Gleaning Knowledge from Data in the Intensive Care Unit

Pinsky, Michael R.; Dubrawski, Artur

doi:10.1164/rccm.201404-0716cp

Cited by 45 publications

(24 citation statements)

References 23 publications

(10 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…However, previous comparison studies have suggested that machine learning methods can be more accurate than traditional logistic regression across a wide variety of subject areas (6). Thus, when the stakes are high, it is important to consider techniques beyond standard regression to optimize accuracy (7). …”

Section: Introductionmentioning

confidence: 99%

Multicenter Comparison of Machine Learning Methods and Conventional Regression for Predicting Clinical Deterioration on the Wards

Churpek

Yuen

Winslow

et al. 2016

Critical Care Medicine

468

437

View full text Add to dashboard Cite

OBJECTIVE Machine learning methods are flexible prediction algorithms that may be more accurate than conventional regression. We compared the accuracy of different techniques for detecting clinical deterioration on the wards in a large, multicenter database. DESIGN Observational cohort study. SETTING Five hospitals, from November 2008 until January 2013. PATIENTS Hospitalized ward patients INTERVENTIONS None MEASUREMENTS AND MAIN RESULTS Demographic variables, laboratory values, and vital signs were utilized in a discrete-time survival analysis framework to predict the combined outcome of cardiac arrest, intensive care unit transfer, or death. Two logistic regression models (one using linear predictor terms and a second utilizing restricted cubic splines) were compared to several different machine learning methods. The models were derived in the first 60% of the data by date and then validated in the next 40%. For model derivation, each event time window was matched to a non-event window. All models were compared to each other and to the Modified Early Warning score (MEWS), a commonly cited early warning score, using the area under the receiver operating characteristic curve (AUC). A total of 269,999 patients were admitted, and 424 cardiac arrests, 13,188 intensive care unit transfers, and 2,840 deaths occurred in the study. In the validation dataset, the random forest model was the most accurate model (AUC 0.80 [95% CI 0.80–0.80]). The logistic regression model with spline predictors was more accurate than the model utilizing linear predictors (AUC 0.77 vs 0.74; p<0.01), and all models were more accurate than the MEWS (AUC 0.70 [95% CI 0.70–0.70]). CONCLUSIONS In this multicenter study, we found that several machine learning methods more accurately predicted clinical deterioration than logistic regression. Use of detection algorithms derived from these techniques may result in improved identification of critically ill patients on the wards.

show abstract

Section: Introductionmentioning

confidence: 99%

Multicenter Comparison of Machine Learning Methods and Conventional Regression for Predicting Clinical Deterioration on the Wards

Churpek

Yuen

Winslow

et al. 2016

Critical Care Medicine

468

437

View full text Add to dashboard Cite

show abstract

“…23 This and other analytic approaches to aid clinical decisions can be incorporated into intelligent and integrated monitoring systems. 24, 25 In addition, they may help the bedside caregiver to decide to call for help in a more objective and “defensible” manner, since there are human factors which also serve as barriers to recognizing instability and increased failure-to-rescue. 26 In one study of 118 ward nurses, 30% reported hesitancy to call a MET.…”

Section: Discussionmentioning

confidence: 99%

Temporal distribution of instability events in continuously monitored step-down unit patients: Implications for Rapid Response Systems

et al. 2015

Self Cite

View full text Add to dashboard Cite

AIM Medical Emergency Teams (MET) activations are more frequent during daytime and weekdays, but whether due to greater patient instability, proximity from admission time, or caregiver concentration is unclear. We sought to determine if instability events, when they occurred, varied in their temporal distribution. METHODS Monitoring data were recorded (frequency 1/20Hz) in 634 SDU patients (41,635 monitoring hours). Vital sign excursion beyond our MET trigger thresholds defined alerts. The resultant 1,399 alerts from 216 patients were tallied according to clock hour and time elapsed since admission. We fit patient ID (n=216), clock hour, time since SDU admission, and alert present into a null model and three mixed effect logistic regression models: clock hour, hours elapsed since admission, and both clock hour and time elapsed since admission as fixed effect covariates. We performed likelihood ratio tests on these models to assess if, among all alerts, there were proportionally more alerts for any given clock hour, or proximity to admission time. RESULTS Only time elapsed since admission (p<0.001), and not clock hour adjusting for time elapsed since admission (p=0.885), was significant for temporal disproportion. Results were unchanged if the first 24 hours following admission were excluded from the models. CONCLUSION Although instability alerts are distributed most frequently within 24 hours after SDU admission in unstable patients, they are otherwise not more likely to distribute proportionally more frequently during certain clock hours. If MET utilization peaks do not coincide with admission time peaks, other variables contributing to unrecognized instability should be explored.

show abstract

“…To date, the data are mainly collected for patient routine care or for a specific purpose such as a trial and are not used “outside.” This is progressively changing and big data is currently being implemented at patients’ bedside and in clinical research. For example, “fused parametric measures” are routinely used to determine the level of severity [49]. In clinical trials, “pooled trial data” seems to be an inevitable evolution that might be very useful in the future.…”

Section: Main Textmentioning

confidence: 99%

Perspective on optimizing clinical trials in critical care: how to puzzle out recurrent failures

et al. 2016

View full text Add to dashboard Cite

BackgroundCritical care is a complex field of medicine, especially because of its diversity and unpredictability. Mortality rates of the diseases are usually high and patients are critically ill, admitted in emergency, and often have several overlapping diseases. This makes research in critical care also complex because of patients’ conditions and because of the numerous ethical and regulatory requirements and increasing global competition. Many clinical trials in critical care have thus failed and almost no drug has yet been developed to treat intensive care unit (ICU) patients. Learning from the failures, clinical trials must now be optimized.Main bodySeveral aspects can be improved, beginning with the design of studies that should take into account patients’ diversity in the ICU. At the site level, selection should reflect more accurately the potential of recruitment. Management of all players that can be involved in the research at a site level should be a priority. Moreover, training should be offered to all staff members, including the youngest. National and international networks are also part of the future as they create a collective synergy potentially improving the efficacy of sites. Finally, computerization is another area that must be further developed with the appropriate tools.ConclusionClinical research in the ICU is thus a discipline in its own right that still requires tailored approaches. Changes have to be initiated by the investigators themselves as they know all the specificities of the field.

show abstract

Gleaning Knowledge from Data in the Intensive Care Unit

Cited by 45 publications

References 23 publications

Multicenter Comparison of Machine Learning Methods and Conventional Regression for Predicting Clinical Deterioration on the Wards

Multicenter Comparison of Machine Learning Methods and Conventional Regression for Predicting Clinical Deterioration on the Wards

Temporal distribution of instability events in continuously monitored step-down unit patients: Implications for Rapid Response Systems

Perspective on optimizing clinical trials in critical care: how to puzzle out recurrent failures

Contact Info

Product

Resources

About