By: Andy Wong Ming JunIn a dusty rural town 50 kilometers south of Manila, an elderly male patient sits in a simple clinic as he seeks medical treatment for the painful red spots on his ankles. During his medical consultation, he notices his attending doctor running a description of his ailment symptoms through ChatGPT before diagnosing his medical condition, prescribing the appropriate medication to alleviate his symptoms, and sending him on his way in a matter of minutes.This illustrates the growing impact of artificial intelligence (AI) on healthcare delivery and service quality in lower-income regions, from Africa to Southeast Asia. Physicians with limited training or experience are increasingly using their phones to call up AI as a supplementary tool for diagnosis and prescription decisions.Beyond that, AI is supporting faster screening, improving diagnostic accuracy, and reducing prescription errors. According to a recent study, AI-assisted tuberculosis screening reduced misdiagnosis rates by more than 30 percent and improved predictive forecasting of seasonal disease outbreaks by more than 80 percent.AI adoption within public healthcare systems has seen remarkable growth in the past few years. However, this exponential growth in routine clinical adoption remains highly uneven between countries and health systems along socioeconomic wealth and median income lines. In the Asia-Pacific, the growing usage of healthcare AI serves more than anything else to highlight how the world’s most populous region continues to see some of the largest unmet healthcare needs and the strongest potential for technology to compensate for shortages of human expertise.According to the World Health Organization, over the past two years, healthcare AI across the world has experienced increasing policy attention, experimentation, and early implementation. As far back as 2021, the WHO listed six guiding principles within a groundbreaking global report on AI healthcare ethics meant to provide fundamental guardrails for AI adoption in improving healthcare provisions. These six guiding principles focus on keeping human autonomy at the heart of any AI-enabled healthcare system and medical decision-making framework, with appropriate transparency and accountability in using appropriately inclusive and representative datasets for safeguarding the public interest across diverse societies.In the Asia-Pacific, AI has already been used to improve the speed and accuracy of diagnosis and screening, while also highlighting applications in clinical care and public health. For instance, last year in the Philippines’ Sulu province an hub-and-spoke teleradiology model system with potential AI application pathways was introduced with specific intent to address diagnostic-imaging gaps in one of the archipelago country’s most underserved healthcare areas. Like many other lower-income countries, particularly in the Asia-Pacific with a record of outsized but binary healthcare professional contributions outside of their own borders, the Philippines is increasingly reliant on telemedicine to provide the necessary healthcare infrastructure to overcome geographic and specialist-access barriers, while AI is developing as a separate but increasingly relevant layer of its digital-health ecosystem.Another example can be found in Singapore’s adoption of the SELENA+ AI software system into its national diabetic-retinopathy screening infrastructure since 2020. Singapore has one of the highest rates of childhood myopia in the Asia-Pacific with almost one in three children requiring the usage of glasses by the age of 7 and significant diabetic retinopathy along similar statistical lines. The exponential growth in ophthalmologic screening access enabled by a centralized national tele-ophthalmology infrastructure with trained human graders examining retinal photographs at a central reading center in Singapore saw annual screening numbers balloon to over 100,000 by 2019.That has provided major impetus for the development of SELENA+ in 2018 as a deep-learning AI system built on analyzing over 500, 000 retinal images and corresponding locally-generated datasets taken for screening diabetic retinopathy, glaucoma, and age-related macular degeneration health issues. After 2021, SELENA+ has been widely recognized as a successful example of AI being used to extend scarce specialist healthcare capacity to cope with a wider screened population with quicker diagnostic triaging.According to Medicins Sans Frontiers (MSF), since 2022, AI has been successfully leveraged in expanding the healthcare workforce capability in lower-income countries to analyze and interpret antimicrobial susceptibility testing beyond pure reliance on fully-trained microbiologists through its Antibiogo offline smartphone application. MSF has also successfully used AI to enable greater throughput and accuracy of tuberculosis community screening in the Philippines since 2024.AI is increasingly being considered as a component of future health-system infrastructure especially in enabling lower-income countries to catch up to their higher-income developed counterparts, but many developing economies remain at the stage of establishing the digital-health foundations needed for large-scale deployment. The two main obstacles can be roughly classified into hardware capability and software data relevancy, given the high energy demands of datacenters critical for AI computational processing and the still-entrenched Western-centric datasets used as default by predominantly Western AI companies in their model developments.The issue of high energy demands in existing AI processing infrastructure is an acutely growing concern for the Asia-Pacific, particularly when multiple countries have already been driven to energy rationing and curtailing in-person office work due to the ongoing price shock fallout from Iran’s disruption of Middle Eastern maritime energy imports since February this year. According to the UN, the Asia-Pacific is projected to account for 85% of global power-demand growth in 2026 with much of it driven by increased electricity demands from data centers.Furthermore, evidence for a positive-impact growing adoption of generative AI and their corresponding large language models in routine healthcare information provision, clinical decision support, and administrative tasks continue to remain limited. This is an issue that will be resolved in time, but only if conscious effort is put into progressive integration of AI-assisted diagnostics and triage into existing physical and telemedicine infrastructure, rather than a fully established AI-enabled, but solely telemedicine model.