Throughout the history of medicine, serendipity has played an unexpected yet pivotal role in drug repurposing, leading to remarkable discoveries that have saved lives and transformed patient outcomes. One notable instance is the discovery of sildenafil, originally developed as an anti-hypertensive, which was later found to have potent effects on erectile dysfunction, leading to the blockbuster drug Viagra. Similarly, minoxidil, initially intended for hypertension management, revealed an unforeseen side effect of stimulating hair growth and found its place in the treatment of baldness. These instances underscore the unpredictability and potential of repurposing existing drugs, often originating from unexpected observations, to address diverse medical challenges.

But what if we could transform these unexpected, serendipitous discoveries, into a re-usable methodology that does not involve luck, in order to develop therapies which could benefit even more patients?

The need for drug repurposing

In the world of healthcare, creating new drugs from scratch is time-consuming, expensive, and full of challenges. Drug repurposing or, simply, drug repositioning/rediscovery or reprofiling is a strategy to identify advanced uses for pre approved drugs or existing medications. The basic idea of repurposing is to discover new useful activity for a distinct malady in an older clinically used drug or one that failed in later stages of development (Chong and Sullivan, 2007).

The societal imperative for innovative treatments is compelling. Millions of patients worldwide suffer from diseases that lack effective therapeutic options.

For those of us - and we are many - who face a pressing need for a therapy, either for ourselves or a dear one, drug repurposing can be a source of hope, improved quality of life and, I dare to add: happiness.

The economic implications of drug development are also significant. The conventional route of creating novel drugs involves exorbitant costs, lengthy timelines, and high failure rates. Drug repurposing and repositioning offers an avenue to optimize these resources. Repurposing does not only potentially accelerate time to market but also makes novel therapies accessible to a broader patient population.

Let’s delve into how AI is speeding up the process of discovering new uses for existing drugs.

Machine Learning's Role in Small Molecule Exploration

AI/ML in drug discovery is not new, but thanks to the availability of data, the speed of discovery of new techniques, and the availability of computing power, it has become over the past years a necessary tool for handling complex molecular data, including structures, activities, and clinical information. Its efficiency allows it to find connections that might have been missed using traditional methods. This makes it a powerful tool for discovering small molecules with new uses. AI-based drug repurposing however faces its own set of challenges. Ensuring the quality and diversity of input data for AI algorithms is crucial for accurate predictions. Interpretability of predictions can often be a real challenge. Multiple data sources (literature, patents, readouts, -omics data, EHR records…) can lead to complex results that are difficult to interpret and validate in a wet lab, resulting sometimes in more questions and hypotheses than answers. This, as a result, delays the time to discovery and, as a consequence, the time to treatment.

How we tackle drug repurposing and repositioning using AI

At Kantify, our unwavering mission is to expedite the discovery of life-changing remedies for patients in need. Our efforts and energy over the past two years have been dedicated to crafting an artificial intelligence capable of comprehending the intricate interactions between small molecule drugs and human biology. This AI marvel, aptly named Sapian, consistently astounds us with its unparalleled ability to propel scientific progress and unveil concealed insights. Sapian stands as a remarkably accurate tool, hinging on millions of meticulously collected data points, yet remarkably capable of delivering intelligible answers even to those unfamiliar with of machine learning.

We are still early in our journey. There is so much to uncover, and so many diseases still do not have a therapy. However, the proof of the pudding is in eating it, and we have had extremely promising results, showing that with our drug discovery technology, we can actively contribute to drug repurposing, at scale.

For example, in 2022, we started with technological developments and interesting collaborations focused on repurposing, such as our work with the team of Dr. Xavier Nissan at the French Research Centre for Stem Cell research on Finding repurposed treatments for rare neuromuscular disorders, resulting in the discovery and validation of treatments that could help patients facing severe and currently untreatable degenerative disorders.

This project is far from a one-off. For example, in a recent project whose results we haven’t published yet, we used our technology to discover a new, promising indication for a stage II oncology drug, which we subsequently validated in vitro. More surprisingly, we believe to have elucidated parts of the mechanism of action of what makes this drug effective in the oncological setting. For the drug, the exact mechanism of action was only partially understood by the company developing the drug candidate.

Bye bye, serendipity?

We should certainly not abandon serendipitous discoveries. But because AI is actionable, repeatable, and scalable, we see it as a promising and exciting complement to serendipity.

Through collaborations with drug hunters, AI specialists like us have the potential to uncover faster and more effective treatments.

If you are interested to better understand how to use our AI technology for a repurposing case, or to start a conversation, please get in touch with us via the form below.