Opyl (ASX:OPL) is on a mission to improve the $US19.2 billion ($24.9 billion) global clinical trials market with two innovative platforms that will be launching soon.

Clinical trials are the lifeblood of early-stage biotech, pharmaceutical and medtech companies – a huge risk and a huge expense for them.

The estimated 30,000 clinical trials registered each year each cost an average of $US19 million ($24.6 million) to run, with 11 to 40 per cent of that spent on recruitment and retention of participants.

And fewer than 14 per cent of drug candidates that undergo clinical trials eventually win approval, according to a 2019 analysis by MIT researchers. The success rate for oncology (anti-cancer) therapies is even smaller, just 3.4 per cent.

“My view is that there is massive waste in clinical research because of poor trial design and poor patient recruitment,” says Opyl chief executive Michelle Gallaher.

“Good medical research discoveries often don’t make it to the market because they are not given the best chance to succeed, either due to poor design or poor implementation… that’s billions of dollars a year, not delivering a return for investors or patients.”

Patient recruitment platform

While Opyl is already working with a number of global and local pharmaceutical companies on social media insights and market intelligence projects,, the company expects to take a major step forward this month with the beta launch of a precision clinical trial recruitment web-based interface – an AI-based platform for matching the right patients with clinical trials anywhere in the world.

Just 13 per cent of clinical trials use social media to recruit, according to Opyl, which has an in-house team of healthcare digital marketing and Search Engine Optimisation (SEO) experts on staff.

“The growth in optimising social media channels as a recruitment strategy has taken off during COVID,” said Gallaher.

“Community awareness and understanding of clinical trials has grown during COVID19.  With the increase in social media use and the public’s interest in trials primed, we now have an almost perfect global environment to launch a clinical trial efficiency platform.”

Opyl will use its advanced in-house social media listening and precision marketing skills and proprietary tools to funnel patients and volunteers to the web platform, where it can create a profile free of charge, and receive information about research study opportunities in their geographic area and area of therapeutic interest.

They can link in immediately to trial opportunities, quickly determining if they are eligible to join the study or not.  Patients can register interest to be  sent push notifications if a new clinical trial becomes available that matches their self-selected filter.

The beta launch late in March will begin the citicial collection of patients registering on the site, building a database asset that is key to the success of the platform, with a full launch due late in May.

Predicting clinical trial success

The company’s next platform could be even more disruptive: an artificial intelligence-based tool that evaluates clinical trial designs and implementation strategies and predicts their odds of success before they begin and as they progress.

The Opyl platform ingests and continuously analyses hundreds of thousands of past clinical trials, current registered active clinical trials as well as a selection of supporting data from various sources such as company announcements, and identifies patterns of success in therapeutic areas and within classes of drugs or devices.

“There’s a real need particularly amongst small to medium sized biotech and medtech companies who lack experience or access to experience in trial design and implementation insight,” said Gallaher.

“Opyl’s protocol design and prediction platform provides an evidence-based systems approach to design and decision making that is fast, reliable, explainable and affordable.”

When biotech and medtech companies report ‘inconclusive‘ or mixed trial results in meeting primary or surrogate endpoints (the key measures of the success of the drug or device) it’s sometimes due to poor study design and selection of core elements such as; choice of trial sites, trial enrolment targets, dosing regimen, or maybe selection of principle investigator or sponsor that has impacted negatively on the probability of success.

The high trial failure rate, common reporting of inconclusive results, repeat clinical trials and frustration from investors and technology developers suggests a strong market demand for an affordable AI-based optimisation and prediction model, Gallaher says.

While it’s still in development, the machine learning tool successfully achieved an important validation last year.

In early September, Opyl made national and international news with its announcement that its AI prediction platform had completed a major proof-of-concept study by evaluating 475 COVID-19 vaccines and therapies in clinical trials, and identified vaccines that were most likely to success in completing the trials and achieving regulatory approval.

The company’s official announcement didn’t list the vaccine candidates in order of likely success– Gallaher didn’t want to bet the future of her $6 million company on a still-in-development artificial intelligence algorithm.

But she told the Australian Financial Review for a story published September 8 that the leading candidates were the Moderna/US National Institute of Allergy and Infectious Diseases vaccine, followed by the BioNTech/Pfizer vaccine and then the Oxford University/AstraZeneca jab.

And, as we all know by now, the tool was right.

“All of our COVID-19 vaccine predictions have come true,” Gallaher says. “We correctly predicted that the mRNA vaccines would be the first to complete trials and achieve market approval.

“And we predicted it would be the  BioNTech/Pfizer vaccine next, followed by the AstraZeneca/Oxford vaccine … and we also predicted the drug candidatesthat would likely fail including the much anticipated local cadidiate developed by the University of Queensland,” Gallaher says. (That vaccine was abandoned in December after some trial participants generated antibodies that would cause a false positive on HIV tests).

A tool that can accurately predict the likelihood of success of clinical trials and offer optimisation information has vast health and economic implications – for patients, investors, for strategic government procurement, for medical research teams and for the biopharma companies themselves.

So far Opyl’s tool is about 84 per cent accurate, says Gallaher, adding that the company is continuing to work on improving that even further.

The algorithm has been initially trained on millions of data points taken from an online US government database of clinical trials called clinicaltrials.gov, as well as other proprietary and supporting clinical trial, health and organisational data sources.

“We’re not revealing where all of our data sources come from or how we integrate and validate them because that’s a bit of our IP secret sauce,” Gallaher says, “other than to say the information is reliable, authoritative, continuously updated and provided with consent.”

Greater challenge ahead

But although Opyl has made remarkable progress in less than 15 months, Gallaher says the work ahead will be much harder.

“Vaccines are actually quite easy to predict within our model,” Gallaher says. “We really picked a pretty easy target – I say that with a wry smile on my face because I don’t mean to be arrogant, but picking and trying to understand which of the COVID vaccines were going to come on top was probably the easiest trial we could do in the space, but had the greatest relevance and newsworthiness for us.”

Opyl is now training the tool to predict trials in the oncology (cancer) area, which are known to have the lowest success rate of any therapeutic.

“It’s much, much harder to predict oncology candidates, it’s much more high risk, and very, few drugs make it to the market in that space.”

The company is working with clinical trial subject matter experts and its team of data scientists to further train the tool and to increase the number of protocol variables from 150 to over 500 improving its reliability and specificity.

“We’re trying to pick the next killer data trial for ourselves,  trying to break our own model, because we need to know exactly where it won’t work,” Gallaher says.

“Our next data target will be significantly harder to predict than vaccines.”

This article was developed in collaboration with Opyl, a Stockhead advertiser at the time of publishing.

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