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.
The Real-time Outbreak and Disease Surveillance system collects chief complaints as free text and uses a naïve Bayesian classifier called CoCo to classify the complaints into syndromic categories. CoCo 3.0 has been trained on 28,990 manually clas-sified chief complaints. The free text chief com-plaints are challenging to work with, due to problems caused by linguistic variations such as synonyms, abbreviations, acronyms, truncations, concatenations, misspellings and typographic errors. Failure to correct these word variations may result in missed cases, thereby decreasing sensitivity of detection.
Objective
To determine whether preprocessing chief complaints before automatically classifying them into syndromic categories improves classification performance.
We will convene a consultative meeting on chief complaint classifiers and standardized syndromic definitions in Pittsburgh, PA, from September 24-25,
To determine whether preprocessing chief complaints before automatically classifying them into syndromic categories improves classification performance.
Syndromic surveillance systems often classify patients into syndromic categories based on emergency department (ED) chief complaints. There exists no standard set of syndromes for syndromic surveillance, and the available syndromic case definitions demonstrate substantial heterogeneity of findings constituting the definition. The use of fever in the definition of syndromic categories is arbitrary and unsystematic. We determined whether chief complaints accurately represent whether a patient has any of five febrile syndromes: febrile respiratory, febrile gastrointestinal, febrile rash, febrile neurological, or febrile hemorrhagic.
Automated syndromic surveillance systems often classify patients into syndromic categories based on free-text chief complaints. Chief complaints (CC) demonstrate low to moderate sensitivity in identifying syndromic cases. Emergency Department (ED) reports promise more detailed clinical information that may increase sensitivity of detection. Objective: Compare classification of patients based on chief complaints against classification from clinical data described in ED reports for identifying patients with an acute lower respiratory syndrome.
Twelve years into the 21st century, after publication of hundreds of articles and establishment of numerous biosurveillance systems worldwide, there is no agreement among the disease surveillance community on most effective technical methods for public health data monitoring. Potential utility of such methods includes timely anomaly detection, threat corroboration and characterization, follow-up analysis such as case linkage and contact tracing, and alternative uses such as providing supplementary information to clinicians and policy makers. Several factors have impeded establishment of analytical conventions. As immediate owners of the surveillance problem, public health practitioners are overwhelmed and understaffed. Goals and resources differ widely among monitoring institutions, and they do not speak with a single voice. Limited funding opportunities have not been sufficient for cross-disciplinary collaboration driven by these practitioners. Most academics with the expertise and luxury of method development cannot access surveillance data. Lack of data access is a formidable obstacle to developers and has caused talented statisticians, data miners, and other analysts to abandon the field. The result is that older research is neglected and repeated, literature is flooded with papers of varying utility, and the decision-maker seeking realistic solutions without detailed technical knowledge faces a difficult task. Regarding conventions, the disease surveillance community can learn from older, more established disciplines, but it also poses some unique challenges. The general problem is that disease surveillance lies on the fringe of disparate fields (biostatistics, statistical process control, data mining, and others), and poses problems that do not adequately fit conventional approaches in these disciplines. In its eighth year, the International Society of Disease Surveillance is well positioned to address the standardization problem because its membership represents the involved stakeholders including progressive programs worldwide as well as resource-limited settings, and also because best practices in disease surveillance is fundamental to its mission. The proposed panel is intended to discuss how an effective, sustainable technical conventions group might be maintained and how it could support stakeholder institutions.
Objective
The panel will present the problem of standardizing analytic methods for public health disease surveillance, enumerate goals and constraints of various stakeholders, and present a straw-man framework for a conventions group.
Recently, a growing number of studies have made use of Twitter to track the spread of infectious disease. These investigations show that there are reliable spikes in traffic related to keywords associated with the spread of infectious diseases like Influenza [1], as well as other Syndromes [2]. However, little research has been done using Social Media to monitor chronic conditions like Asthma, which do not spread from sufferer to sufferer. We therefore test the feasibility of using Twitter for Asthma surveillance, using techniques from NLP and machine learning to achieve a deeper understanding of what users Tweet about Asthma, rather than relying only on keyword search.
Objective
We present a Content Analysis project using Natural Language Processing to aid in Twitter-based syndromic surveillance of Asthma
Presenter
Wendy Chapman, PhD, Associate Professor, Division of Biomedical Informatics, UCSD School of Medicine
Date
Thursday, November 18, 2010
Host
ISDS Research Committee
Characterizing mentions found in clinical texts that support, refute, or represent uncertainty for suspected pneumonia is one area where automated Natural Language Processing (NLP) screening algorithms could be improved. Mentions of uncertainty and negation commonly occur in clinical texts, and opportunities exist to extend existing algorithms [1] and taxonomies [2]. In general there are three main sources of uncertainty found in healthcare: 1) probability or risk; 2) ambiguity – lack of reliability, credibility or adequacy of the information; and, 3) complexity – aspects of the phenomenon that make it difficult to comprehend [3].
Objective
We sought to identify relevant evidence that supports, refutes or contributes uncertainty when reviewing cases of suspected pneumonia and characterize their interaction with uncertainty phenomena found in clinical texts.