Can AI help with achieving orphan drug status?

Welcome back to our blog series where we share some of the highlights and key takeaways from our fireside chat with CEO of Simbec-Orion, Fabrice Chartier, and CEO of, Joern Klinger.

Our last blog post covered how AI can demonstrate the potential of the pipeline to investors and how AI could disrupt clinical development as we know it today.

This third post in the series continues the discussion on how early utilisation of AI could help de-risk and accelerate clinical development, by discussing if AI can help with achieving orphan drug status, recent trends in clinical trial design, how AI can integrate with existing processes, and what data’s platform utilises to predict clinical outcomes.

Can causal modelling be utilised to identify an opportunity to pursue orphan drug status?

Joern Klinger, CEO,
Yes absolutely. Of the 28,000 drugs out there, approximately 14,000 no longer have any patent protection, and around 4000 of those have already been deemed safe and have proof of mechanism after a successful phase 1 trial. You can imagine many of those could be matched up with rare or orphan diseases and work for those diseases, which is something we are doing right now, to analyse that treasure trove of potential and predict if any existing drugs could be effective for rare or orphan diseases.

Regulators have been discussing the potential of AI and alternate sources of data to streamline clinical trials and enable more flexible, agile clinical trial design[1] – what shifts have we seen in clinical trial design recently?

Fabrice Chartier, CEO, Simbec-Orion
At Simbec-Orion, we are running an increasing amount of studies involving both cohorts of healthy volunteers and patients, which illustrates the flexible nature of modern clinical trial design and an appetite to accelerate clinical development, within regulatory guidelines.

We have seen significant changes in clinical trial design over the past 10 years. Adaptative design, flexible design, risk-based design – all these notions are now commonly used in clinical development. The limit of this type of design is that data-based decisions are required during the study. The data may indicate something which you were unable to predict, and this can sometimes require more time, more funds, or both -to complete the study.

With the use of AI, more data will be available faster to support clinical development teams and help ensure the right decisions are made earlier. To make the best use of supplementary AI data, we need both the right tools to generate the data and the right people to utilise those insights in a practical and effective way.

I believe that we will start to see the adoption of AI as an additional tool used to support clinical development. We were impressed by the insights’s AI-enabled causal modelling could offer, and whilst I am sure this space will grow over time, I have not yet seen another AI algorithm that has been used successfully to predict clinical outcomes in the same way.

When looking at modelling, there is going to be a reliance on the quality and quantity of data available. Can you give us a bit of information about your data, and what drug developers should be looking for if they are considering causal modelling?

Joern Klinger, CEO,
We work with both large public biobank data and private data from pharma companies. A lot of voices in the industry have been claiming that the private data that pharma uses internally is superior, but the opposite is true and it’s not even close. The biobanks have done a great job of cleaning up the raw data. We combine that base of data with smaller individual data sets for specific diseases to get a good balance. We always use two independent data sets for each modelling problem.

Our next blog will discuss whether there is support for rare indications and how implementing AI in clinical trials can improve outcomes for patients. orphan drug status

[1] Wieland, A (2020) Regulatory Science & Clinical Research: Where Do We Stand (And Where Are We Going)?

Back to Blog Archive