AI’s already found so many applications – from commerce to travel to entertainment – that we shouldn’t be surprised when a new development reveals a previously unforeseen use for the new technology. AI has seen plenty of interest and support from large companies and high net worth individuals, including our own founder, tech billionaire and entrepreneur Tej Kohli (net worth: £6 billion), who set up Tej Kohli Ventures to channel his resources into this exciting field. Recently one particularly interesting new application for AI has been discovered – and it may be the first step on the path to a cure for cancer. Researchers have found that AI can be used to develop exciting new drugs which can tackle cancer effectively. Here’s how the new technique works, and what impact it could have on the future of disease.
How AI Can Help Develop New Drugs
First, let’s look at how pharmaceutical companies will usually go about designing a new drug. It’s important to bear in mind that the range of potential drugs is vast. Simply based on the possible arrangements of small organic molecules, there are far more than we could ever reasonably test in a research laboratory – one report put the number at 1060 distinct possible drugs. As a comparison, the number of grains of sand on earth is about 7.5 x 1018. The number of drugs we could research in principle isn’t just more than all the grains of sand our planet holds, it’s astronomically larger than that.
Of course, most of those drugs won’t actually have an effect that we’re looking for. That makes searching for a useful drug a fiendishly difficult task – if you were doing it at random, simply testing each and every possible drug, then it would be like looking for a needle in a haystack the size of the planet.
At the heart of modern pharmaceutical research is an attempt to cut down on the potential drugs to test – if we can eliminate broad classes of the possibilities out there, we can focus on the options that are more likely to work, saving immense time and effort. At the moment, though, it’s a haphazard process at best. Sometimes, we can proceed from new insights into a disease to decide what area of pharmaceuticals to investigate. Researchers may test many different variations of a molecular compound to see if any work. But often, we just get lucky, and find that a drug which had been used for something else has an unexpected yet beneficial side-effect.
Even when we use our knowledge of a disease to direct our efforts, there is a certain amount of guesswork involved. Much pharmaceutical research requires ‘brute forcing’ the problem – testing drug after drug until we find one that works. Medical insights can cut down the space of drugs we have to test, but not completely. Only 1% of cancer drugs that reach clinical trials ever prove to be effective.
With that in mind, we’re now in a position to understand the real importance of AI to fighting disease. Researchers have used an algorithm to discover a new drug – BPM 31510 – which is expected to be effective in tackling cancer. The algorithm worked by taking data from human tissue samples, both healthy and cancerous, and processing it until it reached a possible drug for treating cancer. Because the analysis happened digitally, the AI was able to consider and discard far more possibilities than could have been done otherwise, before any drug made it to clinical testing. The AI is better than human researchers at filtering out useless drugs to find the ones we need. It’s better at sorting the worthwhile treatments from the bad because it can analyse lots of data very quickly.
BPM 31510 has now reached Phase II trials, and there’s reason to be optimistic about its prospects. But regardless of the fate of this particular drug, the applications of AI to medicine are very exciting. The ability to filter through far more potential drugs than human researchers means it will have a core place at medical studies in the future. The first approved AI-developed drug may be just around the corner.