Team 6 Quantum grey


Dashboard :

Live recording :

Predictive Models for VL Suppression

Seasonality is impacting the incidence of several diseases. For HIV in particular, immunoresponsiveness might alter in function of the season. But aside from this intrinsic factor, several exogenous factors with seasonal patterns are believed to impact HIV incidence as well. These variables can include socio-demographic factors such as seasonal patterns in population mobility, food availability or alterations in sexual activity. In our project, we studied trends in viral load suppression. To do this, data from Kenya for the period 2015-2017 for the 47 different counties was used.  Different machine learning algorithms were trained to predict the suppression levels on basis of the respective historical data. In addition, historical patterns were subdivided and the inter-variable cross predictive capability was used to further enhance these models. Also -where available- data for longitudinal exogenous variables were also included in the models. The relative scarcity of these additional data at the right level of granularity and for the right historical period was found to be a major challenge

We believe that the methodology applied can be deployed to obtain a higher level of accuracy to predict HIV incidence at sub-county level. Additional effort is needed to obtain longitudinal exogenous variables in a consistent process and with the right level of granularity.