Faculty Collaboration Database

Distinct clinical profiles and post-transplant outcomes among kidney transplant recipients with lower education levels: uncovering patterns through machine learning clustering. Ren Fail 2023;45(2):2292163

Date

12/13/2023

Pubmed ID

38087474

Pubmed Central ID

PMC11001364

DOI

10.1080/0886022X.2023.2292163

Scopus ID

2-s2.0-85179645749 (requires institutional sign-in at Scopus site)   1 Citation

Abstract

BACKGROUND: Educational attainment significantly influences post-transplant outcomes in kidney transplant patients. However, research on specific attributes of lower-educated subgroups remains underexplored. This study utilized unsupervised machine learning to segment kidney transplant recipients based on education, further analyzing the relationship between these segments and post-transplant results.

METHODS: Using the OPTN/UNOS 2017-2019 data, consensus clustering was applied to 20,474 kidney transplant recipients, all below a college/university educational threshold. The analysis concentrated on recipient, donor, and transplant features, aiming to discern pivotal attributes for each cluster and compare post-transplant results.

RESULTS: Four distinct clusters emerged. Cluster 1 comprised younger, non-diabetic, first-time recipients from non-hypertensive younger donors. Cluster 2 predominantly included white patients receiving their first-time kidney transplant either preemptively or within three years, mainly from living donors. Cluster 3 included younger re-transplant recipients, marked by elevated PRA, fewer HLA mismatches. In contrast, Cluster 4 captured older, diabetic patients transplanted after prolonged dialysis duration, primarily from lower-grade donors. Interestingly, Cluster 2 showcased the most favorable post-transplant outcomes. Conversely, Clusters 1, 3, and 4 revealed heightened risks for graft failure and mortality in comparison.

CONCLUSIONS: Through unsupervised machine learning, this study proficiently categorized kidney recipients with lesser education into four distinct clusters. Notably, the standout performance of Cluster 2 provides invaluable insights, underscoring the necessity for adept risk assessment and tailored transplant strategies, potentially elevating care standards for this patient cohort.

Author List

Thongprayoon C, Miao J, Jadlowiec C, Mao SA, Mao M, Leeaphorn N, Kaewput W, Pattharanitima P, Garcia Valencia OA, Tangpanithandee S, Krisanapan P, Suppadungsuk S, Nissaisorakarn P, Cooper M, Cheungpasitporn W

Author

Matthew Cooper MD Chief, Director, Professor in the Surgery department at Medical College of Wisconsin

MESH terms used to index this publication - Major topics in bold

Educational Status
Graft Rejection
Graft Survival
Humans
Kidney Transplantation
Living Donors
Machine Learning
Tissue and Organ Procurement
Transplant Recipients