UC San Diego

Health departments are using HIV data to monitor HIV growth in real time. The main purpose of this monitoring is to come up with policies for efficient allocation of medical resources. In order to achieve the efficient medical resources allocation, a method should be established for predicting where future transmissions of HIV will occur using the partial information of the transmission history. Validity of these predictions are of paramount importance as it affects the policy for allocation of medical resources. Indeed, the more accurate the prediction is, the more efficiently preventive care or other resources can be allocated to the network.

The focus of this work is on community-level monitoring of HIV spread prevention. We have modeled the sexual network as communities of individuals and proposed community-level methods for prediction. Then, we have compared predictive power of the proposed methods in different settings of the network.