Sensitivity and Specificity of a Novel Classifier for the Early Diagnosis of Dengue

Abstract

BackgroundDengue is the commonest arboviral disease of humans. An early and accurate diagnosis of dengue can support clinical management, surveillance and disease control and is central to achieving the World Health Organisation target of a 50% reduction in dengue case mortality by 2020.Methods5729 children with fever of <72hrs duration were enrolled into this multicenter prospective study in southern Vietnam between 2010-2012. A composite of gold standard diagnostic tests identified 1692 dengue cases. Using statistical methods, a novel Early Dengue Classifier (EDC) was developed that used patient age, white blood cell count and platelet count to discriminate dengue cases from non-dengue cases.ResultsThe EDC had a sensitivity of 74.8% (95%CI: 73.0-76.8%) and specificity of 76.3% (95%CI: 75.2-77.6%) for the diagnosis of dengue. As an adjunctive test alongside NS1 rapid testing, sensitivity of the composite test was 91.6% (95%CI: 90.4-92.9%).ConclusionsWe demonstrate that the early diagnosis of dengue can be enhanced beyond the current standard of care using a simple evidence-based algorithm. The results should support patient management and clinical trials of specific therapies.

Evaluation of the importance of fever with respect to dengue prognosis according to the 2009 WHO classification: a retrospective study

“…Dengue is an infectious disease that exhibits acute evolution [1]. The incidence of dengue has increased across all age groups [2] to become a clear public health problem [3].…”

Subspace-Based Method to Improve Classification Accuracy of High-Dimensional Data

“…After combining useful features in the dataset, we applied classification algorithms to evaluate the accuracy of the proposed model. Table IV also shows that, relative to sensitivity, the proposed model outperforms a previous method (74.48 %) in [5]. The better sensitivity indicates that HSC model can find the patient more exactly.…”

“…where the m patients will be determined by the values of vector h  . Assume that we can find the group of patients who truly have dengue, determining the patients have disease based on (6) provides better result than the (5). Thus, the performance of diagnosis will be improved.…”

“…For example, Tanner et al [4] used the data of 1200 patients within 72 hours of fever onset to distinguish patients Manuscript with dengue illness using a decision tree algorithm. Another study [5] conducted an experiment using the data of 5726 children with fever (less than 72 hours from onset). They applied statistic methods and obtained better sensitivity accuracy than traditional methods, like nonstructural protein 1(NS1) or nonstructural protein antigen (NS1 Ag) strip test.…”

Chikungunya and Zika Virus Cases Detected against a Backdrop of Endemic Dengue Transmission in Vietnam

“…Study participants were enrolled at the outpatient clinics of seven hospitals in southern Vietnam between October 2010 and December 2014. The detailed inclusion and exclusion criteria and study procedures are described elsewhere 12 . Reverse transcription polymerase chain reaction (RT-PCR) tests for ZIKV 13 and CHIKV 14 were performed on stored frozen plasma samples from study participants that were tested negative in a dengue virus (DENV) RT-PCR assay that has been described extensively elsewhere 15 .…”