Less is more: Detecting clinical deterioration in the hospital with machine learning using only age, heart rate, and respiratory rate.

AIM: We sought to develop a machine learning analytic (eCART Lite) for predicting clinical deterioration using only age, heart rate, and respiratory data, which can be pulled in real time from patient monitors and updated continuously without need for additional inputs or cumbersome electronic health record integrations. METHODS: We utilized a multicenter dataset of adult admissions from five hospitals. We trained a gradient boosted machine model using only current and 24-hour trended heart rate, respiratory rate, and patient age to predict the probability of intensive care unit (ICU) transfer, death, or the combined outcome of ICU transfer or death. The area under the receiver operating characteristic curve (AUC) was calculated in the validation cohort and compared to those for the Modified Early Warning Score (MEWS), National Early Warning Score (NEWS), and eCARTv2, a previously-described, 27-variable, cubic spline, logistic regression model without trends. RESULTS: Of the 556,848 included admissions, 19,509 (3.5%) were transferred to an ICU and 5764 (1.0%) died within 24 hours of a ward observation. eCART Lite significantly outperformed the MEWS, NEWS, and eCART v2 for predicting ICU transfer (0.79 vs 0.71, 0.74, and 0.78, respectively; p < 0.01) and the combined outcome (0.80 vs 0.72, 0.76, and 0.79, respectively; p < 0.01). Two of the strongest predictors were respiratory rate and heart rate. CONCLUSION: Using only three inputs, we developed a tool for predicting clinical deterioration that is similarly or more accurate than commonly-used algorithms, with potential for use in inpatient settings with limited resources or in scenarios where low-cost tools are needed.

Use of serial analysis of gene expression to generate kidney expression libraries.

Chronic renal disease initiation and progression remain incompletely understood. Genome-wide expression monitoring should clarify mechanisms that cause progressive renal disease by determining how clusters of genes coordinately change their activity. Serial analysis of gene expression (SAGE) is a technique of expression profiling, which permits simultaneous, comparative, and quantitative analysis of gene-specific, 9- to 13-bp sequence tags. Using SAGE, we have constructed a tag expression library from ROP-+/+ mouse kidney. Tag sequences were sorted by abundance, and identity was determined by sequence homology searching. Analyses of 3,868 tags yielded 1,453 unique kidney transcripts. Forty-two percent of these transcripts matched mRNA sequence entries with known function, 35% of the transcripts corresponded to expressed sequence tag (EST) entries or cloned genes, whose function has not been established, and 23% represented unidentified genes. Previously characterized transcripts were clustered into functional groups, and those encoding metabolic enzymes, plasma membrane proteins (transporters/receptors), and ribosomal proteins were most abundant (39, 14, and 12% of known transcripts, respectively). The most common, kidney-specific transcripts were kidney androgen-regulated protein (4% of all transcripts), sodium-phosphate cotransporter (0.3%), renal cytochrome P-450 (0.3%), parathyroid hormone receptor (0.1%), and kidney-specific cadherin (0.1%). Comprehensively characterizing and contrasting gene expression patterns in normal and diseased kidneys will provide an alternative strategy to identify candidate pathways, which regulate nephropathy susceptibility and progression, and novel targets for therapeutic intervention.