Machine learning models reveal neurocognitive impairment type and prevalence are associated with distinct variables in HIV/AIDS.

摘要: Neurocognitive impairment (NCI) among HIV-infected patients is heterogeneous in its reported presentations and frequencies. To determine the prevalence of NCI associated subtypes as well predictive variables, we investigated with HIV/AIDS receiving universal health care. Recruited adult subjects underwent a neuropsychological (NP) test battery established normative (sex-, age-, education-matched) values together assessment their demographic clinical variables. Three patient groups were identified including neurocognitively normal (NN, n = 246), HIV-associated neurocognitive disorders (HAND, n = 78), impaired-other (NCI-OD, n = 46). Univariate, multiple logistic regression machine learning analyses applied. Univariate showed variables differed significantly between birth continent, quality life, substance use, PHQ-9. Multiple models revealed again for PHQ-9 score, VACS index, head injury. Random forest (RF) disclosed that classification algorithms distinguished HAND from NN NCI-OD area under curve (AUC) 0.87 0.77, respectively. Relative importance plots derived RF model exhibited distinct variable rankings status both versus comparisons. Thus, was frequently detected (33.5%) although (21%) lower than several earlier reports underscoring potential contribution other factors to NCI. Machine uncovered related individual types not by univariate or analyses, highlighting value approaches understanding HIV/AIDS.