The system accepts as input the HIV-1 genotype (mandatory) as a list of mutations or as a whole sequence and any of the following information
when available: patient age and sex, route of infection, baseline viral load and CD4 cell count, the number of previous treatment lines, and binary indicators of previous use of the individual NRTIs, NNRTIs and PIs. These optional Thiazovivin purchase input variables have been shown to increase the accuracy of at least one of the three individual engines. For the EuResist system vs. expert (EVE) comparison, 12 top-level international HIV-1 drug resistance experts were invited to take part in the study, and enrolment was closed when the first 10 declared their availability. Experts were recruited among scientists with highly documented activity in the selleck field based on long-standing and relevant visibility as authors of peer-reviewed articles and presentations at HIV-specific international conferences. All of the EuResist data come from patients treated in Europe. Six of the experts contacted were chosen from Europe and, in order
to determine whether working in a different region with possibly different drug prescription attitudes could have an impact on predicting treatment outcome for European patient cases, six experts from non-European countries were invited to participate. The 10 experts composing the final panel are listed as coauthors of the study (C.A.B, F.B.-V., P.R.H., L.M., M.O., C.F.P., P.P., D.P, R.W.S. and A.-M.V.). A total of 25 TCEs were randomly extracted from a subset of the EIDB validation data set (i.e. the cases were excluded from training the EuResist system) for which the treatment regimen consisted of exactly three drugs (a ritonavir-boosted PI being considered a single drug), the baseline viral load was at least 10 000 copies/mL and the baseline
genotype included at least one major resistance mutation according to the contemporary International AIDS Urease Society (IAS) definition [2]. The TCEs were provided via an online interactive questionnaire that could be partially filled in and saved for later completion. Each of the experts received a private username and password that could be used to view and fill in the questionnaire anonymously. Only European or non-European origin was retained by the system; the identities of the individual experts could not be determined. Upon completing and closing the questionnaire, the expert was given a result page where she/he could see her/his own choices together with the actual outcomes and the EuResist predictions.