Drug Development on Fast Track with A.I. and Deep Learning

HAIFA, ISRAEL (August 22, 2018) – Dr. Kira Radinsky and Shahar Harel of the Technion-Israel Institute of Technology Computer Science Department have developed a smart system for the development of new drugs. Founded on artificial intelligence and deep learning, the system is expected to dramatically shorten and reduce the costs of drug development. It will be presented this week during the KDD 2018 conference in London.

Drug production is a costly and lengthy process. How costly and lengthy? Costs of half a billion to 2.5 billion dollars per drug, over 10-15 years are common numbers in the world of pharmacology.

Dr. Kira Radinksy

Dr. Kira Radinksy

In the past, new drugs were serendipitously discovered, with the discovery of penicillin being the most famous example. But modern-day processes are computerized and more systematic, beginning with a screen of many molecules and selection of those that have the greatest therapeutic potential.

The problem is that between 1023  and 1060 molecules show therapeutic potential. In comparison, there are an estimated 1022 stars in the galaxy. Therefore, the development process is typically shortened by narrowing down the initial breadth to molecules that feature the desired properties, which still leaves an enormous number of molecules.

In the study that will be presented at KDD 2018, the Technion researchers formulated a new approach for the generation of therapeutic molecule candidates. According to Shahar Harel, their working hypothesis is that “the drug development-related organic chemistry vocabulary is similar to that of a natural language. The system that we developed, which uses artificial intelligence and deep learning, acquired this language based on hundreds of thousands of molecules. In addition, we supplied it with the chemical composition of all drugs approved up until 1950, which served as the prototypes upon which it generated new variations – new potential drugs. In order to generate a creative system, we deliberately introduced “noise”, which yields diversity. Namely, the system will generate many variations of existing drugs.”

Shahar Harel

Shahar Harel

Thus, the system is based on a pharmacological language, data on existing drugs and a creativity-promoting mechanism. When they instructed the system to propose 1000 drugs based upon old drugs, the researchers were surprised to discover that 35 of the new drugs generated by the system are existing, FDA-approved drugs developed and approved after 1950. In other words, the investigators demonstrated the system’s efficiency in developing ‘rational’ or valid drugs.

“We are essentially presenting here an algorithm which addresses the creative stage of drug development – the molecule discovery stage,” said Harel. “This capacity leans upon our mathematical innovation, which enables the computer to understand the chemical language and to generate new molecules based upon a prototype.”

According to Dr. Radinsky, “What we’ve presented here is not only a means of streamlining existing methods, but also entirely new drug development and scientific practice paradigms. Instead of seeking out specific correlations based upon hypotheses we formulate, we allow the computer to identify these connections from within a massive sample size, without guidance. The computer is not smarter than man, but it can cope with huge amounts of data and find unexpected correlations. This is how we managed to find (in another study) the unknown side effects of various drugs and drug combinations, and now, an innovative drug development mechanism. This is a novel type of science which is not built upon hypotheses tested in an experiment, rather, upon data that generated the research hypothesis.”

The significance of this breakthrough is particularly great in the face of Eroom’s Law, which asserts that the number of new drugs approved by the FDA should decline at a rate of approximately 50% every 9 years (ratio between the number of new drugs and the investment in research and development). The term Eroom was coined in 2012 in an article published in Nature Reviews Drug Discovery, and is a reverse order of Moore, the name of Gordon Moore, one of the founders of Intel. Moore observed that the number of transistors in a dense integrated circuit doubles every two years. In contrast, Eroom’s Law notes that each year, fewer and fewer drugs are marketed.

Dr. Radinksy projects that “this new development will accelerate and reduce costs of development of new and effective drugs, thereby shortening the time patients will have to wait for the drugs. In addition, this breakthrough is expected to lead to the development of drugs that would not have been generated with the conventional pharmacological paradigm.”