Asian Journal of Current Research

Nipah virus infection is a high-consequence zoonosis of South and South-East Asia for which laboratory confirmation rests on two assay families whose informativeness moves in opposite directions across the course of illness. Reverse transcription quantitative polymerase chain reaction performs best while viral nucleic acid remains detectable in blood and the oropharynx, whereas enzyme-linked immunosorbent assays for immunoglobulin M and immunoglobulin G become informative as nucleic acid disappears. Prevailing practice interprets these assays as independent dichotomous results obtained at a single moment, discarding quantitative magnitude, specimen provenance and the day of illness on which each specimen was taken. This review critically evaluates whether the longitudinal structure of the two assay families can be exploited jointly, and what explainable artificial intelligence would need to deliver before such a model could be used in outbreak conditions. Human clinical studies describing the temporal behaviour of molecular and serological markers were appraised alongside the methodological literature on diagnostic accuracy without an adequate reference standard, and the computational literature on temporal and multimodal fusion and on explanation methods in clinical decision support. The evidence consistently indicates an ordered transition in which clearance of viral nucleic acid from blood precedes the appearance of virus-specific immunoglobulin G, and that the combination of detectable viraemia with absent immunoglobulin G at presentation carries adverse prognostic weight. Confidence in these observations is nevertheless constrained by very small samples, over-representation of the minority who survive, the absence of an international nucleic-acid standard, and geographical concentration of evidence in Bangladesh and Kerala. No clinical prediction model that fuses molecular and serological results over time was identified for this infection. A conceptual dynamic framework is developed that separates established temporal relationships from hypothesised gains, specifies the evaluation and governance requirements that would apply, and identifies prospective serial-sampling cohorts as the rate-limiting evidential step rather than model architecture.