AI has a “transformative potential” to help design new drugs and diagnose diseases early, the UK’s Bioindustry Association (BIA) said in a report.
Tech chiefs want faster access to patient data to help AI models learn quickly, a biotech conference in London heard, raising concerns over confidentiality.
BIA chief executive Steve Bates on Wednesday said Britain was second only to the US in terms of financing for data-driven life sciences. Several companies are developing AI-influenced cancer treatments.
One AI drug discovery company, Exscientia, has raised almost $500 million while a second concern, Benevolent AI, has a collaboration with Covid-19 vaccine maker AstraZeneca.
“The UK is a fantastic place to be at the heart of this,” Mr Bates said. “This is the place where much of this is going to happen.”
An annual report on the state of AI published last week by US investors Air Street Capital predicted "significant breakthroughs in the foreseeable future as a result of AI's use in the sciences".
It said AI had shown promise in breast cancer screening by learning when a computer model could be trusted and when human intervention is needed.
The UK biotech industry is urging Mr Sunak to let drug regulators draw up “nuanced, context-specific” rules and allow companies to play with untested AI systems in a “regulatory sandbox”.
Ministers have promised a “pro-innovation approach” that ensures AI technology does not “infringe on the privacy of individuals” or undermine human rights.
“The introduction of additional tools such as ethical guidelines, evaluation frameworks and data standards will also be beneficial in promoting responsible AI use,” the UK industry’s report said.
“Education for clinicians, researchers and patients is paramount to mitigate overconfidence in AI tools and ensure their efficacious utilisation.”
Britain’s National Health Service is widely regarded as overstretched and is invariably a top voter concern at elections. At the same time, the UK is regarded as a leader in high-tech fields such as genomic sequencing.
There are predictions that modern computing could usher in an era of personalised medicine in which doctors experiment on your AI-powered virtual clone, known as a “digital twin”.
This would require AI to crunch through vast amounts of patient information. NHS England recently moved to replace a policy of “data access as default” with a scheme granting entry to approved users.
"What could make a really transformational difference is the way in which we can access data in a way that both protects patient confidentiality but also allows us to access data at scale," said Emily Jefferson, chief technology officer of Health Data Research UK.
"How do you make sure that you can have AI-trained models that can actually be exported from that environment without them encoding potentially sensitive data?
"In this space, I think we need a real transformational programme to enable industry globally to come and securely access UK-wide data sets but in a safe way that the public are happy with and they feel their data isn't being sold to big pharma, which is always that pushback."