Use of artificial intelligence for health science in low- and middle-income countries: NIH portfolio landscape, gaps and opportunities

Abstract

OBJECTIVES: To analyse the landscape of active US National Institutes of Health (NIH) artificial intelligence (AI) health research grants, with emphasis on studies conducted in low- and middle-income countries (LMICs), to characterise use cases, health challenges addressed and gaps relevant to the ethical and responsible application of AI-enabled health science. DESIGN: Descriptive portfolio analysis of NIH-funded AI health research grants. SETTING: NIH research portfolio analysis, with a focus on global health studies in LMICs. PARTICIPANTS: None. Data are derived from active NIH-funded grants involving AI applications in health research, as of 31 January 2025. INTERVENTIONS: Not applicable (portfolio analysis). PRIMARY AND SECONDARY OUTCOME MEASURES: Primary measures included the proportion and funding of AI health research grants focused on LMICs and their thematic use cases. Secondary measures compared LMIC-focused and high-income country (HIC)-focused grants by research focus and health area and identified gaps relevant to ethical and responsible AI use in global health. RESULTS: Of 1850 active NIH AI health research grants, 97 (5.2%) focused on LMICs, representing US$40.2 million (2.4%) of the total US$1.66 billion portfolio. compared with HICs, LMIC-based studies emphasised diagnostics and treatment (72.2% vs 66.8%), health system optimisation (18.6% vs 15.6%), disease surveillance and outbreak response (14.4% vs 8.8%), and telemedicine and remote care (7.2% vs 4.4%). HIC-based grants more frequently addressed public health education (10.4% vs 8.2%) and ethics and data governance (12.8% vs 7.2%). All settings emphasised data science training and capacity strengthening, as well as basic research and early-stage AI-augmented tools. LMIC-based studies most often targeted non-communicable diseases (39%), communicable diseases (30%) and health system strengthening (24%). 31 awards were made directly to LMIC-based principal investigators (1.7% of the portfolio), most commonly in South Africa, Kenya and Uganda. CONCLUSIONS: NIH investment in peer-reviewed AI-enabled health research is expanding globally. LMIC-focused studies prioritise areas aligned with pressing global health needs, including outbreak detection, disease surveillance, diagnostics and treatment, health system optimisation and remote care. Greater attention to ethics, data governance and public health communication, alongside support for digital infrastructure and meaningful collaboration, may help strengthen the relevance and sustainability of AI-enabled research for population health.

Ähnliche Arbeiten