Development, Evaluation, and Multisite Deployment of a Machine Learning Decision Tree Algorithm To Optimize Urinalysis Parameters for Predicting Urine Culture Positivity

PittUDT, a recursive partitioning decision tree algorithm for predicting urine culture (UC) positivity based on macroscopic and microscopic urinalysis (UA) parameters, was developed in support of a broader system-wide diagnostic stewardship initiative to increase appropriateness of UC testing. Refle...

Celý popis

Uložené v:
Podrobná bibliografia
Vydané v:Journal of clinical microbiology Ročník 61; číslo 6; s. e0029123
Hlavní autori: Seheult, Jansen N, Stram, Michelle N, Contis, Lydia, Pontzer, Raymond E, Hardy, Stephanie, Wertz, William, Baxter, Carla M, Ondras, Michael, Kip, Paula L, Snyder, Graham M, Pasculle, A William
Médium: Journal Article
Jazyk:English
Vydavateľské údaje: United States 20.06.2023
Predmet:
Decision Trees
Humans
Machine Learning
Middle Aged
Retrospective Studies
ROC Curve
Urinalysis - methods
Urinary Tract Infections - diagnosis
Urinary Tract Infections - microbiology
Urine - microbiology
reflex protocol
PittUDT algorithm
machine learning
receiver operating characteristic
urine culture
ISSN:1098-660X, 1098-660X
On-line prístup:Zistit podrobnosti o prístupe
Tagy: Pridať tag
Žiadne tagy, Buďte prvý, kto otaguje tento záznam!
Abstract PittUDT, a recursive partitioning decision tree algorithm for predicting urine culture (UC) positivity based on macroscopic and microscopic urinalysis (UA) parameters, was developed in support of a broader system-wide diagnostic stewardship initiative to increase appropriateness of UC testing. Reflex algorithm training utilized results from 19,511 paired UA and UC cases (26.8% UC positive); the average patient age was 57.4 years, and 70% of samples were from female patients. Receiver operating characteristic (ROC) analysis identified urine white blood cells (WBCs), leukocyte esterase, and bacteria as the best predictors of UC positivity, with areas under the ROC curve of 0.79, 0.78, and 0.77, respectively. Using the held-out test data set (9,773 cases; 26.3% UC positive), the PittUDT algorithm met the prespecified target of a negative predictive value above 90% and resulted in a 30 to 60% total negative proportion (true-negative plus false-negative predictions). These data show that a supervised rule-based machine learning algorithm trained on paired UA and UC data has adequate predictive ability for triaging urine specimens by identifying low-risk urine specimens, which are unlikely to grow pathogenic organisms, with a false-negative proportion under 5%. The decision tree approach also generates human-readable rules that can be easily implemented across multiple hospital sites and settings. Our work demonstrates how a data-driven approach can be used to optimize UA parameters for predicting UC positivity in a reflex protocol, with the intent of improving antimicrobial stewardship and UC utilization, a potential avenue for cost savings.
AbstractList PittUDT, a recursive partitioning decision tree algorithm for predicting urine culture (UC) positivity based on macroscopic and microscopic urinalysis (UA) parameters, was developed in support of a broader system-wide diagnostic stewardship initiative to increase appropriateness of UC testing. Reflex algorithm training utilized results from 19,511 paired UA and UC cases (26.8% UC positive); the average patient age was 57.4 years, and 70% of samples were from female patients. Receiver operating characteristic (ROC) analysis identified urine white blood cells (WBCs), leukocyte esterase, and bacteria as the best predictors of UC positivity, with areas under the ROC curve of 0.79, 0.78, and 0.77, respectively. Using the held-out test data set (9,773 cases; 26.3% UC positive), the PittUDT algorithm met the prespecified target of a negative predictive value above 90% and resulted in a 30 to 60% total negative proportion (true-negative plus false-negative predictions). These data show that a supervised rule-based machine learning algorithm trained on paired UA and UC data has adequate predictive ability for triaging urine specimens by identifying low-risk urine specimens, which are unlikely to grow pathogenic organisms, with a false-negative proportion under 5%. The decision tree approach also generates human-readable rules that can be easily implemented across multiple hospital sites and settings. Our work demonstrates how a data-driven approach can be used to optimize UA parameters for predicting UC positivity in a reflex protocol, with the intent of improving antimicrobial stewardship and UC utilization, a potential avenue for cost savings.PittUDT, a recursive partitioning decision tree algorithm for predicting urine culture (UC) positivity based on macroscopic and microscopic urinalysis (UA) parameters, was developed in support of a broader system-wide diagnostic stewardship initiative to increase appropriateness of UC testing. Reflex algorithm training utilized results from 19,511 paired UA and UC cases (26.8% UC positive); the average patient age was 57.4 years, and 70% of samples were from female patients. Receiver operating characteristic (ROC) analysis identified urine white blood cells (WBCs), leukocyte esterase, and bacteria as the best predictors of UC positivity, with areas under the ROC curve of 0.79, 0.78, and 0.77, respectively. Using the held-out test data set (9,773 cases; 26.3% UC positive), the PittUDT algorithm met the prespecified target of a negative predictive value above 90% and resulted in a 30 to 60% total negative proportion (true-negative plus false-negative predictions). These data show that a supervised rule-based machine learning algorithm trained on paired UA and UC data has adequate predictive ability for triaging urine specimens by identifying low-risk urine specimens, which are unlikely to grow pathogenic organisms, with a false-negative proportion under 5%. The decision tree approach also generates human-readable rules that can be easily implemented across multiple hospital sites and settings. Our work demonstrates how a data-driven approach can be used to optimize UA parameters for predicting UC positivity in a reflex protocol, with the intent of improving antimicrobial stewardship and UC utilization, a potential avenue for cost savings.
PittUDT, a recursive partitioning decision tree algorithm for predicting urine culture (UC) positivity based on macroscopic and microscopic urinalysis (UA) parameters, was developed in support of a broader system-wide diagnostic stewardship initiative to increase appropriateness of UC testing. Reflex algorithm training utilized results from 19,511 paired UA and UC cases (26.8% UC positive); the average patient age was 57.4 years, and 70% of samples were from female patients. Receiver operating characteristic (ROC) analysis identified urine white blood cells (WBCs), leukocyte esterase, and bacteria as the best predictors of UC positivity, with areas under the ROC curve of 0.79, 0.78, and 0.77, respectively. Using the held-out test data set (9,773 cases; 26.3% UC positive), the PittUDT algorithm met the prespecified target of a negative predictive value above 90% and resulted in a 30 to 60% total negative proportion (true-negative plus false-negative predictions). These data show that a supervised rule-based machine learning algorithm trained on paired UA and UC data has adequate predictive ability for triaging urine specimens by identifying low-risk urine specimens, which are unlikely to grow pathogenic organisms, with a false-negative proportion under 5%. The decision tree approach also generates human-readable rules that can be easily implemented across multiple hospital sites and settings. Our work demonstrates how a data-driven approach can be used to optimize UA parameters for predicting UC positivity in a reflex protocol, with the intent of improving antimicrobial stewardship and UC utilization, a potential avenue for cost savings.
Author Hardy, Stephanie
Wertz, William
Kip, Paula L
Pontzer, Raymond E
Seheult, Jansen N
Stram, Michelle N
Contis, Lydia
Snyder, Graham M
Baxter, Carla M
Pasculle, A William
Ondras, Michael
Author_xml – sequence: 1
  givenname: Jansen N
  orcidid: 0000-0002-6850-7495
  surname: Seheult
  fullname: Seheult, Jansen N
  organization: Department of Laboratory Medicine and Pathology, Mayo Clinic, Rochester, Minnesota, USA
– sequence: 2
  givenname: Michelle N
  orcidid: 0000-0002-9956-9453
  surname: Stram
  fullname: Stram, Michelle N
  organization: Department of Forensic Medicine, NYU Langone Health, New York, New York, USA
– sequence: 3
  givenname: Lydia
  surname: Contis
  fullname: Contis, Lydia
  organization: Department of Pathology, University of Pittsburgh, Pittsburgh, Pennsylvania, USA
– sequence: 4
  givenname: Raymond E
  surname: Pontzer
  fullname: Pontzer, Raymond E
  organization: Infection Control and Hospital Epidemiology, UPMC, Pittsburgh, Pennsylvania, USA
– sequence: 5
  givenname: Stephanie
  surname: Hardy
  fullname: Hardy, Stephanie
  organization: Laboratory Service Center, UPMC, Pittsburgh, Pennsylvania, USA
– sequence: 6
  givenname: William
  surname: Wertz
  fullname: Wertz, William
  organization: Laboratory Service Center, UPMC, Pittsburgh, Pennsylvania, USA
– sequence: 7
  givenname: Carla M
  surname: Baxter
  fullname: Baxter, Carla M
  organization: Wolff Center, UPMC, Pittsburgh, Pennsylvania, USA
– sequence: 8
  givenname: Michael
  surname: Ondras
  fullname: Ondras, Michael
  organization: Laboratory Service Center, UPMC, Pittsburgh, Pennsylvania, USA
– sequence: 9
  givenname: Paula L
  orcidid: 0000-0001-8337-4777
  surname: Kip
  fullname: Kip, Paula L
  organization: Wolff Center, UPMC, Pittsburgh, Pennsylvania, USA
– sequence: 10
  givenname: Graham M
  orcidid: 0000-0001-5562-8880
  surname: Snyder
  fullname: Snyder, Graham M
  organization: Division of Infectious Diseases, University of Pittsburgh, Pittsburgh, Pennsylvania, USA
– sequence: 11
  givenname: A William
  orcidid: 0000-0001-5540-2056
  surname: Pasculle
  fullname: Pasculle, A William
  organization: Clinical Microbiology Laboratory, UPMC, Pittsburgh, Pennsylvania, USA
BackLink https://www.ncbi.nlm.nih.gov/pubmed/37227272$$D View this record in MEDLINE/PubMed
BookMark eNpNkEtPwzAQhC0E4n3jjPbIgYDtOI8eUXlKRfTQStyqbbIBI8cOtlOp_CP-JSkPidnDzOHTjLQHbNs6S4ydCH4hhCwv36r2gnM5EolMt9i-4KMyyXP-vP0v77GDEN44F0pl2S7bSwspi-H22ec1rci4riUbz-FmhabHqJ09B7Q1PPYm6qAjwTV1xq03FLgGEB6xetWWYELorbYvA1ANpLMw80RwZV6c1_G1hZmDpy7qVn8QzL22aNZBB5iix5Yi-QCN8zD1VOsqboo2EMF4WO49wdQN83ql4_qI7TRoAh3_-iGb397MxvfJ5OnuYXw1STDN05ggbwouqjqT1NRcZWqZ19jkBamMoypEmSmRNxKLEW9EmXJOqsbvXGWlEEoesrOf3s67955CXLQ6VGQMWnJ9WMhSjGQxSA7o6S_aL1uqF53XLfr14u-98gunfoD1
CitedBy_id crossref_primary_10_3390_antibiotics14030256
crossref_primary_10_1016_j_cca_2024_119854
crossref_primary_10_1007_s10096_024_05027_y
crossref_primary_10_1016_j_ijmm_2025_151652
crossref_primary_10_3390_jpm15050200
crossref_primary_10_1017_ice_2025_10265
crossref_primary_10_1038_s41598_025_16677_z
crossref_primary_10_3389_fendo_2025_1593735
crossref_primary_10_1016_j_biotechadv_2024_108480
ContentType Journal Article
DBID CGR
CUY
CVF
ECM
EIF
NPM
7X8
DOI 10.1128/jcm.00291-23
DatabaseName Medline
MEDLINE
MEDLINE (Ovid)
MEDLINE
MEDLINE
PubMed
MEDLINE - Academic
DatabaseTitle MEDLINE
Medline Complete
MEDLINE with Full Text
PubMed
MEDLINE (Ovid)
MEDLINE - Academic
DatabaseTitleList MEDLINE - Academic
MEDLINE
Database_xml – sequence: 1
  dbid: NPM
  name: PubMed
  url: http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed
  sourceTypes: Index Database
– sequence: 2
  dbid: 7X8
  name: MEDLINE - Academic
  url: https://search.proquest.com/medline
  sourceTypes: Aggregation Database
DeliveryMethod no_fulltext_linktorsrc
Discipline Medicine
Biology
EISSN 1098-660X
ExternalDocumentID 37227272
Genre Journal Article
GroupedDBID ---
.55
0R~
18M
29K
2WC
39C
4.4
53G
5GY
5RE
5VS
ABOCM
ABPPZ
ACGFO
ADBBV
AENEX
AGVNZ
ALMA_UNASSIGNED_HOLDINGS
AOIJS
BAWUL
BTFSW
CGR
CS3
CUY
CVF
D-I
DIK
DU5
E3Z
EBS
ECM
EIF
F5P
FRP
GX1
H13
HYE
HZ~
KQ8
L7B
NPM
O9-
OK1
P2P
P6G
RHF
RHI
RNS
RPM
RSF
TR2
UCJ
W8F
WOQ
X7M
YIF
ZCA
~KM
7X8
AAGFI
ID FETCH-LOGICAL-a363t-a0f701cd52efd0454b6daf67e450a47185416f2a790f18300e4da90f18c581142
IEDL.DBID 7X8
ISICitedReferencesCount 8
ISICitedReferencesURI http://www.webofscience.com/api/gateway?GWVersion=2&SrcApp=Summon&SrcAuth=ProQuest&DestLinkType=CitingArticles&DestApp=WOS_CPL&KeyUT=000993048000001&url=https%3A%2F%2Fcvtisr.summon.serialssolutions.com%2F%23%21%2Fsearch%3Fho%3Df%26include.ft.matches%3Dt%26l%3Dnull%26q%3D
ISSN 1098-660X
IngestDate Thu Sep 04 17:01:31 EDT 2025
Wed Feb 19 02:22:51 EST 2025
IsDoiOpenAccess false
IsOpenAccess true
IsPeerReviewed true
IsScholarly true
Issue 6
Keywords reflex protocol
PittUDT algorithm
machine learning
receiver operating characteristic
urine culture
Language English
LinkModel DirectLink
MergedId FETCHMERGED-LOGICAL-a363t-a0f701cd52efd0454b6daf67e450a47185416f2a790f18300e4da90f18c581142
Notes ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 23
ORCID 0000-0002-9956-9453
0000-0002-6850-7495
0000-0001-8337-4777
0000-0001-5540-2056
0000-0001-5562-8880
OpenAccessLink https://pmc.ncbi.nlm.nih.gov/articles/PMC10281150/pdf/jcm.00291-23.pdf
PMID 37227272
PQID 2819277772
PQPubID 23479
ParticipantIDs proquest_miscellaneous_2819277772
pubmed_primary_37227272
PublicationCentury 2000
PublicationDate 2023-06-20
PublicationDateYYYYMMDD 2023-06-20
PublicationDate_xml – month: 06
  year: 2023
  text: 2023-06-20
  day: 20
PublicationDecade 2020
PublicationPlace United States
PublicationPlace_xml – name: United States
PublicationTitle Journal of clinical microbiology
PublicationTitleAlternate J Clin Microbiol
PublicationYear 2023
SSID ssj0014455
Score 2.4759717
Snippet PittUDT, a recursive partitioning decision tree algorithm for predicting urine culture (UC) positivity based on macroscopic and microscopic urinalysis (UA)...
SourceID proquest
pubmed
SourceType Aggregation Database
Index Database
StartPage e0029123
SubjectTerms Decision Trees
Humans
Machine Learning
Middle Aged
Retrospective Studies
ROC Curve
Urinalysis - methods
Urinary Tract Infections - diagnosis
Urinary Tract Infections - microbiology
Urine - microbiology
Title Development, Evaluation, and Multisite Deployment of a Machine Learning Decision Tree Algorithm To Optimize Urinalysis Parameters for Predicting Urine Culture Positivity
URI https://www.ncbi.nlm.nih.gov/pubmed/37227272
https://www.proquest.com/docview/2819277772
Volume 61
WOSCitedRecordID wos000993048000001&url=https%3A%2F%2Fcvtisr.summon.serialssolutions.com%2F%23%21%2Fsearch%3Fho%3Df%26include.ft.matches%3Dt%26l%3Dnull%26q%3D
hasFullText
inHoldings 1
isFullTextHit
isPrint
link http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwpV1NT9wwELXKp7iUsnyU8qGpxHEtvE7iJKcKFVa97JLDIu1t5dhjoOomsECl9h_xL_E4WZYLUqXmEOUwiSyNMx75Pb_H2IkwJkFf-Lg1ieOxURkvdY48ypUtrZYmVo3ZRDocZuNxXrQbbg8trXJeE0OhtrWhPfJTAnxk6i_57e6ek2sUoauthcYSW4l8K0OzOh0vUIQ4Dq6nPdLMVEqM58R3mZ3-NFPaS8l7nIyK3msuwyLT3_zf4X1iH9v2Es6a-bDFPmDVYWuN4eSfDlsftFD6Nnt-QxfqwsWr6HcXdGUhnMulb8M5kiUwRUHtQMMgsC8RWmHWax_QuPTAaIYIZ7-u_bAeb6YwquHS16Pp7V-Eq9ltK34ChSY6GGl6gu-XoZjRiIh7HYIQGo1PhCKwycjZYodd9S9G33_w1reB60hFj1wLl4qesYlEZ0nir1RWO5VinAhNi2Hiu0AndZoL5yuKEBhbHZ5NktHZ3l22XNUVfmaAiXI6yaPSKBFbQoyyUgtSY8qUMDbfZ1_n6Zj4_4LADl1h_fQwWSRkn-01OZ3cNQIekyiVkgDoL__w9gHbIId5YodJcchWnK8KeMRWzW-fhtlxmHD-PiwGL-7x4zU
linkProvider ProQuest
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=Development%2C+Evaluation%2C+and+Multisite+Deployment+of+a+Machine+Learning+Decision+Tree+Algorithm+To+Optimize+Urinalysis+Parameters+for+Predicting+Urine+Culture+Positivity&rft.jtitle=Journal+of+clinical+microbiology&rft.au=Seheult%2C+Jansen+N&rft.au=Stram%2C+Michelle+N&rft.au=Contis%2C+Lydia&rft.au=Pontzer%2C+Raymond+E&rft.date=2023-06-20&rft.issn=1098-660X&rft.eissn=1098-660X&rft.volume=61&rft.issue=6&rft.spage=e0029123&rft_id=info:doi/10.1128%2Fjcm.00291-23&rft.externalDBID=NO_FULL_TEXT
thumbnail_l http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/lc.gif&issn=1098-660X&client=summon
thumbnail_m http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/mc.gif&issn=1098-660X&client=summon
thumbnail_s http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/sc.gif&issn=1098-660X&client=summon