Ask The Expert about AI in Drug Development with Quibim
Drug development is the process of developing a new pharmaceutical drug once a lead compound has been identified through drug discovery.
We believe that we will witness revolutionary changes for patients in the next five years as artificial intelligence is integrated into drug discovery and development.
In this article, we’ve sat down with Quibim, our portfolio company, to discuss their approach to using AI to discover and support the development of novel drugs.
Quibim designs pioneering tools that unlock imaging data to improve patient outcomes. Its main purpose is the application of AI techniques to medical images from MRI, CT, and PET to unlock new data which can be transformed into actionable predictions for life sciences and providers.
Dr Angel Alberich - Bayarri, PhD, Founder and CEO of Quibim, talks us through their approach to using AI to transform imaging data into actionable insights.
Dr. Angel Alberich - Bayarri, Phd, Founder and CEO of Quibim
Using AI to transform imaging data into actionable insights
Dr. Angel Alberich-Bayarri, Ph.D., Quibim: We see the future of medicine mainly being backed by three pillars:
Finding patients who benefit the most from specific treatments.
We are aware that biomarker-based medicines account for 25-42% of drugs approved by the FDA in the past eight years.
Maximising drug development success
Imaging data can accelerate trials by 30-40%, identifying patients likely to experience benefit or adverse events. That means there is a test that also indicates treatment, and there is more and more evidence that real-world data can be used to reduce drug development costs by around 30%.
Enabling precision medicine
Imaging feature-based diagnostics can reach more patients by addressing the limitations of traditional diagnostics. Shortening clinical trials from around 10 years to seven years has been possible thanks to the real-world data that we might find in electronic records.
At the same time, imaging can be considered real-world data. It's a standard of care, and it's universally accessible. We perform image-based searches to enable precision medicine.
We know that molecular biology is the established technique to advance companion diagnostics. But imaging can also play a significant role.
Current challenges in precision medicine
We've seen that genomics and pathomics have led to significant breakthroughs - they are now the standard of care for many treatments, for example in lung cancer.
But molecular biology application depends on a procedure that includes analysing a biopsy sample. But the sample is not representative of the huge heterogeneity of the lesion and of the patient.
Around 64% of patients who could be eligible for advanced lung cancer treatments or personalised treatments are not being tested, one of the reasons is because the doctors need to make decisions fast and can’t wait for weeks for a result. So they just start the treatment. There is also a big problem in terms of lack of accessibility for patients do to the high costs of tests.
How can we use imaging to complement these deficits?
Despite 10 Million daily imaging exams worldwide, less than 0.1% are used in clinical research or creating Al models.
Most of the developments in imaging have been centred on using AI for data acquisition to accelerate ways of reconstructing images, or AI has been applied to workflows to flag exams that need to be reported faster or more immediately because of urgency.
We still think that there is a missed opportunity to connect those imaging patterns with what is going to happen to the patient, not just becoming a workflow improvement tool, not only trying to help radiologists but also trying to find patterns that might be helpful from a prediction/prognosis standpoint in a tumour board or for clinicians.
At Quibim, we are on a mission to be able to convert imaging into companion diagnostics in the future. We are working towards enabling clinicians to indicate a treatment after the image analysis.
But right now, we are concentrating on providing organ and lesion AI segmentation models and establishing research collaborations with BioPharma to create tools for drug development.
These tools are mainly used in clinical trials and would allow for better patient selection as candidates for trials.
Our next step is to convert these tools into not only radiology software, but software that any oncologist or immunology expert could use in clinical routine to predict survival rates or treatment response in patients.
Quibim covers the use of imaging from population to patient, translating research to clinical practice.
Our product QP Insights is a platform that allows to load a vast amount of imaging data and convert it to patterns linking with clinical endpoints.
Once we have this evidence on new biomarkers, new methods, and new developments, we convert it into products like QP-Prostate and QP-Brain that can analyse the data of a single individual.
Additionally, we are building predictive models that will likely become diagnostic tools in the coming few years.
Quibim’s development process creates a steady pipeline of products and features along the Al value chain
We are quite advanced in developing data imaging products to diagnose prostate, lung, breast and colorectal cancer. A new brain imaging product is imminent, and in the coming months, we will be focusing on liver imaging.
On the research side, we are engaged at a European level with the European Commission in many initiatives, the biggest one being The European Cancer Imaging initiative, Cancer Image Europe.
The future of drug development using AI promises to revolutionise the pharmaceutical industry by significantly reducing the time and cost associated with discovering new drugs and treatments.
Keep an eye on our blog posts and sign up for our monthly newsletter to get the latest updates on AI drug development.