Using Supervised Machine Learning to Classify Real Alerts and Artifact in Online Multisignal Vital Sign Monitoring Data.

OBJECTIVE: The use of machine-learning algorithms to classify alerts as real or artifacts in online noninvasive vital sign data streams to reduce alarm fatigue and missed true instability. DESIGN: Observational cohort study. SETTING: Twenty-four-bed trauma step-down unit. PATIENTS: Two thousand one hundred fifty-three patients. INTERVENTION: Noninvasive vital sign monitoring data (heart rate, respiratory rate, peripheral oximetry) recorded on all admissions at 1/20 Hz, and noninvasive blood pressure less frequently, and partitioned data into training/validation (294 admissions; 22,980 monitoring hours) and test sets (2,057 admissions; 156,177 monitoring hours). Alerts were vital sign deviations beyond stability thresholds. A four-member expert committee annotated a subset of alerts (576 in training/validation set, 397 in test set) as real or artifact selected by active learning, upon which we trained machine-learning algorithms. The best model was evaluated on test set alerts to enact online alert classification over time. MEASUREMENTS AND MAIN RESULTS: The Random Forest model discriminated between real and artifact as the alerts evolved online in the test set with area under the curve performance of 0.79 (95% CI, 0.67-0.93) for peripheral oximetry at the instant the vital sign first crossed threshold and increased to 0.87 (95% CI, 0.71-0.95) at 3 minutes into the alerting period. Blood pressure area under the curve started at 0.77 (95% CI, 0.64-0.95) and increased to 0.87 (95% CI, 0.71-0.98), whereas respiratory rate area under the curve started at 0.85 (95% CI, 0.77-0.95) and increased to 0.97 (95% CI, 0.94-1.00). Heart rate alerts were too few for model development. CONCLUSIONS: Machine-learning models can discern clinically relevant peripheral oximetry, blood pressure, and respiratory rate alerts from artifacts in an online monitoring dataset (area under the curve > 0.87).

Increasing Cardiovascular Data Sampling Frequency and Referencing It to Baseline Improve Hemorrhage Detection.

We hypothesize that knowledge of a stable personalized baseline state and increased data sampling frequency would markedly improve the ability to detect progressive hypovolemia during hemorrhage earlier and with a lower false positive rate than when using less granular data. DESIGN: Prospective temporal challenge. SETTING: Large animal research laboratory, University Medical Center. SUBJECTS: Fifty-one anesthetized Yorkshire pigs. INTERVENTIONS: Pigs were instrumented with arterial, pulmonary arterial, and central venous catheters and allowed to stabilize for 30 minutes then bled at a constant rate of either 5 mL·min-1 (n = 13) or 20 (n = 38) until mean arterial pressure decreased to 40 or 30 mm Hg in the 5 and 20 mL·min-1 pigs, respectively. MEASUREMENTS AND MAIN RESULTS: Data during the stabilization period served as baseline. Hemodynamic variables collected at 250 Hz were used to create predictive models of "bleeding" using featurized beat-to-beat and waveform data and compared with models using mean unfeaturized hemodynamic variables averaged over 1-minute as simple hemodynamic metrics using random forest classifiers to identify bleeding with or without baseline data. The robustness of the prediction was evaluated in a leave-one-pig-out cross-validation. Predictive performance of models was compared by their activity monitoring operating characteristic and receiver operating characteristic profiles. Primary hemodynamic threshold data poorly identified bleed onset unless very stable initial baseline reference data were available. When referenced to baseline, bleed detection at a false positive rates of 10-2 with time to detect 80% of pigs bleeding was similar for simple hemodynamic metrics, beat-to-beat, and waveform at about 3-4 minutes. Whereas when universally baselined, increasing sampling frequency reduced latency of bleed detection from 10 to 8 to 6 minutes, for simple hemodynamic metrics, beat-to-beat, and waveform, respectively. Some informative features differed between simple hemodynamic metrics, beat-to-beat, and waveform models. CONCLUSIONS: Knowledge of personal stable baseline data allows for early detection of new-onset bleeding, whereas if no personal baseline exists increasing sampling frequency of hemodynamic monitoring data improves bleeding detection earlier and with lower false positive rate.

Learning temporal rules to forecast instability in continuously monitored patients.

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.