Learning temporal rules to forecast instability in continuously monitored patients

Abstract

Inductive machine learning, and in particular extraction of association rules from data, has been successfully used in multiple application domains, such as market basket analysis, disease prognosis, fraud detection, and protein sequencing. The appeal of rule extraction techniques stems from their ability to handle intricate problems yet produce models based on rules that can be comprehended by humans, and are therefore more transparent. Human comprehension is a factor that may improve adoption and use of data-driven decision support systems clinically via face validity. In this work, we explore whether we can reliably and informatively forecast cardiorespiratory instability (CRI) in step-down unit (SDU) patients utilizing data from continuous monitoring of physiologic vital sign (VS) measurements. We use a temporal association rule extraction technique in conjunction with a rule fusion protocol to learn how to forecast CRI in continuously monitored patients. We detail our approach and present and discuss encouraging empirical results obtained using continuous multivariate VS data from the bedside monitors of 297 SDU patients spanning 29 346 hours (3.35 patient-years) of observation. We present example rules that have been learned from data to illustrate potential benefits of comprehensibility of the extracted models, and we analyze the empirical utility of each VS as a potential leading indicator of an impending CRI event.

Artificial intelligence and machine learning in emergency medicine

“…CRI decreases in a linear fashion following voluntary haemorrhage, and the addition of CRI to the monitor screen has been associated with earlier identification of impending instability . Another ML model trained to assess vital signs issues cardiovascular instability alerts for intensive care unit stepdown patients on average 17 min and 51 s before onset of a cardiovascular instability event …”

“…21,22 Another ML model trained to assess vital signs issues cardiovascular instability alerts for intensive care unit stepdown patients on average 17 min and 51 s before onset of a cardiovascular instability event. 23 Sepsis outcomes improve significantly with earlier detection and treatment. ML applied to heart rate and BP dynamics can independently predict sepsis 4 h prior to clinical onset.…”

Explaining an increase in predicted risk for clinical alerts

“…To explain predictions on EHR data (the focus of this paper), a variety of approaches have been explored starting with generalized additive models [11] over discretized features, or feature crosses fitted using gradient boosted decision trees. Subsequent work has extracted more complex discretized features that also incorporate temporal aspects using the maximum information gain criteria [43]. To explain a patient's risk, statistics of these discretized features can be used such as the odds ratio or the Rothman index [12].…”

Automated Blood Pressure Control

“…Artificial intelligence, in terms of machine learning, can be used to predict hypotension before it clinically becomes apparent by analyzing subtle hemodynamic changes that precede clinical hypotension. [80][81][82][83] A hypotension prediction algorithm-resulting in a unit-less "hypotension prediction index" (HPI)-was proposed by Hatib et al 84 The machine learning algorithm was trained in a cohort of OR and ICU patients and analyzes characteristics of the AP waveform. The HPI uses "a MAP of less than 65 mm Hg for at least 1 minute" to define hypotension and can take values between 0 and 100, with higher numbers indicating a higher risk of hypotension.…”