My View: Integrating AI into NZ’s healthcare system

VIEW - Robyn Whittaker, director evidence research & clinical trials, Te Whatu Ora

At Health New Zealand | Te Whatu Ora, we are working to integrate artificial intelligence (AI) into the healthcare system in a responsible, ethical, and effective manner. Our National AI and Algorithm Expert Advisory Group (NAIAEAG) and the AI Lab within Te Whatu Ora are key components of this effort, providing governance, evaluation, and guidance on AI adoption in public healthcare.

Governance

The formation of a centralised AI group became necessary as New Zealand transitioned from 20 separate district health boards into a unified national health organisation. This aligns with recommendations from the Prime Minister’s Chief Science Advisor, which emphasised leadership, governance, and evaluation of AI tools before implementation. Our focus is on ensuring that AI is used in a way that is safe, equitable, and beneficial for both clinicians and consumers.

Governance is crucial because AI has well-documented risks, including issues of privacy, bias, and its impact on the clinician-patient relationship. Historical data used in AI models can perpetuate biases, so we must be diligent in mitigating these risks. Our goal is to strike a balance between enabling AI innovation and ensuring that its deployment is ethical, trustworthy, and aligned with public health values.

Evaluation

One of our key initiatives has been the development of a comprehensive AI evaluation framework. Initially created at the district health board level, this framework has evolved into a national checklist incorporating nine perspectives: consumer, Māori, equity, ethics, clinical, data, technical, legal, and the fundamental question of whether AI is even appropriate for a given problem. This ensures that every AI tool is rigorously assessed before deployment.

NAIAEAG operates by reviewing AI proposals at two stages: during initial conceptualisation and before final endorsement. Our multidisciplinary team, which includes both internal and external experts, evaluates proposals against our checklist to identify potential risks, challenges, and necessary safeguards. We emphasise real-world implementation considerations, including how AI integrates into clinical workflows and its impact on patient interactions.

Guidance and implementation

Beyond individual AI proposals, we provide general guidance on AI-related issues. For example, we have published advice on the use of large language models in healthcare settings and are continuously updating our recommendations to reflect technological advancements. We also collaborate with the Ministry of Health and other government agencies to ensure alignment on AI policy and regulation.

Looking ahead, we are working on several critical initiatives. We aim to establish clear pathways for evaluating AI tools, differentiating between high-risk and low-risk applications to streamline the approval process. 

We are also maintaining a national register of AI tools used within Te Whatu Ora to track their implementation and impact. Additionally, we are supporting research into new AI evaluation methodologies, particularly for generative AI, and studying patient perspectives on AI in healthcare.

The future 

Education and public engagement are areas where we need to expand our efforts. As AI adoption grows, it is essential that both healthcare professionals and the general public understand AI’s capabilities, limitations, and ethical considerations. 

We are watching overseas developments through involvement in international networks. 

Through ongoing research, governance, and collaboration, we are committed to ensuring that AI serves as a valuable and responsible tool within New Zealand’s healthcare system.

This View is based on Robyn Whittaker’s presentation to the AI in Health Workshop at HiNZ Digital Health Week and has been created with the help of ChatGPT.

Register now for the Digital Health AI Summit this May 20-21 in Wellington.

Read more VIEWS

Return to eHealthNews.nz home page