A reinforcement learning model to inform optimal decision paths for HIV elimination

The ‘Ending the HIV Epidemic (EHE)’ national plan aims to reduce annual HIV incidence in the United States from 38,000 in 2015 to 9,300 by 2025 and 3,300 by 2030. Diagnosis and treatment are two most effective interventions, and thus, identifying corresponding optimal combinations of testing and retention-in-care rates would help inform implementation of relevant programs. Considering the dynamic and stochastic complexity of the disease and the time dynamics of decision-making, solving for optimal combinations using commonly used methods of parametric optimization or exhaustive evaluation of pre-selected options are infeasible. Reinforcement learning (RL), an artificial intelligence method, is ideal; however, training RL algorithms and ensuring convergence to optimality are computationally challenging for large-scale stochastic problems. We evaluate its feasibility in the context of the EHE goal. We trained an RL algorithm to identify a ‘sequence’ of combinations of HIV-testing and retention-in-care rates at 5-year intervals over 2015–2070 that optimally leads towards HIV elimination. We defined optimality as a sequence that maximizes quality-adjusted-life-years lived and minimizes HIV-testing and care-and-treatment costs. We show that solving for testing and retention-in-care rates through appropriate reformulation using proxy decision-metrics overcomes the computational challenges of RL. We used a stochastic agent-based simulation to train the RL algorithm. As there is variability in support-programs needed to address barriers to care-access, we evaluated the sensitivity of optimal decisions to three cost-functions. The model suggests to scale-up retention-in-care programs to achieve and maintain high annual retention-rates while initiating with a high testing-frequency but relaxing it over a 10-year period as incidence decreases. Results were mainly robust to the uncertainty in costs. However, testing and retention-in-care alone did not achieve the 2030 EHE targets, suggesting the need for additional interventions. The results from the model demonstrated convergence. RL is suitable for evaluating phased public health decisions for infectious disease control.

[1]  Norma Terrin,et al.  Poverty, Hunger, Education, and Residential Status Impact Survival in HIV , 2011, AIDS and Behavior.

[2]  R. Scholz,et al.  Theoretical Biology and Medical Modelling Models of Epidemics: When Contact Repetition and Clustering Should Be Included , 2022 .

[3]  D. Hutton,et al.  Optimization Models for HIV/AIDS Resource Allocation: A Systematic Review. , 2020, Value in health : the journal of the International Society for Pharmacoeconomics and Outcomes Research.

[4]  Jennifer A. Pellowski,et al.  A pandemic of the poor: social disadvantage and the U.S. HIV epidemic. , 2013, The American psychologist.

[5]  Catherine Kreatsoulas,et al.  Machine learning in social epidemiology: Learning from experience , 2018, SSM - population health.

[6]  Xinyu Zhang,et al.  Multi-Objective Reinforcement Learning for Infectious Disease Control with Application to COVID-19 Spread , 2020, ArXiv.

[7]  M. Irfan Uddin,et al.  Optimal policy learning for COVID-19 prevention using reinforcement learning , 2020, J. Inf. Sci..

[8]  L. Kumaranayake.  The economics of scaling up: cost estimation for HIV/AIDS interventions , 2008, AIDS.

[9]  Abhijit Gosavi,et al.  Simulation-Based Optimization: Parametric Optimization Techniques and Reinforcement Learning , 2003 .

[10]  Harshad Khadilkar,et al.  Optimising Lockdown Policies for Epidemic Control using Reinforcement Learning , 2020, Transactions of the Indian National Academy of Engineering.

[11]  G. Marks,et al.  Estimating the Cost of Increasing Retention in Care for HIV-Infected Patients: Results of the CDC/HRSA Retention in Care Trial , 2015, Journal of acquired immune deficiency syndromes.

[12]  A. Lansky,et al.  Recommendations for HIV Screening of Gay, Bisexual, and Other Men Who Have Sex with Men — United States, 2017 , 2017, MMWR. Morbidity and mortality weekly report.

[13]  Margaret L Brandeau,et al.  Optimal investment in HIV prevention programs: more is not always better , 2009, Health care management science.

[14]  Atlanta,et al.  Revised recommendations for HIV testing of adults, adolescents, and pregnant women in health-care settings. , 2006, MMWR. Recommendations and reports : Morbidity and mortality weekly report. Recommendations and reports.

[15]  Emine Yaylali,et al.  Impact of Improved HIV Care and Treatment on PrEP Effectiveness in the United States, 2016–2020 , 2018, Journal of acquired immune deficiency syndromes.

[16]  L. Kumaranayake,et al.  Cost Effectiveness and Resource Allocation , 2018 .

[17]  Carlos Del Rio,et al.  The spectrum of engagement in HIV care and its relevance to test-and-treat strategies for prevention of HIV infection. , 2011, Clinical infectious diseases : an official publication of the Infectious Diseases Society of America.

[18]  Chaitra Gopalappa,et al.  Progression and Transmission of HIV/AIDS (PATH 2.0) , 2017, Medical decision making : an international journal of the Society for Medical Decision Making.

[19]  A. Chandra,et al.  HIV risk-related behaviors in the United States household population aged 15-44 years: data from the National Survey of Family Growth, 2002 and 2006-2010. , 2012, National health statistics reports.

[20]  S. Sansom,et al.  Cost Effectiveness of the National HIV/AIDS Strategy Goal of Increasing Linkage to Care for HIV-Infected Persons , 2012, Journal of acquired immune deficiency syndromes.

[21]  Andrew J. Schaefer,et al.  The Optimal Time to Initiate HIV Therapy Under Ordered Health States , 2008, Oper. Res..

[22]  Brian T. Denton,et al.  Optimizing the simultaneous management of blood pressure and cholesterol for type 2 diabetes patients , 2014, Eur. J. Oper. Res..

[23]  S. Parbhoo.  A Reinforcement Learning Design for HIV Clinical Trials , 2014 .

[24]  M. Jabbour,et al.  Chromatid recommensuration after segmental duplication , 2009, Theoretical Biology and Medical Modelling.

[25]  Jiming Liu,et al.  Incorporating causal factors into reinforcement learning for dynamic treatment regimes in HIV , 2019, BMC Medical Informatics and Decision Making.

[26]  A. Lansky,et al.  Estimating the Number of Heterosexual Persons in the United States to Calculate National Rates of HIV Infection , 2015, PLoS ONE.

[27]  Oluwole Daniel Makinde,et al.  Impact of optimal control on the treatment of HIV/AIDS and screening of unaware infectives , 2013 .

[28]  Angela B. Hutchinson,et al.  Cost-Effectiveness of Finding New HIV Diagnoses Using Rapid HIV Testing in Community-Based Organizations , 2008, Public health reports.

[29]  Max Buot,et al.  Beyond Race and Place: Distal Sociological Determinants of HIV Disparities , 2014, PloS one.

[30]  Peter Stone,et al.  Reinforcement Learning for Optimization of COVID-19 Mitigation policies , 2020, ArXiv.

[31]  David R. Holtgrave,et al.  U.S. HIV Incidence and Transmission Goals, 2020 and 2025. , 2017, American journal of preventive medicine.

[32]  Emily L. Tucker,et al.  Effects of Reaching National Goals on HIV Incidence, by Race and Ethnicity, in the United States , 2017, Journal of public health management and practice : JPHMP.

[33]  U. Netlogo Wilensky,et al.  Center for Connected Learning and Computer-Based Modeling , 1999 .

[34]  Richard S. Sutton,et al.  Reinforcement Learning: An Introduction , 1998, IEEE Trans. Neural Networks.

[35]  Yi Zhou,et al.  Epileptic foci localization based on mapping the synchronization of dynamic brain network , 2019, BMC Medical Informatics and Decision Making.

[36]  Mauricio Arango,et al.  COVID-19 Pandemic Cyclic Lockdown Optimization Using Reinforcement Learning , 2020, ArXiv.

[37]  S. Sansom,et al.  Vital Signs: HIV Transmission Along the Continuum of Care — United States, 2016 , 2019, MMWR. Morbidity and mortality weekly report.

[38]  Julio S. G. Montaner,et al.  Optimizing an HIV testing program using a system dynamics model of the continuum of care , 2015, Health Care Management Science.

[39]  R. Shrestha,et al.  Cost Effectiveness of HIV Prevention Interventions in the U.S. , 2016, American journal of preventive medicine.

[40]  Tamer Khattab,et al.  Reinforcement learning-based decision support system for COVID-19 , 2021, Biomedical Signal Processing and Control.

[41]  Srivatsan Srinivasan,et al.  Evaluating Reinforcement Learning Algorithms in Observational Health Settings , 2018, ArXiv.

[42]  M. Friend,et al.  Estimated HIV incidence and prevalence in the United States 2010–2015 , 2018 .

[43]  L. Koenig,et al.  Vital Signs: Human Immunodeficiency Virus Testing and Diagnosis Delays — United States , 2017, MMWR. Morbidity and mortality weekly report.