AI shows promise for early diagnosis of mosquito-borne viral infections, review finds

Artificial intelligence and machine learning models show encouraging diagnostic performance for the early detection of dengue, chikungunya, and Zika, according to a systematic review and meta-analysis published in PLOS Digital Health.

The pooled estimates were high across the reported metrics. The random-effects pooled area under the receiver operating characteristic curve was 0.954 (95% CI 0.936–0.968), pooled accuracy was 0.904 (95% CI 0.888–0.918), positive predictive value was 0.903 (95% CI 0.876–0.925), and negative predictive value was 0.882 (95% CI 0.859–0.902). A bivariate random-effects model yielded pooled sensitivity of 0.861 (95% CI 0.809–0.900) and specificity of 0.902 (95% CI 0.871–0.927). Heterogeneity was substantial for the univariate outcomes, with I² values ranging from 95.8% to 97.7%.

The authors stress that these pooled figures should be read as descriptive summaries of a highly heterogeneous evidence base rather than as definitive estimates of real-world performance. Most of the included studies were judged to be at high overall risk of bias when assessed with the Prediction model Risk Of Bias Assessment Tool (PROBAST), and calibration and external validation were uncommon across the literature. The review included only model-development or validation studies reporting diagnostic performance, and the estimates therefore reflect the published modelling literature rather than prospective clinical deployment.

The evidence base was also geographically and pathogen-wise narrow. Most eligible studies came from Asia and overwhelmingly focused on dengue, and no eligible studies evaluated West Nile virus, yellow fever, or Rift Valley fever. Early diagnosis of mosquito-borne viral infections remains challenging because initial symptoms are often nonspecific and confirmatory testing may be delayed or unavailable, which is the clinical gap these models aim to address. The authors conclude that prospective, externally validated, well-calibrated, and geographically diverse studies are needed before routine clinical implementation of such artificial intelligence tools can be recommended.

Subject of Research: Diagnostic performance of artificial intelligence models for early diagnosis of mosquito-borne viral infections

Article Title: Diagnostic performance of artificial intelligence in early diagnosis of mosquito-borne viral infections: Systematic review and meta-analysis

Article References: Pennisi, F., Pinto, A., Cozzolino Cangiano, C., Rezza, G., Signorelli, C., Baldo, V., & Gianfredi, V. (2026). Diagnostic performance of artificial intelligence in early diagnosis of mosquito-borne viral infections: Systematic review and meta-analysis. PLOS Digital Health, 5(10), e0001796. https://doi.org/10.1371/journal.pdig.0001796

Image Credits: AI Generated

DOI: 10.1371/journal.pdig.0001796

Keywords: artificial intelligence, machine learning, dengue, chikungunya, Zika, systematic review, meta-analysis, diagnostic performance, mosquito-borne viruses, risk of bias, early diagnosis, Diagnostic