Research conducted at the speed of light, synthetic personas responding instantly to complex questioning and algorithms rifling through the deep recesses of data for insights we didn’t even know we needed. It is a terabyte goldmine.

With its capacity and reach growing daily, the algorithms and machinery of computational learning herald a new age where remote, platforms, cyber and digital are the workhorse words of everyday healthcare.

The promise to tackle and treat conditions with targeted and economically viable therapies is beguiling, but it dissipates without the human touch. Artificial intelligence will always be…artificial.

And it is prone to mistakes – some of them alarming – as the EMA’s guiding principles of using large language models (LLMs) – the computer programmes that are trained on huge data sets – warns, “The use of LLMs is not without risk. LLMs have shown surprising failure at seemingly trivial tasks, returning irrelevant or inaccurate responses – known as hallucinations.”

How we make the best of AI is the echoing thought that greets every new announcement of AI’s potential and regulators around the world now view it as one of their prime purposes.

Laura Squire, Chief Quality and Access Officer at the UK’s MHRA, underscored the need for overwatch and partnership as she welcomed the UK government’s £500m commitment to become a science and technology superpower by 2030.

She commented, “The quest to control its (AI) capabilities becomes more focused and essential every day and regulators around the world are collaborating with pharma to bring in the right controls that provide directional guidelines without compromising potential.”

Juan Equihua, Head of AI at Havas Lynx, the global healthcare communications agency, sees an exciting future for AI as a tool to facilitate and elevate the creativity and decision-making that is at the core of the company’s service.

“We embrace AI’s potential but we are also pragmatic. There are so many ways it can be used to empower people to do their jobs better, principally by freeing up their time to do what they do best – the creativity and the critical thinking,” he says.

“LLMs cannot make decisions but they are good at summarising and collating research, taking away the mundane work, and these capabilities allow us to focus on the more strategic tasks such as delivering efficient and effective campaigns for our clients. AI is there to enhance creativity and deliver meaningful high-quality outcomes.

“AI should augment our skills and workflows, not replace human creativity and judgment.”

AI is here to stay and its influence is spreading across healthcare and industry. Havas Lynx is maximising its capacity to process large data quickly, uncover patterns and insights that might be missed by human analysis and automate repetitive tasks.
Juan adds, “Our synthetic personas have been particularly effective in simulating diverse healthcare professional (HCP) experiences and outcomes, allowing us to anticipate and address potential communication challenges before they arise, as well as significantly reducing time to test campaigns.”

But a critical point for him is that AI is subordinate to human decision-making. He comments, “We follow a principle of AI augmentation rather than replacement. This means that AI tools are used to support and enhance human decision-making, not to replace it.

“We work in a human-first approach, which means any AI output is ultimately a human responsibility, and AI content is not produced and published in its generic form.”

Juan also contends that AI has a role to play in improving patient engagement by facilitating targeted approaches, adding, “In essence, the automation enabled by generative AI allows us to design communications and products that are no longer segment-based, but rather individual-centric. This creates a paradoxical effect where technology is actually fostering greater human connection through tailored interactions that acknowledge and respond to each person’s unique needs.

“Ultimately, as we integrate generative AI into healthcare, it’s not about replacing human interaction, but rather amplifying its benefits by providing more personalised and empathetic communication experiences. By doing so, we can build trust and confidence with patients, ultimately improving their overall health and well-being.”

Juan advocates making AI a core business priority. “The potential is exciting but many companies do not give it enough resource or importance and that is why projects fail. Companies that use AI successfully embrace it and ask how it can make their business model better,” he says.

The deployment of AI at Havas Lynx is governed by overriding principles of compliant ethical use, a patient-centric approach, AI augmenting human creativity and critical thinking, continuous learning and adaptation, and accountability and transparency.

Transparency is a fundamental of any effective AI-influenced programme or campaign, says Lina Eliasson, a partner at Sprout Health Solutions, which specialises in patient engagement across the full drug development pathway, from preclinical patient insights to patient-reported outcomes in clinical trials, as well as post-launch patient and caregiver support.

“AI has a great ability to speed up the identification of knowledge gaps, but transparency is essential in any engagement with patients. If any AI-enabled technologies are being used, they must be disclosed upfront,” says Lina, a specialist in behavioural science and health outcomes.

Although AI is sometimes viewed as a harbinger of a diluted patient voice, Lina believes that it can empower them. “We would never move away from having the patient voice at the centre of what we do, but AI can provide a lot of benefits, not just in its ability to process huge amounts of data quickly.

“It can reduce the burden on the patient communities. For example, much of our work at Sprout focuses on rare diseases, and we find that those patients often have fatigue from being asked the same question over and over again by different companies.

“Through AI-enabled technologies, we can more easily collate all the data and knowledge that is available across online forums, grey literature and published articles, and laser-focus new research on specific topics that might not have been covered in detail – or at all.

“Not having to ask patients the same questions benefits both the patient community and the research insights we generate.”

Lina believes that AI will gain greater traction and allay fears over privacy and accuracy as it becomes more widespread and its influence has a direct effect on healthcare system efficiency and patient outcomes.

“I think there is an inherent mistrust in new technologies and there has been a lot of scaremongering around AI specifically. Over time, people will become more accepting and appreciative of the fact that we can speed up research and bring medical products to market quicker,” adds Lina.

“We are already seeing it action, particularly in medical devices and diagnostics. For instance, we developed patient information for a programme that uses AI to identify individuals at risk of a debilitating rare disease from health records. Those at-risk individuals are offered a diagnostic test that help them receive an accurate diagnosis and effective treatment. The introduction of AI is clearly hugely valuable to these patients.”

Regulators such as the US Food and Drug Administration (FDA) and European Medicines Agency (EMA) are playing a positive role in advancing AI by providing comprehensive guidance and also through their engagement with pharma companies throughout clinical development to ensure the machine learning element is valid.

Payers are also on a collaborative AI pathway. NICE in the UK issued a positioning statement that highlighted benefits and potential risks of AI methods in health technology assessment (HTA) and invited early collaboration, stating, “Submitting organisations considering using AI methods should engage with NICE to discuss their plans. When appropriate, early engagement could be sought through NICE Advice. At later stages of evidence development, organisations should discuss their plans with appropriate NICE technical teams.”

Lina observes, “Pharma is typically conservative and I think that regulators and payers will drive the changes by setting parameters to govern responsible use of AI in drug development. There will of course be a few industry trailblazers who will make big advances and they will be working with regulators and with HTA agencies to drive change.

AI is the biggest topic at conferences these days and the discussions are about efficiency, being able to bring drugs to market more quickly, and being able to reduce the burden and risks to patients in clinical trials.

“Regulators have a lot of interaction with pharma throughout drug development pathways. At Sprout, we help our clients to get regulatory feedback on their strategies to capture patient-reported outcomes within their programmes. Regulators are thoughtful in their feedback and we expect their guidance on the use of AI will help facilitate patient-experience data generation. It is clear that payers like NICE also want companies to come to them early if they are thinking of using AI or AI-generated data to inform cost-effectiveness analysis and reimbursement decisions.

“Although it appears to be a slow revolution in pharma, the use of AI in health research will change the process and speed by which we can collect and evaluate data, improving our ability to accurately spotlight the patient voice, from preclinical work through to real-world, in-market programmes.”

AI promises to be a catalyst, from proof of concept and clinical trials to patient involvement, but it is paramount that controls and direction are in place to ensure they amplify patient needs and the patient voice rather than drowning them out.

References are available on request.

Danny Buckland is a freelance journalist specialising in the healthcare industry