AI in healthcare: navigating potential and regulation in a fast-growing field

Artificial intelligence (AI) is revolutionising healthcare, from diagnostics to patient care, but regulation is struggling to keep pace, according to a new paper from City St George’s, University of London.

Published in LSE Public Policy Review, the research by Dr Divya Srivastava, Assistant Professor in Research Methods at City St George’s, highlights the pressing need for increased international collaboration grounded on robust evidence. It also proposes specific measures for researchers and decision-makers to address AI’s potential and pitfalls in health systems worldwide.

The growing impact of AI in health

AI has already made significant inroads in healthcare, particularly in diagnostics. From radiology to cardiology, AI algorithms support medical professionals by interpreting complex data, detecting diseases, and recommending treatments. This includes its use in radiology for image-based diagnostics and in screening tools for diseases like cancer and heart disease.

“With the right data and frameworks, AI has the potential to dramatically improve patient outcomes and reduce medical errors,” said Dr Srivastava. The paper also mentions that AI could reduce miscommunication errors by up to one-third.

Regulatory challenges with adaptive AI

One of the biggest challenges is regulating adaptive AI systems, which evolve as they learn from new data. Unlike traditional technologies, these models are not static, presenting unique difficulties for policymakers. In response, organisations such as the US Food and Drug Administration (FDA) have focused on AI’s functional use, rather than specific technical components, to create more flexible guidelines. In addition, the paper notes that while AI's potential is substantial, only 1% of health apps and digital tools are backed by rigorous evidence.

“Traditional regulation cannot keep pace with adaptive AI systems, which require a flexible, ongoing approach. Collaboration across sectors and borders is crucial if we’re to ensure safe and effective AI applications in healthcare,” added Dr Srivastava.

Addressing data privacy and bias

For AI to perform accurately, vast and diverse datasets are essential. However, this raises issues of data privacy, security, and potential bias. There are also concerns over whether AI models adequately represent all population groups. This could lead to biased healthcare outcomes, potentially limiting AI’s effectiveness for certain demographic groups. To address these concerns, Dr Srivastava’s research calls for privacy-protected, diverse data sources and cross-country ethical guidelines, such as those proposed by European and North American radiology societies. Some researchers are exploring blockchain and other privacy-enhancing technologies to help manage sensitive health data.

Strengthening international cooperation

The study also highlights the need for enhanced global cooperation to create regulatory standards, ethical guidelines, and frameworks for evidence generation as AI technologies often transcend national boundaries. Organisations like the World Health Organization (WHO) and the Organisation for Economic Cooperation and Development (OECD) are creating forums to foster shared standards and best practices.

Dr Srivastava said: “AI is a borderless technology, and international cooperation is essential. Only through collaboration can we achieve consistent, evidence-based standards for AI in healthcare.”

Towards a ‘Total Product Lifecycle’ approach

Overall, Dr Srivastava’s research suggests a “total product lifecycle” (TPLC) approach for AI in healthcare, advocating continuous evaluation and monitoring beyond the initial approval stage.

“While AI offers immense potential to improve healthcare outcomes, its success relies on robust, adaptable policies and cross-border partnerships. Overall, we need to ensure a proactive, evidence-driven approach, with ongoing international collaboration, and this will be essential for AI to remain a positive force in global health,” said Dr Srivastava.