30 - AI: Show real

Having lots of regulatory qualifications for new drugs makes perfect sense. It’s widely known, however, just how much work goes into each submission and the amount of ‘paperwork’ this creates – albeit virtual paperwork. I’ll never forget a presentation I saw in 2015, showing a submission consisting of 1,485 folders, placed in 375 boxes, weighing a total of over four tonnes!

Let’s also not forget how long it takes the regulators to then sift through each submission before it can go to market. It’s a massively time-consuming process, all driven by the fact that you can’t truly see how well drugs work without rigorous statistical evidence.

But what if you could – in real time? This is where the use of real-world data (RWD) comes in. Rather than confining it to numbers and code, we must think of data as a medium which gives us the opportunity to observe what’s really happening. In the context of healthcare, we have the opportunity to constantly examine what happens to patients when we give them new treatments.

RWD can dramatically reduce the timeline for bringing life-saving products to market – we only need to recall how quickly COVID-19 vaccines were made available to see its benefits. But it does require a wider industry mind-shift.

Of course, phase 1, deeming a drug or vaccine safe, and phase 2, clarifying dosing limits, are absolutely essential. After that, if RWD is able to give us a perfect insight into how effectively each new drug works, it makes the time typically spent on phase 3 seem unnecessary.

Instead, a never-ending ‘phase 4’ can be used to identify the nuances of the drug, in the hands of well-informed clinicians. Those that work well will rise to the top, while other less effective drugs will become scarcely used. It would encourage a real growth mindset and present hitherto unchartered drug development opportunities.

To work, the regulatory requirements must step away from just receiving vast amounts of submission content. Instead, it could revolve around an ever-expanding store of data, regularly updated by pharma companies – fed by RWD and inspected by the regulator.

This changing terrain also presents opportunities for clinicians to make analytically driven decisions. I don’t yet see a time when we won’t have an expert evaluating that information but if technology is able to give data-led observations using RWD, clinicians can combine this with their own expertise to improve the patient experience and their outcomes.

AI-powered technology could also create ‘medication alignment scoring’ for patients. For example, if a certain drug is most effective for a specific condition, but only if it’s taken consistently at exactly the right time, then it will only work if patients are reliable at taking their medication. AI might predict whether other drugs are less sensitive to timing, thereby helping patients who struggle with adherence.

There are also opportunities for change in the clinical trial process too, and I’m all for encouraging confidence in decentralised clinical trials. Driving meaningful observation, however, requires a consistent semantic and syntax for RWD which enables collection and aggregation, at scale.

Once these conversations start and a new, global standard established, AI in healthcare can become an inspiring reality.