, Iyyer, Rodriguez, Boyd-Graber (bib0025) 2018
McCoy, Min, Linzen (bib0026) 2020
Kent, Kitsios (bib0001) 2009; 10
.
Wallach, Sullivan, Trepanowski, Sainani, Steyerberg, Ioannidis (bib0007) 2017; 177
Tonekaboni, Joshi, McCradden, Goldenberg (bib0029) 2019
Truwit, Bernard, Steingrub, Matthay (bib0018) 2014; 370
Kent, Steyerberg, van Klaveren (bib0008) 2018; 363
Lessov-Schlaggar, Rubin, Schlaggar (bib0023) 2016; 10
Lahav O, Mastronarde N, van der Schaar M. What is Interpretable? Using Machine Learning to Design Interpretable Decision-Support Systems2018 November 01, 2018:[arXiv:1811.10799 p.]. Available from
Matthay, Zemans, Zimmerman, Arabi, Beitler, Mercat (bib0004) 2019; 5
Seymour, Kennedy, Wang, Chang, Elliott, Xu (bib0013) 2019; 321
Sinha, Calfee, Delucchi (bib0020) 2021; 49
Kent (10.1016/j.ebiom.2021.103697_bib0001) 2009; 10
Famous (10.1016/j.ebiom.2021.103697_bib0011) 2017; 195
Kent (10.1016/j.ebiom.2021.103697_bib0008) 2018; 363
Calfee (10.1016/j.ebiom.2021.103697_bib0010) 2018; 6
Truwit (10.1016/j.ebiom.2021.103697_bib0018) 2014; 370
10.1016/j.ebiom.2021.103697_bib0027
10.1016/j.ebiom.2021.103697_bib0028
Kent (10.1016/j.ebiom.2021.103697_bib0024) 2020; 172
Matthay (10.1016/j.ebiom.2021.103697_bib0004) 2019; 5
Sanchez-Pinto (10.1016/j.ebiom.2021.103697_bib0022) 2018; 154
McCoy (10.1016/j.ebiom.2021.103697_bib0026) 2020
Calfee (10.1016/j.ebiom.2021.103697_bib0009) 2014; 2
Athey (10.1016/j.ebiom.2021.103697_bib0014) 2016; 113
Seymour (10.1016/j.ebiom.2021.103697_bib0013) 2019; 321
Brower (10.1016/j.ebiom.2021.103697_bib0016) 2004; 351
Andreopoulos (10.1016/j.ebiom.2021.103697_bib0019) 2009; 10
Varadhan (10.1016/j.ebiom.2021.103697_bib0005) 2013
Sinha (10.1016/j.ebiom.2021.103697_bib0012) 2018; 44
Burke (10.1016/j.ebiom.2021.103697_bib0006) 2015; 351
Feng (10.1016/j.ebiom.2021.103697_bib0025) 2018
Wiedemann (10.1016/j.ebiom.2021.103697_bib0017) 2006; 354
Greene (10.1016/j.ebiom.2021.103697_bib0021) 2016; 17
Lessov-Schlaggar (10.1016/j.ebiom.2021.103697_bib0023) 2016; 10
Sinha (10.1016/j.ebiom.2021.103697_bib0020) 2021; 49
Kent (10.1016/j.ebiom.2021.103697_bib0002) 2007; 298
Wallach (10.1016/j.ebiom.2021.103697_bib0007) 2017; 177
Marshall (10.1016/j.ebiom.2021.103697_bib0003) 2014; 20
Tonekaboni (10.1016/j.ebiom.2021.103697_bib0029) 2019
Wager (10.1016/j.ebiom.2021.103697_bib0015) 2018; 113
References_xml – volume: 113
  start-page: 7353
  year: 2016
  end-page: 7360
  ident: bib0014
  article-title: Recursive partitioning for heterogeneous causal effects
  publication-title: Proc Natl Acad Sci U S A
– reference: Lipton ZC. The Doctor Just Won't Accept That!2017 November 01, 2017:[arXiv:1711.08037 p.]. Available from:
– volume: 195
  start-page: 331
  year: 2017
  end-page: 338
  ident: bib0011
  article-title: Acute respiratory distress syndrome subphenotypes respond differently to randomized fluid management strategy
  publication-title: Am J Respir Crit Care Med
– volume: 20
  start-page: 195
  year: 2014
  end-page: 203
  ident: bib0003
  article-title: Why have clinical trials in sepsis failed?
  publication-title: Trends Mol Med
– volume: 351
  start-page: h5651
  year: 2015
  ident: bib0006
  article-title: Three simple rules to ensure reasonably credible subgroup analyses
  publication-title: BMJ
– volume: 354
  start-page: 2564
  year: 2006
  end-page: 2575
  ident: bib0017
  article-title: Comparison of two fluid-management strategies in acute lung injury
  publication-title: N Engl J Med
– volume: 351
  start-page: 327
  year: 2004
  end-page: 336
  ident: bib0016
  article-title: Higher versus lower positive end-expiratory pressures in patients with the acute respiratory distress syndrome
  publication-title: N Engl J Med
– volume: 172
  start-page: 35
  year: 2020
  end-page: 45
  ident: bib0024
  article-title: The predictive approaches to treatment effect heterogeneity (PATH) statement
  publication-title: Ann Intern Med
– year: 2020
  ident: bib0026
  article-title: BERTs of a feather do not generalize together: Large variability in generalization across models with similar test set performance
  publication-title: ArXiv
– year: 2018
  ident: bib0025
  article-title: Pathologies of neural models make interpretation difficult
  publication-title: EMNLP
– volume: 6
  start-page: 691
  year: 2018
  end-page: 698
  ident: bib0010
  article-title: Acute respiratory distress syndrome subphenotypes and differential response to simvastatin: secondary analysis of a randomised controlled trial
  publication-title: Lancet Respir Med
– volume: 44
  start-page: 1859
  year: 2018
  end-page: 1869
  ident: bib0012
  article-title: Latent class analysis of ARDS subphenotypes: a secondary analysis of the statins for acutely injured lungs from sepsis (SAILS) study
  publication-title: Intensive Care Med
– volume: 10
  start-page: 297
  year: 2009
  end-page: 314
  ident: bib0019
  article-title: A roadmap of clustering algorithms: finding a match for a biomedical application
  publication-title: Brief Bioinform
– volume: 10
  start-page: 48
  year: 2009
  ident: bib0001
  article-title: Against pragmatism: on efficacy, effectiveness and the real world
  publication-title: Trials
– volume: 298
  start-page: 1209
  year: 2007
  end-page: 1212
  ident: bib0002
  article-title: Limitations of applying summary results of clinical trials to individual patients: the need for risk stratification
  publication-title: JAMA
– volume: 154
  start-page: 1239
  year: 2018
  end-page: 1248
  ident: bib0022
  article-title: Big data and data science in critical care
  publication-title: Chest
– volume: 10
  start-page: 267
  year: 2016
  ident: bib0023
  article-title: The fallacy of univariate solutions to complex systems problems
  publication-title: Front Neurosci
– volume: 5
  start-page: 18
  year: 2019
  ident: bib0004
  article-title: Acute respiratory distress syndrome
  publication-title: Nat Rev Dis Primers
– volume: 321
  start-page: 2003
  year: 2019
  end-page: 2017
  ident: bib0013
  article-title: Derivation, validation, and potential treatment implications of novel clinical phenotypes for sepsis
  publication-title: JAMA
– volume: 363
  start-page: k4245
  year: 2018
  ident: bib0008
  article-title: Personalized evidence based medicine: predictive approaches to heterogeneous treatment effects
  publication-title: BMJ
– volume: 2
  start-page: 611
  year: 2014
  end-page: 620
  ident: bib0009
  article-title: Subphenotypes in acute respiratory distress syndrome: latent class analysis of data from two randomised controlled trials
  publication-title: Lancet Respir Med
– volume: 17
  start-page: 43
  year: 2016
  end-page: 50
  ident: bib0021
  article-title: Adapting bioinformatics curricula for big data
  publication-title: Brief Bioinform
– volume: 177
  start-page: 554
  year: 2017
  end-page: 560
  ident: bib0007
  article-title: Evaluation of evidence of statistical support and corroboration of subgroup claims in randomized clinical trials
  publication-title: JAMA Intern Med
– reference: .
– reference: Lahav O, Mastronarde N, van der Schaar M. What is Interpretable? Using Machine Learning to Design Interpretable Decision-Support Systems2018 November 01, 2018:[arXiv:1811.10799 p.]. Available from:
– year: 2013
  ident: bib0005
  article-title: Estimation and reporting of heterogeneity of treatment effects
  publication-title: Developing a Protocol for Observational Comparative Effectiveness Research: A User's Guide
– start-page: 359
  year: 2019
  end-page: 380
  ident: bib0029
  article-title: What clinicians want: contextualizing explainable machine learning for clinical end use
  publication-title: Proceedings of the 4th Machine Learning for Healthcare Conference; Proceedings of Machine Learning Research: PMLR
– volume: 113
  start-page: 1228
  year: 2018
  end-page: 1242
  ident: bib0015
  article-title: Estimation and inference of heterogeneous treatment effects using random forests
  publication-title: J Am Stat Assoc
– volume: 49
  start-page: e63
  year: 2021
  end-page: e79
  ident: bib0020
  article-title: Practitioner's guide to latent class analysis: methodological considerations and common pitfalls
  publication-title: Crit Care Med
– volume: 370
  start-page: 2191
  year: 2014
  end-page: 2200
  ident: bib0018
  article-title: Rosuvastatin for sepsis-associated acute respiratory distress syndrome
  publication-title: N Engl J Med
– volume: 2
  start-page: 611
  issue: 8
  year: 2014
  ident: 10.1016/j.ebiom.2021.103697_bib0009
  article-title: Subphenotypes in acute respiratory distress syndrome: latent class analysis of data from two randomised controlled trials
  publication-title: Lancet Respir Med
  doi: 10.1016/S2213-2600(14)70097-9
– volume: 44
  start-page: 1859
  issue: 11
  year: 2018
  ident: 10.1016/j.ebiom.2021.103697_bib0012
  article-title: Latent class analysis of ARDS subphenotypes: a secondary analysis of the statins for acutely injured lungs from sepsis (SAILS) study
  publication-title: Intensive Care Med
  doi: 10.1007/s00134-018-5378-3
– year: 2018
  ident: 10.1016/j.ebiom.2021.103697_bib0025
  article-title: Pathologies of neural models make interpretation difficult
  publication-title: EMNLP
– volume: 113
  start-page: 1228
  issue: 523
  year: 2018
  ident: 10.1016/j.ebiom.2021.103697_bib0015
  article-title: Estimation and inference of heterogeneous treatment effects using random forests
  publication-title: J Am Stat Assoc
  doi: 10.1080/01621459.2017.1319839
– volume: 177
  start-page: 554
  issue: 4
  year: 2017
  ident: 10.1016/j.ebiom.2021.103697_bib0007
  article-title: Evaluation of evidence of statistical support and corroboration of subgroup claims in randomized clinical trials
  publication-title: JAMA Intern Med
  doi: 10.1001/jamainternmed.2016.9125
– year: 2013
  ident: 10.1016/j.ebiom.2021.103697_bib0005
  article-title: Estimation and reporting of heterogeneity of treatment effects
– volume: 172
  start-page: 35
  issue: 1
  year: 2020
  ident: 10.1016/j.ebiom.2021.103697_bib0024
  article-title: The predictive approaches to treatment effect heterogeneity (PATH) statement
  publication-title: Ann Intern Med
  doi: 10.7326/M18-3667
– volume: 351
  start-page: h5651
  year: 2015
  ident: 10.1016/j.ebiom.2021.103697_bib0006
  article-title: Three simple rules to ensure reasonably credible subgroup analyses
  publication-title: BMJ
  doi: 10.1136/bmj.h5651
– volume: 10
  start-page: 267
  year: 2016
  ident: 10.1016/j.ebiom.2021.103697_bib0023
  article-title: The fallacy of univariate solutions to complex systems problems
  publication-title: Front Neurosci
  doi: 10.3389/fnins.2016.00267
– volume: 351
  start-page: 327
  issue: 4
  year: 2004
  ident: 10.1016/j.ebiom.2021.103697_bib0016
  article-title: Higher versus lower positive end-expiratory pressures in patients with the acute respiratory distress syndrome
  publication-title: N Engl J Med
  doi: 10.1056/NEJMoa032193
– ident: 10.1016/j.ebiom.2021.103697_bib0027
– volume: 354
  start-page: 2564
  issue: 24
  year: 2006
  ident: 10.1016/j.ebiom.2021.103697_bib0017
  article-title: Comparison of two fluid-management strategies in acute lung injury
  publication-title: N Engl J Med
  doi: 10.1056/NEJMoa062200
– start-page: 359
  year: 2019
  ident: 10.1016/j.ebiom.2021.103697_bib0029
  article-title: What clinicians want: contextualizing explainable machine learning for clinical end use
– year: 2020
  ident: 10.1016/j.ebiom.2021.103697_bib0026
  article-title: BERTs of a feather do not generalize together: Large variability in generalization across models with similar test set performance
  publication-title: ArXiv
– volume: 195
  start-page: 331
  issue: 3
  year: 2017
  ident: 10.1016/j.ebiom.2021.103697_bib0011
  article-title: Acute respiratory distress syndrome subphenotypes respond differently to randomized fluid management strategy
  publication-title: Am J Respir Crit Care Med
  doi: 10.1164/rccm.201603-0645OC
– volume: 370
  start-page: 2191
  issue: 23
  year: 2014
  ident: 10.1016/j.ebiom.2021.103697_bib0018
  article-title: Rosuvastatin for sepsis-associated acute respiratory distress syndrome
  publication-title: N Engl J Med
  doi: 10.1056/NEJMoa1401520
– volume: 363
  start-page: k4245
  year: 2018
  ident: 10.1016/j.ebiom.2021.103697_bib0008
  article-title: Personalized evidence based medicine: predictive approaches to heterogeneous treatment effects
  publication-title: BMJ
  doi: 10.1136/bmj.k4245
– volume: 20
  start-page: 195
  issue: 4
  year: 2014
  ident: 10.1016/j.ebiom.2021.103697_bib0003
  article-title: Why have clinical trials in sepsis failed?
  publication-title: Trends Mol Med
  doi: 10.1016/j.molmed.2014.01.007
– volume: 6
  start-page: 691
  issue: 9
  year: 2018
  ident: 10.1016/j.ebiom.2021.103697_bib0010
  article-title: Acute respiratory distress syndrome subphenotypes and differential response to simvastatin: secondary analysis of a randomised controlled trial
  publication-title: Lancet Respir Med
  doi: 10.1016/S2213-2600(18)30177-2
– volume: 10
  start-page: 48
  year: 2009
  ident: 10.1016/j.ebiom.2021.103697_bib0001
  article-title: Against pragmatism: on efficacy, effectiveness and the real world
  publication-title: Trials
  doi: 10.1186/1745-6215-10-48
– volume: 321
  start-page: 2003
  issue: 20
  year: 2019
  ident: 10.1016/j.ebiom.2021.103697_bib0013
  article-title: Derivation, validation, and potential treatment implications of novel clinical phenotypes for sepsis
  publication-title: JAMA
  doi: 10.1001/jama.2019.5791
– volume: 5
  start-page: 18
  issue: 1
  year: 2019
  ident: 10.1016/j.ebiom.2021.103697_bib0004
  article-title: Acute respiratory distress syndrome
  publication-title: Nat Rev Dis Primers
  doi: 10.1038/s41572-019-0069-0
– volume: 10
  start-page: 297
  issue: 3
  year: 2009
  ident: 10.1016/j.ebiom.2021.103697_bib0019
  article-title: A roadmap of clustering algorithms: finding a match for a biomedical application
  publication-title: Brief Bioinform
  doi: 10.1093/bib/bbn058
– volume: 154
  start-page: 1239
  issue: 5
  year: 2018
  ident: 10.1016/j.ebiom.2021.103697_bib0022
  article-title: Big data and data science in critical care
  publication-title: Chest
  doi: 10.1016/j.chest.2018.04.037
– ident: 10.1016/j.ebiom.2021.103697_bib0028
– volume: 298
  start-page: 1209
  issue: 10
  year: 2007
  ident: 10.1016/j.ebiom.2021.103697_bib0002
  article-title: Limitations of applying summary results of clinical trials to individual patients: the need for risk stratification
  publication-title: JAMA
  doi: 10.1001/jama.298.10.1209
– volume: 17
  start-page: 43
  issue: 1
  year: 2016
  ident: 10.1016/j.ebiom.2021.103697_bib0021
  article-title: Adapting bioinformatics curricula for big data
  publication-title: Brief Bioinform
  doi: 10.1093/bib/bbv018
– volume: 49
  start-page: e63
  issue: 1
  year: 2021
  ident: 10.1016/j.ebiom.2021.103697_bib0020
  article-title: Practitioner's guide to latent class analysis: methodological considerations and common pitfalls
  publication-title: Crit Care Med
  doi: 10.1097/CCM.0000000000004710
– volume: 113
  start-page: 7353
  issue: 27
  year: 2016
  ident: 10.1016/j.ebiom.2021.103697_bib0014
  article-title: Recursive partitioning for heterogeneous causal effects
  publication-title: Proc Natl Acad Sci U S A
  doi: 10.1073/pnas.1510489113
SSID ssj0001542358
Score 2.3975832
Snippet Heterogeneity in Acute Respiratory Distress Syndrome (ARDS), as a consequence of its non-specific definition, has led to a multitude of negative randomised...
AbstractBackgroundHeterogeneity in Acute Respiratory Distress Syndrome (ARDS), as a consequence of its non-specific definition, has led to a multitude of...
Background: Heterogeneity in Acute Respiratory Distress Syndrome (ARDS), as a consequence of its non-specific definition, has led to a multitude of negative...
SourceID doaj
pubmedcentral
proquest
pubmed
crossref
elsevier
SourceType Open Website
Open Access Repository
Aggregation Database
Index Database
Enrichment Source
Publisher
StartPage 103697
SubjectTerms Advanced Basic Science
ARDS
Clustering
Heterogeneity of treatment effect
Internal Medicine
LCA
machine learning
RCTs
Research paper
Title Comparison of machine learning clustering algorithms for detecting heterogeneity of treatment effect in acute respiratory distress syndrome: A secondary analysis of three randomised controlled trials
URI https://www.clinicalkey.com/#!/content/1-s2.0-S2352396421004916
https://www.clinicalkey.es/playcontent/1-s2.0-S2352396421004916
https://dx.doi.org/10.1016/j.ebiom.2021.103697
https://www.ncbi.nlm.nih.gov/pubmed/34861492
https://www.proquest.com/docview/2606922321
https://pubmed.ncbi.nlm.nih.gov/PMC8645454
https://doaj.org/article/dcd0c2641c8d430e8bbc69ed3dea44b9
Volume 74
WOSCitedRecordID wos000730209800010&url=https%3A%2F%2Fcvtisr.summon.serialssolutions.com%2F%23%21%2Fsearch%3Fho%3Df%26include.ft.matches%3Dt%26l%3Dnull%26q%3D
hasFullText 1
inHoldings 1
isFullTextHit
isPrint
journalDatabaseRights – providerCode: PRVAON
  databaseName: DOAJ Directory of Open Access Journals
  customDbUrl:
  eissn: 2352-3964
  dateEnd: 99991231
  omitProxy: false
  ssIdentifier: ssj0001542358
  issn: 2352-3964
  databaseCode: DOA
  dateStart: 20140101
  isFulltext: true
  titleUrlDefault: https://www.doaj.org/
  providerName: Directory of Open Access Journals
link http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwrV1Nj9MwELVgBRIXxDfhY2UkjkQ0seM43JYVKw5ohQRIvVmuPdkGdRPUtEj7C_dvMWMnoQW0e-GW1LET26_2G3nmDWOvrbW1roo8FaIQqRSLOtV14dPSS-WsVFgviLh-Kk9P9Xxefd5J9UU-YVEeOA7cW-_8zOGunTntpZiBXiycqsALD1bKRQjdQ9azY0zF-GBJMaAhsxx9SKXkKDkUnLuAgtvROswzijpXJPm0sy0F9f693elv9vmnE-XOrnRyj90d6CQ_it24z25A-4DdjgkmLx6yy-MpzSDvan4ePCeBD6kizrhbbUkogS7t6qxbN5vlec-Rx3IPdLpABUtymOkQZ4CEnVqZfNN5dAbhTcut226Ar3-f23PfxDAUPmoivONHvCfz21sstoMYSmgQ8YSVbes7RB14PvjPr_AyZBXpH7FvJx--Hn9Mh9QNqVMi36SlUr5QdZbVQoAHiVaOE8JqhThQUCJAQFhSk6tcXuF9Vnrr0Vp2tbAAVojH7KDtWnjKsAOZ9iSWgERT1rqukJMUIC0uJB4qoROWjzNn3KBrTuk1VmZ0YPtuwnQbmm4Tpzthb6ZKP6Ksx9WPvydITI-SJnf4AZFqBqSa65CaMDkCyoxhr7hQY0PN1e8u_1UN-mGx6U1m-tzMzBeCOiE9JxVApP0JU1PNgU9FnnT9K1-NeDc473SEZFvotr1B61dVyCjzLGFPIv6nQRFSI9ercvzgvX_G3qjtl7TNMiiaa9KVK-Sz_zHMz9kd6kp0OXrBDjbrLbxkt9zPTdOvD9nNcq4Pw2LxC4UadVo
linkProvider Directory of Open Access Journals
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=Comparison+of+machine+learning+clustering+algorithms+for+detecting+heterogeneity+of+treatment+effect+in+acute+respiratory+distress+syndrome%3A+A+secondary+analysis+of+three+randomised+controlled+trials&rft.jtitle=EBioMedicine&rft.au=Sinha%2C+Pratik&rft.au=Spicer%2C+Alexandra&rft.au=Delucchi%2C+Kevin+L&rft.au=McAuley%2C+Daniel+F&rft.date=2021-12-01&rft.pub=Elsevier&rft.eissn=2352-3964&rft.volume=74&rft_id=info:doi/10.1016%2Fj.ebiom.2021.103697&rft_id=info%3Apmid%2F34861492&rft.externalDocID=PMC8645454
thumbnail_m http://cvtisr.summon.serialssolutions.com/2.0.0/image/custom?url=https%3A%2F%2Fcdn.clinicalkey.com%2Fck-thumbnails%2F23523964%2FS2352396421X00126%2Fcov150h.gif