[Probabilistic spatial-temporal prediction of total and severe epidemic of dengue in Colombia].

Objective: To establish a new predictive methodology to determine the proportion of severe dengue with respect to the annual total of dengue infections per department based on the probability theory.
Materials and Methods: Based on annual data on the number of infected persons by department in the period 2005-2010, the proportion of cases of severe dengue was calculated with respect to the total for each year. Probability spaces were constructed to evaluate these events in the ranges 0.5 and 0.3. Sets of ranges were determined and probability, mean square deviation and the difference between them were estimated. A prediction of the range of infected people for 2011 was made using the arithmetic average of the values of the last two years.
Results: The range in which the proportion of the number of people infected with severe dengue is included with respect to the total amount in each department was correctly predicted, with an effectiveness of 93.3% for the 0.5 range and 86.7% for the 0.3 range.
Conclusion: A mathematical spatial-temporal self-organization was found in the proportion of severe dengue with respect to the total, which allows establishing useful predictions for decision-making in public health.