CRITERION: Looking for a clinical approach to predicting trial outcomes? AI is there for you

Last Monday marked Clinical Trials Day, when the world’s healing profession celebrated the work of Royal Navy surgeon James Lind.

The medical mariner was credited with carrying out the world’s first placebo-controlled clinical trial in 1747, randomly dosing sailors with vitamin C to prevent scurvy.

These days about 35,000 clinical trials are carried out globally, at an average cost of more than $US48 million ($70 million) each for a late-stage study.

Given that, a positive trial result can put a rocket up a biotech company’s share price, while a negative one can send it into a tailspin.

In March, shares in ASX-listed kidney drug developer Dimerix (ASX:DXB) soared 60 per cent after positive interim results for its Phase 3 trial to treat focal glomerulosclerosis (FSGS).

In 2018 Innate Therapeutics’ multiple sclerosis trial famously flopped, leading to Donald Trump’s Congressional aide Chris Collins – an Innate director – being jailed for insider trading.

What if trial designers and investors could remove the guesswork? The wonders of artificial intelligence mean that it is possible to predict a trial result with 90 per cent accuracy, according to life-sciences nanocap Opyl (ASX:OPL).

Dubbed as “ChatGPT” for clinical trials, Opyl ‘s tool, trialkey.ai, applies machine learning to more than 350,000 disclosed clinical trials done globally over the last two decades.

Opyl CEO Saurabh Jain says the “staggeringly high” 90 per cent success rate is based on the outcome of 4189 trials, for which the outcome was not known at the time.

Success was measured on whether the trial met one or more of its primary endpoints – key objectives – and whether the results were statistically significant (not a fluke).

Each trial is rated on 700 variables including enrolment numbers, the drug’s mechanism of action and trial design such as dosing.

“The way clinical trials are carried out is incredibly important as to whether they succeed or not,” Jain says.

So what are the prospects for some of the most keenly-awaited advanced trials being carried out by ASX-listed biotechs?

By the end of the year, Paradigm Biopharmaceuticals (ASX:PAR) is due to report on a Phase 2/3 trial for its repurposed drug to treat knee osteoarthritis.

The algo ascribes a high success probability of just under 86 per cent, partly because of the high enrolment of 900 patients.

Opthea (ASX:OPT) has two Phase 3 trials for the eye condition wet aged-related macular degeneration, due to report mid next year. Combining the company’s therapy with different ‘standard of care’ drugs, the trials are rated a solid 63.8 per cent and 52.32 per cent chance of success.

Sector star Neuren Pharmaceuticals (ASX:NEU) last year won US approval for its drug to treat the rare childhood condition Rett syndrome.

On Friday the stock went into trading halt pending next week’s results of a second trial, for the rare Pitt Hopkins syndrome.

The algo comes up with a success score of 18.5 per cent. But a key rider is that a low prediction score still might be respectable, depending on the difficulty of the indication (illness) pursued and the relative track record of competitor trials.

In this case, Neuren’s chances aren’t out of whack with many comparative trials. In any event, investors don’t have to wait long to know if the machine is on the money.

While trailkey.ai could be a useful investor tool, Opyl’s revenue focus is on advising drug companies on the optimum trial design, such as avoiding expensive patient over recruitment.

If only we could predict whether the $2.7 million market cap Opyl emulates the fate of UK quasi-rival Pharma Intelligence, acquired last year for an eye-watering £1.9 billion ($3.6 billion).

This story does not constitute financial product advice. You should consider obtaining independent advice before making any financial decision

The views, information, or opinions expressed in the interviews in this article are solely those of the interviewees and do not represent the views of Stockhead. Stockhead does not provide, endorse or otherwise assume responsibility for any financial product advice contained in this article.