Comparison of time series clustering methods for identifying novel subphenotypes of patients with infection

被引：14

作者：

Bhavani, Sivasubramanium, V ^{[1
,2
,12
]}

Xiong, Li ^{[3
]}

Pius, Abish ^{[4
]}

Semler, Matthew ^{[5
]}

Qian, Edward T. ^{[5
]}

Verhoef, Philip A. ^{[6
,7
]}

Robichaux, Chad ^{[8
]}

Coopersmith, Craig M. ^{[2
,9
]}

Churpek, Matthew M. ^{[10
,11
]}

机构：

[1] Emory Univ, Dept Med, Atlanta, GA 30322 USA

[2] Emory Crit Care Ctr, Atlanta, GA USA

[3] Emory Univ, Dept Comp Sci, Atlanta, GA 30322 USA

[4] Univ Pittsburgh, Dept Computat & Syst Biol, Sch Med, Pittsburgh, PA USA

[5] Vanderbilt Univ, Dept Med, Nashville, TN USA

[6] Univ Hawaii, John A Burns Sch Med, Dept Med, Honolulu, HI USA

[7] Hawaii Permanente Med Grp, Honolulu, HI USA

[8] Emory Univ, Dept Biomed Informat, Atlanta, GA 30322 USA

[9] Emory Univ, Dept Surg, Atlanta, GA 30322 USA

[10] Univ Wisconsin, Dept Med, Madison, WI USA

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

[12] Emory Univ, Sch Med, Div Pulm Allergy Crit Care & Sleep Med, 615 Michael St, Atlanta, GA 30322 USA

来源：

JOURNAL OF THE AMERICAN MEDICAL INFORMATICS ASSOCIATION | 2023年 / 30卷 / 06期

关键词：

infection; sepsis; vital signs; subphenotypes; phenotypes; SEPSIS;

D O I：

10.1093/jamia/ocad063

中图分类号：

TP [自动化技术、计算机技术];

学科分类号：

0812 ;

摘要：

Objective: Severe infection can lead to organ dysfunction and sepsis. Identifying subphenotypes of infected patients is essential for personalized management. It is unknown how different time series clustering algo-rithms compare in identifying these subphenotypes.Materials and Methods: Patients with suspected infection admitted between 2014 and 2019 to 4 hospitals in Emory healthcare were included, split into separate training and validation cohorts. Dynamic time warping (DTW) was applied to vital signs from the first 8 h of hospitalization, and hierarchical clustering (DTW-HC) and partition around medoids (DTW-PAM) were used to cluster patients into subphenotypes. DTW-HC, DTW-PAM, and a previ-ously published group-based trajectory model (GBTM) were evaluated for agreement in subphenotype clusters, trajectory patterns, and subphenotype associations with clinical outcomes and treatment responses.Results: There were 12473 patients in training and 8256 patients in validation cohorts. DTW-HC, DTW-PAM, and GBTM models resulted in 4 consistent vitals trajectory patterns with significant agreement in clustering (71-80% agreement, P< .001): group A was hyperthermic, tachycardic, tachypneic, and hypotensive. Group B was hyperthermic, tachycardic, tachypneic, and hypertensive. Groups C and D had lower temperatures, heart rates, and respiratory rates, with group C normotensive and group D hypotensive. Group A had higher odds ratio of 30-day inpatient mortality (P< .01) and group D had significant mortality benefit from balanced crystalloids compared to saline (P< .01) in all 3 models.Discussion: DTW-and GBTM-based clustering algorithms applied to vital signs in infected patients identified consistent subphenotypes with distinct clinical outcomes and treatment responses.Conclusion: Time series clustering with distinct computational approaches demonstrate similar performance and significant agreement in the resulting subphenotypes.

引用

页码：1158 / 1166

页数：9

共 27 条

[1] Time-series clustering - A decade review [J].