The present study aimed to develop a sensitive dengue diagnostic clinical algorithm using the resources that are available at the health care institutions, clinical expertise, and identification of signs and symptoms and variables of the hemogram associated with dengue. The performance of the potential algorithms were assessed under routine care. A single group cuasiexperimental study with a Bayesian adaptive design was conducted. Prior and cumulative information was used during intermediate analysis of clinical algorithms performance and according to these results algorithms were modified and then implemented in the subsequent study subjects. Men and women of all ages, seeking care due to fever of less than 15 days were recruited in 3 endemic areas in Colombia. The algorithms were applied by study physicians and blood samples taken for hemogram and dengue reference tests. It was planned that algorithms with high sensitivity (95%) and specificity (80%) were implemented in real life to assess their impact on patients outcomes alone and in combination with dengue rapid diagnostic tests but this was not feasible as not algorithms achieved the targetted performance. The experience of physicians using the algorithm in a mobile device was going to be explored with qualitative methods but this was explored with study physicians.

Inclusion Criteria:

• Fever of less than 15 days

Exclusion Criteria:

• Origin of fever identified

Patient unidentifiable data and detailed algorithms available upon request

Upon request for 10 years

Samples were coded and laboratory personnel were unaware of algorithm results, the participant did not know what algorithm prototype was tested.