Mark Landry addresses the audience before proceeding to the next set of discussions with panelists. Screenshot from PMAC 2025 recording.

Bangkok, Thailand–The Asia eHealth Information Network (AeHIN) led the Prince Mahidol Award Conference (PMAC) 2025 plenary session “Fortifying National Systems for the Age of Artificial Intelligence (AI)” on February 1, 2025, at the Lotus Suite of Centara Grand & Bangkok Convention Centre.

The session followed a fishbowl-style discussion. In addition to the panelists, the audience was free to “jump in and out” of the discussion onstage. Mr. Jai Ganesh Udayasankaran, Executive Director of AeHIN, chaired the session, while Mr. Mark Landry, Programme Officer at the World Health Organization (WHO) Thailand, moderated the discussions.

Prof. Anurag Agrawal from the Digital Transformations for Health Lab (DTH) set the stage by discussing the definition of ‘fortify’ in the age of AI. He stated that fortifying health systems, patients, clinicians, and health systems should all be considered simultaneously. He also mentioned three themes: data, priority, and future, and touched on how training and using AI will determine the implications for health systems, the importance of setting priorities towards applications for global health as well as sound data systems, regional set of priorities, including and involving the young people, and establishing regulatory frameworks.

Experiences from Australia and Mongolia

Prof. Farah Magrabi from Macquarie University and the Australian Alliance for Artificial Intelligence in Healthcare (AAAiH) shared Australia’s national policy roadmap for AI in healthcare and what needs to be done to ensure the sensible and safe adoption of AI in the Australian health system. Australia identified five priority areas with 16 recommendations and integrated AI governance into its existing governance processes for patient safety and digital health. 

Dr. Zoljargal Z Lkhagvajav from the Mongolian Society of Artificial Intelligence in Medicine discussed the ethical principles that guided AI policies and deployment in Mongolia. She highlighted the aspect of transparency in all processes from planning the model, to research, data, and development, and clearly state any limitations or shortcomings to the public and professional because everyone needs to know what AI is and is not, what it can and cannot do, including the generalizability or biases that may be present to be able to use AI responsibly. Dr. Lkhagvajav also mentioned the nomadic lifestyle of those in rural Mongolia, which makes healthcare delivery challenging, so AI solutions must also be tailored to their culture and lifestyle.

AI governance frameworks and prioritizing equity

Dr. Patricia Mechael from HealthEnabled talked about the divide that existed before and still persists today, citing the internet and proliferation of mobile phones and contents are mainly in English, describing the technology ecosystem as a massive digital divide that never corrected itself, so it is not representative of the global population nor designed to promote equity. Dr. Mechael proposed going back to the basics to see the underlying ecosystem and where it is possible to adapt and adjust things because AI systems are as good as their data systems. She added that if the data are not representative of the populations to be served, it will only exacerbate existing inequities.

AI accountability and liability

Prof. Magrabi said that the application of AI should be thought about. While AI is assistive at the moment, like a second pair of eyes or a scribe providing summaries, the liabilities go back to the clinicians. As people automate more, the liabilities are still an open question because they depend on the risk levels and application areas.

Dr. Lkhagvajav talked about the autonomy of clinicians in making decisions, but it could imply that the responsibility would solely be on the clinicians. However, missing at least one of the five rights of clinical decision support can disrupt a clinician’s judgment and lead to mistakes, so there should be certain considerations when clinical support tools and AI algorithms are involved in decisions and mistakes during healthcare delivery. 

Dr. Mechael raised the need for greater digital fluency among the general population and health professionals; the demand for the private sector to be accountable with people’s data and how they are used; transparency and systematic evaluation of AI and what goes into it; and whether the outcomes are fair, equitable, etc. She also mentioned ‘nothing about us without us’, pertaining to people owning and controlling data and educating people how to engage with it.

Prof. Agrawal also talked about autonomous AI, Oxipit, where there is no human in the loop. He stated that, in time, there may be health systems deploying these tools and companies that supply these tools and come up with how to pay for costs and litigation claims that may come up. He also mentioned the need to talk about liabilities further up the chain because of possible scenarios of people using AI on tasks they were not trained for.

Role of civil society in shaping AI governance frameworks

Ms. Indira Dewi Kantiana from the Asia Centre for Health Security at NUS Saw Swee Hock School of Public Health talked about acknowledging the rapid development of AI and how it is driven by private technological corporations, as well as the risk of concentration of power and profit-generation interest. She said that governments cannot work alone, and civil societies can have a role in AI governance. They can be advocates for inclusion and equity. They can also be watchdogs, ensuring that human rights approaches are considered throughout the AI lifecycle process and monitored in accordance with ethical principles.

Ms. Resham Sethi from PATH India discussed a dual governance model in which national and subnational governments work together, and civil society plays a role in bridging the gap between global AI principles and local realities. This includes ethical AI in public health, ensuring that AI policies are not just theoretical but operational, practical, and adaptable, and civil society organizations acting as oversight mechanisms.

Mr. Anis Fuad explains Indonesia’s regulatory sandbox process. Screenshot from PMAC 2025 recording.

Ethical and regulatory standpoint

Mr. Anis Fuad from the University of Gadjah Mada and a member of the governing committee of AeHIN discussed the complexities of liabilities and regulations, particularly the use of AI technology in the healthcare setting. He shared Indonesia’s goal to eradicate malaria by 2030. While digital technologies have improved services and efficiency in various sectors, their application, including in malaria surveillance and remote diagnosis, remains limited due to regulatory gaps. Consequently, tools such as AI-driven detection or telemedicine programs have yet to be widely adopted in malaria-endemic areas. To address these issues, the Ministry of Health, Indonesian Healthtech Association, the University of Gadjah Mada, and partners collaborated to start a regulatory sandbox to explore, discuss, and test digital health technologies. Out of the 18 start-ups that joined the trial sandbox, only six passed the criteria for safe technologies. The Ministry of Health then began a sandbox research initiative on e-malaria and is currently testing several AI-based innovations.

Ms. Kantiana commended Indonesia on processing its regulatory sandbox and the possibility of extending it to AI. She also raised that regulators must first build capacity to avoid a regulatory arbitrage where a regulatory sandbox is introduced, but the safeguards are lowered for the sake of innovation. She stressed the need to ensure that the regulatory sandbox can test and experiment and provide appropriate safeguards not to push for conformity but to benchmark to provide a baseline and move forward.

Dr. Mechael mentioned statistics from the Global Digital Health Monitor (GDHM) that 57% of countries do not have protocol policy frameworks or accepted processes for governing the clinical and patient care use of connected medical devices in health services, and 67% of countries have no emerging technologies and AI policies and strategies relevant to health. She thanked Indonesia for the great example and expressed encouragement to include civil society in the regulatory framework.

Ms. Katiana shared that while civil society plays a critical role in Indonesia’s regulatory sandbox framework, there is no participatory mechanism yet. She added that without a feedback loop, there is no clarity on how civil societies’ voices are heard and incorporated in the process, so in developing a regulatory framework, it is essential to formalize the participation mechanisms so that engagements are not tokenistic.

Dr. Chaminda Weerabaddana from Sri Lanka shared that they do not yet have a clear approach to govern AI in their healthcare system. They also used a participatory approach for their interoperability framework in Sri Lanka, where all vendors, Ministry of Health, Ministry of ICT, and academia came together and created the implementation guide in the workshop. He thinks this approach can also work in developing a governance framework around AI, and may be the better approach instead of having it sent to a legal department for clarity and then taking several years.

Mr. Sean Blaschke from UNICEF reflected on the themes of empowerment, whether designing for or with end-users, and data ownership, whether owned by an individual, the government, or the private sector, and who has control over it. He related it to ethical AI and regulatory frameworks because the current paradigm often states that data is owned by others. Mr. Blaschke added that to get to the core of questions surrounding ethics in AI, the broader issue of who owns and controls the data should be collectively looked at. If the paradigm is that individuals do not own their data, it should also be collectively shifted. 

Mr. Udayasankaran relates his thoughts on AI policies to AeHIN’s work. Screenshot from PMAC 2025 recording.

In the discussion, Mr. Udayasankaran highlighted the importance of health data governance, stating that it will be more meaningful to discuss governance of AI with a clear-cut policy and regulation to govern health data. Reiterating Dr. Chaminda Weerabaddana’s point that establishing policies would entail sitting down together to work on the policies so they can see the day, Mr. Udayasankaran suggested the possibility to align stakeholders relevant for national governance of AI to work together akin to AeHIN’s Convergence approach.

 

Related: