Syndrome classification

Complex, highly parameterized data models are often used to detect syndromic outbreaks. Unfortunately, such models can pose greater maintenance challenges as parameter variations increase. As such, our work focuses on whether day-of-the-week (DoW) effects may (or may not) show little variation across hospitals.

Objective

This paper investigates the existence of the DoW effect across twenty-six hospitals within the Indiana Public Health Emergency Surveillance System. We will consider both the impact of each DoW and the impact of individual hospitals.

Scientists have utilized many chief complaint (CC) classification techniques in biosurveillance including keyword search, weighted keyword search, and naïve Bayes. These techniques may utilize CC-to-syndrome or CC-to-symptom-to-syndrome classification approaches. In the former approach, we classify a CC directly into syndrome categories. In the latter approach, we first classify a CC into symptom categories. Then, we use a syndrome definition, a combination of one or more symptoms, to determine whether or not a chief complaint belongs in a particular syndrome category. One approach to CC-to-symptom-to-syndrome classification uses manually weighted keyword search and Boolean operations to build syndrome classifiers. A limitation to this approach is that it does not address uncertainty in the data and the system is manually parameterized. A CC-tosymptom-to-syndrome approach that is both probabilistic and utilizes machine learning addresses these limitations.

Objective

Design, build and evaluate a symptom-based probabilistic chief complaint classifier for the Real-time Outbreak and Disease Surveillance System.

Previously we developed an “Ngram” classifier for syndromic surveillance of emergency department (ED) chief complaints (CC) in Turkish for bioterrorism. The classifier is developed from a set of ED visits for which both the ICD diagnosis code and CC are available. A computer program calculates the associations of text fragments within the CC (e.g. 3 characters for a “3-gram”) with a syndromic group of ICD codes. The program then generates an algorithm which can be deployed to evaluate chief complaint data in real-time. However, the N-gram method differs from most other classifiers in that it assigns a probability that each visit falls within the syndrome rather than ruling the visit “in” or “out” of the syndrome. It is possible to dichotomize visits “in” or “out” using N-grams by choosing a cut-off sensitivity for the n-grams used, but this affects the specificity of the method. The effect of this trade-off is best measured by a receiveroperator curve.

Objective

Our objective was to determine the sensitivity and specificity of the Ngram CC classifier for individual ED visits. We also wish to compare these results to those obtained when we substituted anglicized characters for 6 problematic Turkish characters.

The Centers for Disease Control and Prevention BioSense project has developed chief complaint (CC) and ICD9 sub-syndrome classifiers for the major syndromes for early event detection and situational awareness. This has the potential to expand the usefulness of syndromic surveillance, but little data exists evaluating this approach. The overall performance of classifiers can differ significantly among syndromes, and presumably among subsyndromes as well. Also, we had previously found that the seasonal pattern of diarrhea was different for patients < 60 months of age (younger) and for patients > 60 months of age (older).

Objective

Using chart review as the criterion standard to estimate the sensitivity, specificity, positive predictive value and negative predictive value of New York State hospital emergency department CC classifiers for patients < 60 months of age and > 60 months of age for the gastrointestinal (GI) syndrome and the following GI sub-syndromes: “abdominal pain”, “nausea-vomiting” and “diarrhea”.

This paper describes a novel approach to the statistical classification of free-text chief complaints for the purpose of syndromic surveillance.

To understand the types of false positive cases identified by an Influenza-like illness (ILI) text classifier by measuring the prevalence of ILI-related concepts that are negated, hypothetical, include explicit mention of temporality, experienced by someone other than the patient, or described in templated text that is difficult to process.

Objective To examine sub-syndrome distributions among BioSense emergency department (ED) chief complaint and final diagnosis based data and to observe patterns by hospital system, age, and gender.

The goal of this project is to create a set of clinical condition categories based on explicit criteria for use in biosurveillance programs. The categories will be defined and keywords and ICD-9-CM diagnosis codes for implementation will be proposed.

We evaluated several classifications of emergency department (ED) syndromic data to ascertain best syndrome classifications for ILI.

To determine whether preprocessing chief complaints before automatically classifying them into syndromic categories improves classification performance.