Unsupervised phenotyping of sepsis using nonnegative matrix factorization of temporal trends from a multivariate panel of physiological measurements

Abstract

Background Sepsis is a highly lethal and heterogeneous disease. Utilization of an unsupervised method may identify novel clinical phenotypes that lead to targeted therapies and improved care. Methods Our objective was to derive clinically relevant sepsis phenotypes from a multivariate panel of physiological data using subgraph-augmented nonnegative matrix factorization. We utilized data from the Medical Information Mart for Intensive Care III database of patients who were admitted to the intensive care unit with sepsis. The extracted data contained patient demographics, physiological records, sequential organ failure assessment scores, and comorbidities. We applied frequent subgraph mining to extract subgraphs from physiological time series and performed nonnegative matrix factorization over the subgraphs to derive patient clusters as phenotypes. Finally, we profiled these phenotypes based on demographics, physiological patterns, disease trajectories, comorbidities and outcomes, and performed functional validation of their clinical implications. Results We analyzed a cohort of 5782 patients, derived three novel phenotypes of distinct clinical characteristics and demonstrated their prognostic implications on patient outcome. Subgroup 1 included relatively less severe/deadly patients (30-day mortality, 17%) and was the smallest-in-size group (n = 1218, 21%). It was characterized by old age (mean age, 73 years), a male majority (male-to-female ratio, 59-to-41), and complex chronic conditions. Subgroup 2 included the most severe/deadliest patients (30-day mortality, 28%) and was the second-in-size group (n = 2036, 35%). It was characterized by a male majority (male-to-female ratio, 60-to-40), severe organ dysfunction or failure compounded by a wide range of comorbidities, and uniquely high incidences of coagulopathy and liver disease. Subgroup 3 included the least severe/deadly patients (30-day mortality, 10%) and was the largest group (n = 2528, 44%). It was characterized by low age (mean age, 60 years), a balanced gender ratio (male-to-female ratio, 50-to-50), the least complicated conditions, and a uniquely high incidence of neurologic disease. These phenotypes were validated to be prognostic factors of mortality for sepsis patients. Conclusions Our results suggest that these phenotypes can be used to develop targeted therapies based on phenotypic heterogeneity and algorithms designed for monitoring, validating and intervening clinical decisions for sepsis patients.

Methods for Phenotyping Adult Patients in Sepsis and Septic Shock: A Scoping Review

“…No articles blinded outcome assessors to the exposure status of participants or justified sample sizes. Seven articles investigated clustering variables more than once over time (15, 16, 21, 23, 25–27), and 12 articles adjusted for key confounding variables (13, 15, 16, 18–20, 22–25, 28, 29). Loss-to-follow-up measurements were not applicable toward 12 articles given the retrospective nature of these analyses (12, 15–21, 25–27, 29).…”

“…All articles were either retrospective, prospective, or a mixed retrospective analysis and prospective cohort study ( Table 1 ). Two articles retrieved patients from the Medical Information Mart for Intensive Care-III database (16, 21), and two articles used data from the Gene Expression Omnibus (GEO) and ArrayExpress (28, 29). There were no other overlaps in datasets.…”

“…Out of all 17 articles, 11 studies performed sensitivity analyses (13, 16, 18–20, 22–25, 28, 29), and nine studies validated their findings in a separate cohort or had their findings validated by a previous study (Supplemental Table 5, http://links.lww.com/CCX/A963) (13, 15, 19, 20, 22–24, 28, 29).…”

“…Seven articles investigated clustering variables more than once over time (15, 16, 21, 23, 25–27), and 12 articles adjusted for key confounding variables (13, 15, 16, 18–20, 22–25, 28, 29). Loss-to-follow-up measurements were not applicable toward 12 articles given the retrospective nature of these analyses (12, 15–21, 25–27, 29). Given that identified patient subgroups were nominal classifications in 15 articles, the risk of bias field evaluating whether the outcome associated with increasing levels of the independent variable in a dose-dependent fashion was only applicable to two studies (25, 26).…”

“…The remaining articles either retrieved data from multiple studies using different sepsis definitions, as in ArrayExpress or GEO, or used separate definitions for sepsis (18, 23, 26–29). Exclusion criteria aimed to eliminate cases with missing or nonsensical data in seven articles (16, 19–21, 23, 27, 29) and to removing confounding patient populations in 10 articles (12, 15, 17, 18, 21, 22, 24, 27–29). Exclusion criteria were not stated in three articles (13, 25, 26).…”