Dengue fever is a seasonal, vector borne disease which is often deadly. At present, dengue is usually diagnosed through two stages of tests in India. A patient showing the physical symptoms of dengue is first subjected to a screening test, the CBC test. The second test, the dengue serology test, is the truly confirmatory test, but may take up to 10 days to return a correct reading. We propose a simple neural network based model which can detect whether the patient has dengue, with just the preliminary CBC test report’s data. Patient data was collected from a single hospital located in Ghaziabad, India. We found that the system correctly classified the unseen test cases with a significant degree of accuracy. We further propose as future research directions the application and comparison of the modelling results of more pattern recognition techniques for this classification task, testing of the system in real-time hospital conditions, and the inclusion of locality specific factors to build as general and as widely-reproducible a model as possible.