Dear RDI subscriber,
We are writing to you today to tell you about the latest advance with the HIV Treatment Response Predictions system - HIV-TRePS.
HIV-TRePS optimised for use on mobile devices
The system has been re-designed and programmed to work on mobile devices such as mobile phones and tablets and at super-fast speed. Users simply log-in via their usual browser and the user interface automatically resizes for their particular mobile device. This makes the system much more convenient for use 'on the move'.HIV-TRePS optimised speed
We have migrated the system to new servers and streamlined the code so that HIV-TRePS now runs substantially faster than ever before. Typically a case that used to take up to a minute to process before the report was delivered now takes around 20 seconds, over broadband.New models
We have developed new models that predict treatment response without a genotype. The models were trained to predict the probability of virological response to therapy following virological failure using 50,270 treatment change episodes (TCEs) without a genotype, from around the world. The models were tested with an independent global test set (n=3,000). For the first time the models allow the use of baseline data with wider 'windows' - up to to 24 weeks for the treatment change for the baseline CD4 count and up to 16 weeks for the vial load. The models also included for the first time tipranavir, maraviroc and elvitegravir.The models achieved an area under the ROC curve (AUC) value of 0.83 during cross-validation and 0.81 during independent testing. Using the optimum cut-off for predictions of response and failure determined during model development, they achieved sensitivity of 73% and specificity of 76% and overall accuracy of 75% in testing.
The new models achieved similar and not significantly different performance to that of models trained with a subset of the same data using the older, tighter baseline data windows of 12 and 8 weeks.
Sign In or Register now at: https://www.hivrdi.org/treps
The RDI team.
Date published: 13th April 2017