AI can play an important role in transforming the way healthcare is delivered, but only if it can connect siled systems. Gerard Hanratty and Chris Holder explore the potential of digital innovation, how to achieve interoperability, and the necessary safeguards.
Whether deploying robotics for the most precise surgical procedures or using machine learning algorithms to quickly analyze millions of data points in genomic discovery, AI offers great potential for medical innovation.
Of course, it’s not a magic wand that will cure everything. The NHS needs to adapt its procurement systems and wider business practices to incorporate emerging technologies into government policy and regulation of service delivery, while becoming an enabler rather than a barrier to innovation, for example by giving patients more control over their data and allowing healthcare providers to share anonymized patient datasets across borders for research purposes.
Naturally, it is important to ensure proper governance is in place when deploying AI tools. This is despite the fact that there are currently no specific regulations regarding AI.
In our work on AI governance, we have applied guidance from information technology, artificial intelligence, and management systems in establishing a framework for organizations to implement and use AI.
AI opportunities in medical innovation
The direction of travel explains why this is preferred. The UK Government’s ambitions in its 10-year health plan hinge on three major shifts: from hospitals to communities, from analog to digital, and from disease to prevention.
Advances in technology, particularly AI, can support more personalized care. The government is betting big on the NHS app, envisioning it could become a ‘digital health wallet’ – a secure, user-controlled platform to store, manage and share health information.
Taking inspiration from how the banking industry uses real-time metrics and intelligent insights to help individuals manage their finances effortlessly, healthcare providers could implement AI-based incentive initiatives that revolutionize patient self-management.
Extending patient control over their own data and enabling seamless data transfer between different trusts and other healthcare providers is essential to realizing these aspirations. The Life Sciences Sector Plan also highlights the need for further investment in genomics, including DNA analysis, to provide faster and more accurate diagnosis, particularly for rare diseases and cancer. This is another central pillar of a preventive health system supported by early detection and personalized medicine.
AI will reshape genomics by sifting through vast amounts of data and deploying predictive algorithms to understand, for example, how aspirin affects some people compared to others. Pharmaceutical companies can use this information to adapt medicines to be more effective among different groups of people.
Overcoming barriers to digital health innovation in the UK
Achieving a fully interoperable health data ecosystem across the UK requires facing a practical reality. There are 42 integrated care systems in the UK, and there is a fragmented approach to how the NHS collects and stores data.
This raises questions about data ownership, challenges the consistency of how such data assets are used, and ensuring that different systems can effectively communicate with each other.
Additionally, the NHS manages data that describes a patient’s cradle-to-grave journey, but vast amounts of it are unstructured and therefore not yet readily usable for AI-powered health interventions. The NHS Federated Data Platform aims to address this issue by organizing and integrating fragmented NHS data and incorporating analytics to identify patterns and opportunities for improvement.
Sharing data across borders for research and development of new AI-optimized drugs, treatments, and technologies is also a key element of medical innovation. However, the UK currently lacks the relevant regulatory framework to enable effective collaboration with international partners in this way.
Our view is that the UK should explore bilateral data adequacy frameworks with its closest partners, as well as knowledge sharing agreements to enhance interoperability between international health systems.
AI protection
One of the biggest risks when using AI today is data bias. Flaws in how AI training sets are selected can lead to algorithms discriminating against certain demographics and reinforcing social and health inequalities.
The main reason for this is the ‘digital divide’, where those with the lowest levels of participation are characterized by low-grade technology, poor internet connectivity and low digital literacy. This disproportionately affects people from some minority ethnic groups, who are therefore provided with limited digital profiles and are consequently underrepresented in public datasets.
Governments need to identify the best way to reach these groups and ensure adequate access to GPs and banks to help collect anonymous data.
Guardrails to limit AI bias should be built into data contracts for AI healthcare projects. To implement this more broadly, the UK will need to take this into account when establishing its own AI regulations.
What’s next for AI in healthcare?
AI is not going away because there is a lot of investment in this technology. Used correctly, it has the potential to be a tool that transforms everything from patient self-care to drug discovery.
But to realize that potential safely and ethically, the UK government will need to accelerate its own AI regulations and upgrade backend systems to ensure AI models are trained on high-quality, ethically sourced datasets.
