AI for IITRx in HIV

Jhpiego Corporation

AI for IITRx in HIV

Upskilling frontline case managers with predictive AI to deliver high-quality, risk-stratified HIV care and prevent treatment interruptions at scale

Abuja, Federal Capital Territory, Nigeria

NGA

Other, including part of a larger organization (please explain below)

The solution team is a part of a program run by the Jhpiego Corporation, a global not for profit, affiliated by the Johns Hopkins University.

Interruption in Treatment (IIT) among HIV patients on Antiretroviral Therapy (ART) is a major barrier to epidemic control. Globally, nearly 40 million individuals are on ART, and IIT contributes significantly to their morbidity and mortality. In Nigeria, an estimated 25% of the 1.6 million individuals on ART are at risk of disengaging from care. Globally the self-reported rate of non-adherence varies from 10% to 62% Traditional risk assessment methods—reliant on retrospective data reviews, static demographic flags, and subjective clinical judgment—result in delayed, reactive interventions that often come too late to prevent treatment interruptions. An individual’s risk of IIT is not static, it evolves over time. However, frontline health workers lack the tools that factor in this complexity. There is an expressed need for solutions that empower frontline health workers to deliver timely, risk-stratified care, allocating resources to those who are most at risk. Such a solution must be integrated into routine service delivery workflows in real time while remaining scalable and affordable. What is needed is a paradigm shift from reactive case management to proactive, patient-centered HIV care that is both adaptive to patient needs and sustainable within public health systems.

Jhpiego alongside our AI partners Palindrome Data, operationalized the Reducing Interrupted Rx in HIV Care solution through a multi-phase initiative, culminating in an artificial intelligence model embedded within the Case Management App used by frontline health workers in Nigeria. The solution leverages Machine Learning algorithms to predict the risk of treatment interruption for HIV patients, using routinely collected health and socio-demographic information at the point of care. It generates a personalized risk score and alerts for the case managers, categorizing patients as high, medium or low risk. Case managers subsequently receive tailored risk-stratified interventions to choose from, enabling timely and targeted care. The system does not replace human decision-making; it enhances capacity of the case managers, helping them prioritize patients in need of support. A Pilot across 49 sites in 2 states of Nigeria, for over 86,000 patients, demonstrated that case managers at intervention sites spent 76% more time with high-risk patients, resulting in a 4x reduction in treatment interruptions in comparison to control sites. With support from government stakeholders and the Patrick J McGovern Foundation, the solution is now scaling up across 3 states of Nigeria (Kwara, Taraba and Niger, serving over 135,000 patients across 85 ART sites).

The Reducing Interrupted Rx in HIV Care solution serves three key stakeholders: A. Frontline HIV Case Managers: Often overburdened, and operating with limited resources, they struggle to prioritize patients effectively. Our solution equips them with personalized risk assessments, enabling proactive targeted care for those most at risk. B. People living with HIV: Many face stigma, financial hardship and fragmented access to healthcare. The solution helps improve treatment adherence, reducing morbidity and enhancing the overall quality of life through better resultant viral suppression. C. Government Health Agencies: They design programs and strategies at scale, often lacking data-driven insights and optimal resources. The solution has the potential to reduce costs associated with IIT-related complications, allowing for more efficient resource allocation. The solution is currently being scaled to 85 ART sites across 3 Nigerian states, covering 135,000 HIV patients, served by 200 case managers. Expected outcomes include • Reduced rates of treatment interruption • Lower HIV transmission rates, and reduced risk of drug-resistant HIV strains in the community • Reduce burden of work for the case managers • A replicable scalable model for AI-driven HIV care that can inform regional and global scale-up in other high burden low-resource settings.