Laura B Balzer1, Diane V Havlir2, Moses R Kamya3,4, Gabriel Chamie2, Edwin D Charlebois5, Tamara D Clark2, Catherine A Koss2, Dalsone Kwarisiima3, James Ayieko6, Norton Sang6, Jane Kabami3, Mucunguzi Atukunda3, Vivek Jain2, Carol S Camlin7, Craig R Cohen7, Elizabeth A Bukusi6,7, Mark Van Der Laan8, Maya L Petersen8. 1. Department of Biostatistics and Epidemiology, University of Massachusetts, Amherst, Massachusetts, USA. 2. Division of HIV, Infectious Diseases, and Global Medicine, Department of Medicine, University of California, San Francisco, California, USA. 3. Infectious Diseases Research Collaboration, Kampala, Uganda. 4. School of Medicine, Makerere University College of Health Sciences, Kampala, Uganda. 5. Division of Prevention Science, Department of Medicine, University of California, San Francisco, California, USA. 6. Kenya Medical Research Institute, Nairobi, Kenya. 7. Department of Obstetrics, Gynecology, and Reproductive Sciences, University of California, San Francisco, California, USA. 8. Division of Epidemiology and Biostatistics, University of California, Berkeley, California, USA.
Abstract
BACKGROUND: In generalized epidemic settings, strategies are needed to prioritize individuals at higher risk of human immunodeficiency virus (HIV) acquisition for prevention services. We used population-level HIV testing data from rural Kenya and Uganda to construct HIV risk scores and assessed their ability to identify seroconversions. METHODS: During 2013-2017, >75% of residents in 16 communities in the SEARCH study were tested annually for HIV. In this population, we evaluated 3 strategies for using demographic factors to predict the 1-year risk of HIV seroconversion: membership in ≥1 known "risk group" (eg, having a spouse living with HIV), a "model-based" risk score constructed with logistic regression, and a "machine learning" risk score constructed with the Super Learner algorithm. We hypothesized machine learning would identify high-risk individuals more efficiently (fewer persons targeted for a fixed sensitivity) and with higher sensitivity (for a fixed number targeted) than either other approach. RESULTS: A total of 75 558 persons contributed 166 723 person-years of follow-up; 519 seroconverted. Machine learning improved efficiency. To achieve a fixed sensitivity of 50%, the risk-group strategy targeted 42% of the population, the model-based strategy targeted 27%, and machine learning targeted 18%. Machine learning also improved sensitivity. With an upper limit of 45% targeted, the risk-group strategy correctly classified 58% of seroconversions, the model-based strategy 68%, and machine learning 78%. CONCLUSIONS: Machine learning improved classification of individuals at risk of HIV acquisition compared with a model-based approach or reliance on known risk groups and could inform targeting of prevention strategies in generalized epidemic settings. CLINICAL TRIALS REGISTRATION: NCT01864603.
BACKGROUND: In generalized epidemic settings, strategies are needed to prioritize individuals at higher risk of human immunodeficiency virus (HIV) acquisition for prevention services. We used population-level HIV testing data from rural Kenya and Uganda to construct HIV risk scores and assessed their ability to identify seroconversions. METHODS: During 2013-2017, >75% of residents in 16 communities in the SEARCH study were tested annually for HIV. In this population, we evaluated 3 strategies for using demographic factors to predict the 1-year risk of HIV seroconversion: membership in ≥1 known "risk group" (eg, having a spouse living with HIV), a "model-based" risk score constructed with logistic regression, and a "machine learning" risk score constructed with the Super Learner algorithm. We hypothesized machine learning would identify high-risk individuals more efficiently (fewer persons targeted for a fixed sensitivity) and with higher sensitivity (for a fixed number targeted) than either other approach. RESULTS: A total of 75 558 persons contributed 166 723 person-years of follow-up; 519 seroconverted. Machine learning improved efficiency. To achieve a fixed sensitivity of 50%, the risk-group strategy targeted 42% of the population, the model-based strategy targeted 27%, and machine learning targeted 18%. Machine learning also improved sensitivity. With an upper limit of 45% targeted, the risk-group strategy correctly classified 58% of seroconversions, the model-based strategy 68%, and machine learning 78%. CONCLUSIONS: Machine learning improved classification of individuals at risk of HIV acquisition compared with a model-based approach or reliance on known risk groups and could inform targeting of prevention strategies in generalized epidemic settings. CLINICAL TRIALS REGISTRATION: NCT01864603.
Authors: Esther Kissling; Edward H Allison; Janet A Seeley; Steven Russell; Max Bachmann; Stanley D Musgrave; Simon Heck Journal: AIDS Date: 2005-11-18 Impact factor: 4.177
Authors: Julia L Marcus; Leo B Hurley; Douglas S Krakower; Stacey Alexeeff; Michael J Silverberg; Jonathan E Volk Journal: Lancet HIV Date: 2019-07-05 Impact factor: 12.767
Authors: Douglas S Krakower; Susan Gruber; Katherine Hsu; John T Menchaca; Judith C Maro; Benjamin A Kruskal; Ira B Wilson; Kenneth H Mayer; Michael Klompas Journal: Lancet HIV Date: 2019-07-05 Impact factor: 12.767
Authors: Catherine A Koss; James Ayieko; Florence Mwangwa; Asiphas Owaraganise; Dalsone Kwarisiima; Laura B Balzer; Albert Plenty; Norton Sang; Jane Kabami; Theodore D Ruel; Douglas Black; Carol S Camlin; Craig R Cohen; Elizabeth A Bukusi; Tamara D Clark; Edwin D Charlebois; Maya L Petersen; Moses R Kamya; Diane V Havlir Journal: Clin Infect Dis Date: 2018-11-28 Impact factor: 9.079
Authors: Gabriel Chamie; Tamara D Clark; Jane Kabami; Kevin Kadede; Emmanuel Ssemmondo; Rachel Steinfeld; Geoff Lavoy; Dalsone Kwarisiima; Norton Sang; Vivek Jain; Harsha Thirumurthy; Teri Liegler; Laura B Balzer; Maya L Petersen; Craig R Cohen; Elizabeth A Bukusi; Moses R Kamya; Diane V Havlir; Edwin D Charlebois Journal: Lancet HIV Date: 2016-01-26 Impact factor: 12.767
Authors: Susan Gruber; Douglas Krakower; John T Menchaca; Katherine Hsu; Rebecca Hawrusik; Judith C Maro; Noelle M Cocoros; Benjamin A Kruskal; Ira B Wilson; Kenneth H Mayer; Michael Klompas Journal: Stat Med Date: 2020-06-24 Impact factor: 2.373
Authors: J Richard Pilsner; Hachem Saddiki; Brian W Whitcomb; Alexander Suvorov; Germaine M Buck Louis; Sunni L Mumford; Enrique F Schisterman; Oladele A Oluwayiose; Laura B Balzer Journal: Hum Reprod Date: 2022-06-30 Impact factor: 6.353
Authors: Lisa P Spees; Kathleen E Wirth; Shreshth Mawandia; Semo Bazghina-Werq; Jenny H Ledikwe Journal: South Afr J HIV Med Date: 2020-12-14 Impact factor: 2.744
Authors: Katherine M Jia; Hallie Eilerts; Olanrewaju Edun; Kevin Lam; Adam Howes; Matthew L Thomas; Jeffrey W Eaton Journal: J Int AIDS Soc Date: 2022-01 Impact factor: 5.396
Authors: Catherine A Koss; Edwin D Charlebois; James Ayieko; Dalsone Kwarisiima; Jane Kabami; Laura B Balzer; Mucunguzi Atukunda; Florence Mwangwa; James Peng; Yusuf Mwinike; Asiphas Owaraganise; Gabriel Chamie; Vivek Jain; Norton Sang; Winter Olilo; Lillian B Brown; Carina Marquez; Kevin Zhang; Theodore D Ruel; Carol S Camlin; James F Rooney; Douglas Black; Tamara D Clark; Monica Gandhi; Craig R Cohen; Elizabeth A Bukusi; Maya L Petersen; Moses R Kamya; Diane V Havlir Journal: Lancet HIV Date: 2020-02-19 Impact factor: 12.767
Authors: Kathryn Peebles; Thesla Palanee-Phillips; Jennifer E Balkus; Ivana Beesham; Heeran Makkan; Jennifer Deese; Jennifer Smit; Renee Heffron; Charles S Morrison; Neena M Philip; Mookho Malahleha; Margaret Kasaro; Yuthika Naidoo; Tanya Nielson; Krishnaveni Reddy; Philip Kotze; Khatija Ahmed; Helen Rees; Jared M Baeten; Ruanne V Barnabas Journal: J Acquir Immune Defic Syndr Date: 2020-10-01 Impact factor: 3.771
Authors: Shaun Palmer; Maartje Dijkstra; Johannes Cf Ket; Elizabeth W Wahome; Jeffrey Walimbwa; Evanson Gichuru; Elise M van der Elst; Maarten F Schim van der Loeff; Godelieve J de Bree; Eduard J Sanders Journal: J Int AIDS Soc Date: 2020-10 Impact factor: 5.396