Strategies of Managing Repeated Measures: Using Synthetic Random Forest to Predict HIV Viral Suppression Status Among Hospitalized Persons with HIV

• Type of Material: Article
• Authors: Jingxin Liu, University of Miami; Yue Pan, University of Miami; Mindy C. Nelson, University of Miami; Lauren K. Gooden, Columbia University; Lisa R. Metsch, Columbia University; Allan E. Rodriguez, University of Miami; Susan Tross, Columbia University; Carlos del rio, Emory University; Raul N. Mandler, National Institutes of Health; Daniel J. Feaster, University of Miami
• Language: English
• Date: 2023-02-05
• Publisher: Springer Nature
• Publication Version: Author Accepted Manuscript (After Peer Review)
• Copyright Statement: © 2023, The Author(s), under exclusive licence to Springer Science Business Media, LLC, part of Springer Nature
• Final Published Version (URL): https://doi.org/10.1007/s10461-023-04015-1
• Title of Journal or Parent Work: AIDS and Behavior
• Volume: 27
• Issue: 9
• Start Page: 2915
• End Page: 2931
• Grant/Funding Agency: Center for AIDS Research; National Institute on Drug Abuse; University of Miami
• Grant/Funding Information: Funding for this study and by the National Institute on Drug Abuse under the following awards: UG1DA013720, UG1DA013035, UG1DA013034, UG1DA013727, UG1DA020024, UG1DA013732, UG1DA015831, UG1DA015815, and U10DA020036. Support from the University of Miami Center for HIV and Research on Mental Health (CHARM) (P30MH116867) and Center for AIDS Research (CFAR) (P30AI07396) is also acknowledged.
• Keywords: Longitudinal measurement; HIV/AIDS; Predictive model; Machine learning
• Subject - Topics: HIV infections; Machine learning; Public health

Abstract

The HIV/AIDS epidemic remains a major public health concern since the 1980s; untreated HIV infection has numerous consequences on quality of life. To optimize patients’ health outcomes and to reduce HIV transmission, this study focused on vulnerable populations of people living with HIV (PLWH) and compared different predictive strategies for viral suppression using longitudinal or repeated measures. The four methods of predicting viral suppression are (1) including the repeated measures of each feature as predictors, (2) utilizing only the initial (baseline) value of the feature as predictor, (3) using the last observed value as the predictors and (4) using a growth curve estimated from the features to create individual-specific prediction of growth curves as features. This study suggested the individual-specific prediction of the growth curve performed the best in terms of lowest error rate on an independent set of test data.