2013 OutlierDetectionforPatientMonit

• (Hauskrecht et al., 2013) ⇒ Milos Hauskrecht, Iyad Batal, Michal Valko, Shyam Visweswaran, Gregory F Cooper, and Gilles Clermont. (2013). “Outlier Detection for Patient Monitoring and Alerting.” In: Journal of Biomedical Informatics, 46(1). doi:10.1016/j.jbi.2012.08.004

Subject Headings: Outlier Detection.

Cited By

• http://scholar.google.com/scholar?q=%222013%22+Outlier+Detection+for+Patient+Monitoring+and+Alerting

Quotes

Author Keywords

• Machine learning; Clinical alerting; Conditional outlier detection; Medical errors

Abstract

We develop and evaluate a data-driven approach for detecting unusual (anomalous) patient-management decisions using past patient cases stored in electronic health records (EHRs). Our hypothesis is that a patient-management decision that is unusual with respect to past patient care may be due to an error and that it is worthwhile to generate an alert if such a decision is encountered. We evaluate this hypothesis using data obtained from EHRs of 4486 post-cardiac surgical patients and a subset of 222 alerts generated from the data. We base the evaluation on the opinions of a panel of experts. The results of the study support our hypothesis that the outlier-based alerting can lead to promising true alert rates. We observed true alert rates that ranged from 25% to 66% for a variety of patient-management actions, with 66% corresponding to the strongest outliers.

Highlights

• An approach for detecting unusual patient-management decisions from EHR data.
• A decision that is unusual may be due to an error and an alert should be considered.
• Tests the approach using expert reviews on data for postsurgical cardiac patients.
• The results show false alert rates from 25% to 66%.
• The results show stronger outliers are correlated with higher true alert rates.

1. Introduction

Despite numerous improvements in health-care practice, the occurrence of medical errors remains a persistent and serious problem [ 1 ] and [ 2 ]. The well-known Institute of Medicine’s report To Err Is Human – Building a Safer Health System estimated that between 44,000 and 98,000 Americans die each year as a result of medical errors [ 1 ].

…

…

3.1 Conditional anomaly models

Anomaly detection is an active area of current machine learning and data mining research. An outlier (or a deviation or an anomaly) is an observation or a pattern in the data that appears to deviate significantly from other observations or patterns in the same data [ 8 ] and [ 9 ]. Anomaly detection methods have been applied to problems as diverse as monitoring of credit card transactions, detection of network intrusions, and detection of technical system failures.

Standard outlier detection methods try to identify unusual data instances [ 9 ]. In the clinical settings, these would correspond to unusual combination of symptoms, defining, for example, a rare disease. Conditional anomaly detection (CAD) [ 10 ] aims to detect unusual outcomes for a subset of (response) attributes given the values of the remaining (context) attributes. CAD is particularly suitable for detecting unusual outcomes, unusual behaviors, and unusual attribute pairings. In this work we use CAD to detect unusual patient-management actions in the context of an existing patient condition.

3.1.1. Measure of deviation

In our approach, we quantify the deviation of a patient-management action using conditional probability measures. Let y* denote a patient-management action (such as a medication order) and let x* denote information about the current patient state. We say the action y* is conditionally anomalous given x*, if the probability P (y*|x*) is small, or equivalently if the probability 1 − P (y*|x*) is large. We represent the level of anomalousness of action y* for x* in terms of the following anomaly score: equation (1)

Anom(x∗,y∗)=1-P(y∗|x∗).Anom(x∗,y∗)=1-P(y∗|x∗).

…

References

;

	Author	volume	Date Value	title	type	journal	titleUrl	doi	note	year
2013 OutlierDetectionforPatientMonit	Iyad Batal Milos Hauskrecht Michal Valko Shyam Visweswaran Gregory F Cooper Gilles Clermont			Outlier Detection for Patient Monitoring and Alerting