The AI Revolution in Asian Healthcare
Stratifying resources for improved patient outcomes
Sylvia Varela, AVP Asia, AstraZeneca
Sylvia Varela, AVP Asia at AstraZeneca, explores how AI drives clinical outcomes via resource-stratified approaches suited to Asia's healthcare landscape. From early lung cancer triage via existing screening to enhancing heart failure detection in primary care, she discusses the role of AI in cost-effective, scalable solutions that aim to improve early diagnosis and patient outcomes.
The landscape of global health is at a critical juncture. While medical advancements continue to push the boundaries of what's possible, significant disparities in healthcare access and outcomes persist, particularly in low-and middle-income countries (LMICs), impacting access to medicines and patient outcomes. Asia faces unique challenges in delivering equitable and effective healthcare, characterised by varying resource availability and unique population needs. In this context, AI is emerging as a transformative force, not merely as a technological novelty but as a practical driver of improved patient outcomes through resource-stratified solutions. These solutions are designed to optimise the use of existing, often limited, healthcare assets, ensuring that advanced diagnostics and treatments are directed where they are most needed. By intelligently leveraging infrastructure and data, AI is aiming to significantly improve early diagnosis and contribute to more resilient and equitable healthcare systems across the region.
Maximising impact with existing infrastructure: AI’s role in resource stratification
The success of a resource-stratified approach hinges on how limited resources are leveraged to achieve the greatest possible health impact. Rather than demanding entirely new, expensive infrastructure, AI can be integrated with existing, widely accessible tools like chest X-rays or primary care consultations. By acting as an intelligent pre-screening or triage layer, AI can accurately identify individuals at high risk who genuinely require further, more specialised (and often costlier) investigation. This not only optimises the use of scarce resources by directing them where they are most needed but also expands the reach of early detection to broader populations who might otherwise be missed by traditional, resource-intensive screening protocols. This paradigm shift ensures that advanced diagnostics are utilised efficiently, while foundational healthcare services become more effective and equitable.
Lung cancer: Why early detection is crucial
Take the example of lung cancer, a disease for which Asia bears most of the global burden. In 2020, the region accounted for over 1.3 million new cases and 1.1 million fatalities. Cases are set to rise as well, with GLOBOCAN estimating a 64 per cent increase in new cases and a 73 per cent increase in deaths by 2040. Adding to the complexity of lung cancer is that more than 50 per cent of cases are diagnosed at Stage IV, a point where the five-year survival rate plummets to less than 10 per cent. Conversely, if lung cancer is detected at Stage I, patient five-year survival rates increase to more than 68 per cent. The profound difference highlights the critical importance of early detection, but in many LMICs across Asia, limited access to advanced diagnostics means delayed diagnoses and treatment. While low-dose computed tomography (LDCT) is the gold standard for screening and assessing incidental pulmonary nodules (IPNs) — small, non-calcified nodules often discovered incidentally during imaging for other reasons — its cost and accessibility often make it unviable for widespread screening in many resource-limited settings. This is where the strategic integration of AI, particularly with existing infrastructure like chest X-rays (CXRs), offers a game-changing potential.
AI-enabled chest X-ray triage: Expanding the net for lung cancer detection
The current paradigm for lung cancer screening often focuses on individuals with traditional risk factors, primarily current or former smokers between the ages of 50 and 75. Even within this targeted group, conventional CXRs can miss a significant proportion – 19 per cent – of lung cancers presenting as nodules. This narrow focus, while understandable given resource constraints, leaves a vast and at risk population vulnerable.
AI-enabled CXR triaging, on the other hand, can help expand screening programmes beyond traditional risk factors. By integrating AI, routine chest X-rays – millions of which are performed annually – can be analysed for the presence of pulmonary nodules. This allows for the identification of high-risk individuals well beyond the traditional risk group, including never-smokers, those with a family history of lung cancer, or individuals with high exposure to indoor air pollution, who would otherwise be missed by conventional screening protocols. Once screened, only those flagged as high-risk by the AI would then require an LDCT for confirmation, optimising the use of this more expensive and less accessible resource. This proactive approach has the potential to transform routine imaging into an early detection tool that can increase the chances of identifying IPNs that warrant further investigation.
The CREATE study: Validating AI's efficacy in real-world settings
The real-world utility of AI-enabled chest X-ray triaging has been powerfully demonstrated by the CREATE Study, presented at the European Lung Cancer Congress (ELCC) in March 2025. This landmark study, involving over 700 participants across five countries (Egypt, India, Indonesia, Mexico, Turkey) with identified IPNs, compared the AI-enabled CXR triaging tool's ability to predict malignancy risk against radiologist assessments based on LDCT scans and Lung Imaging Reporting and Data System scores.
The results were compelling, validating the use of the AI-driven lung nodule malignancy scoring tool to predict whether detected IPNs were likely benign or malignant. Findings indicated that the AI-enabled CXR triaging tool accurately identified high-risk nodules 54.1 per cent of the time and accurately excluded low-risk nodules 93.5 per cent of the time. Crucially, these results were consistent across all patient subgroups, including never-smokers and individuals aged under 55 years – those who are typically excluded from traditional lung cancer screening programmes.
The CREATE Study findings tell us that AI-enabled triaging of incidental chest X-rays can optimise lung cancer screening workflows across diverse healthcare settings and prove beneficial in resource-limited settings.8 Most importantly, adopting this technology can significantly expand screening populations beyond typical high-risk groups, leading to earlier diagnoses for a broader range of individuals and optimising treatment costs. Notably, a further budget impact model for AI-enabled CXR triage in Vietnam projected that its national implementation would be a cost-neutral intervention, meaning the initial investment would be offset by subsequent cost savings. More importantly, the model projected an additional 3,155 lung cancer cases identified and a staggering 4,742 premature deaths averted due to earlier diagnosis.
Under the World Economic Forum’s EDISON Alliance, which aims to improve lives through digital healthcare access, AstraZeneca and Qure.AI completed over 5 million AI-enabled CXRs across more than 20 countries, including over half in Asia. As a result, lung nodules at high risk for cancer have been identified in nearly 50,000 people, who have subsequently been referred for further testing and potential diagnosis. This approach allows healthcare professionals to focus their limited resources on those patients who truly need further assessment, ensuring that screening programmes are as effective and efficient as possible in shifting populations away from late stage diagnosis to potentially improve survival rates.
Heart Failure: Using AI to identify heart failure in primary care and open new career pathways in healthcare diagnostics
The impact of AI is not limited to oncology. Heart failure, a chronic and progressive condition, represents another significant global health burden, particularly in Asia. Like lung cancer, early detection is crucial for effective heart failure management and improved patient quality of life. Despite symptoms often appearing up to five years before diagnosis in over 40 per cent of patients, heart failure remains the leading cause of hospitalisation in individuals over 65 years.
The recent findings from the Heart2Miss heart failure screening project in Malaysia, unveiled at the European Society of Cardiology’s Heart Failure Congress in 2025, highlight AI's potential in this area. This project utilised a decentralised rapid cardiac ultrasound triage model, leveraging trained bioscience graduates as mobile community sonographers to screen 1,000 high-risk diabetic patients. The model demonstrated a strong ability to identify heart failure within primary care settings, detecting 11 per cent with pre-heart failure and 1 per cent with previously undiagnosed heart failure.
Crucially, this approach not only simplified patient referrals and outpatient management, permitting early interventions, but also significantly eased the workload on tertiary centres: only 1 per cent of these high-risk patients required referral for specialist intervention. Furthermore, by successfully enhancing the diagnostic performance of novice users across over 400 cumulative patient cases, the project demonstrated a clear pathway to build a new diagnostic workforce, overcoming bottlenecks in specialist care due to overwhelming demand. This approach empowers primary care practitioners to detect heart failure early, even in high-risk, asymptomatic individuals at the pre-heart failure stage. It powerfully demonstrates how AI can innovatively alleviate resource constraints while having the potential to maximise healthcare outcomes. Similar projects are now underway in Vietnam, aiming to analyse data from 10,000 patients by early 2026, and in Singapore, where preparations for project data publication are ongoing.
Towards a new era of equitable and effective healthcare
AI is not merely an incremental improvement but a fundamental shift in how we approach healthcare delivery, particularly in environments where healthcare resources are constrained. By enabling AI-powered chest X-ray triaging for lung cancer and innovative heart failure screening in primary care, we are witnessing a paradigm shift towards more equitable access to diagnosis, better patient outcomes, and ultimately, reduced healthcare costs. It can also open pathways to overcome barriers to diagnosis in other prevalent diseases, such as asthma and chronic obstructive pulmonary disorder (COPD). The strategic application of AI allows for the stratification of resources, ensuring that advanced and often expensive diagnostic tools are utilised only when truly necessary, while broader populations benefit from cost-effective, AI-enabled screening.
As we look to the future, the continued integration of AI into healthcare systems across Asia holds great promise. It is a powerful tool that can bridge the gap between medical innovation and real-world impact, empowering healthcare professionals, potentially improving patient lives, and building more resilient and equitable healthcare systems for generations to come. The AI revolution in Asian healthcare is not just about technology; it's about transforming lives and building a healthier future for all.
Author Bio
Sylvia Varela is Area Vice President, Asia at AstraZeneca, where she oversees operations of nine markets in the region. With 25+ years of experience in pharma, she focuses on building win-win partnerships with public and private sectors to pioneer patient-centric programmes, expanding access to innovation, and reaching new patient populations.
