New drugs: artificial intelligence as a development worker?

Status: 01/17/2023 1:11 p.m

The development of new drugs is becoming more and more complex and less and less worthwhile. In the future, artificial intelligence could help pharmaceutical companies bring new products to market faster.

Alexander Fleming discovered one of the most important antibiotics of all, penicillin, because the culture medium of one of his bacterial cultures went moldy. Such accidental discoveries are rare today. Pharmaceutical companies have to invest billions in drug development. Nevertheless, around 90 percent of active ingredients fail in the clinical test phase alone, when development is almost complete.

At the same time, we need more and more new medicines: the growing world population increases the risk of new pathogens developing and spreading. Since the beginning of the 20th century, up to 100 million people have died from the flu, corona and HIV pandemics alone – and bacteria are constantly developing new antibiotic resistance.

AI can significantly shorten the development process

Artificial intelligence trained with data on the efficacy, bioavailability, and side effects of known drugs can help researchers predict these factors in newly discovered drugs.

New active ingredients often fail because it turns out that they inhibit the production of the vital so-called cytochrome P450 (CYP450) in the liver. These are important for processing water-insoluble substances in the liver. They are also involved in the synthesis of certain hormones and other substances.

Since it is often not exactly known why cytochrome production is inhibited, it has only been possible to predict whether an active substance inhibits CYP450 production with a success rate of around 60 to 70 percent. With artificial intelligence, the rate can be increased to 95 percent. As a result, far fewer active ingredients have to go through the preclinical and clinical test phases. This means that in the future, drugs with fewer side effects could be developed faster and more effectively.

A decillion of possible drugs

Many active ingredients have been and are still being discovered in nature. But this reservoir will eventually be exhausted. In addition, synthetic active ingredients can often have a much more targeted effect on a specific pathogen and resistance can be circumvented. It is estimated that there are about a decillion molecules that could be used as active ingredients.

A decillion – that’s a one followed by 60 zeros. An AI can’t study all of these molecules, but it can certainly study more of them faster than humans ever could.

AI molecule design

Artificial intelligence can also help locate these molecules in the first place. With the large amount, it makes little sense to go through all the molecules one by one and test for all possible disease triggers. For example, if it is known that a certain protein is involved in a disease, AI can be used to develop a molecule that binds to this protein and thus renders it harmless.

AI is also used in the development of new mRNA-based cancer therapies. In December, the US company Moderna published a clinical study in which an AI selected surface proteins of skin cancer cells, which were then vaccinated against using mRNA. On January 10, the Mainz-based pharmaceutical manufacturer announced BioNTech, the British AI company InstaDeep to expand drug development in-house using AI. Both technologies, AI and mRNA, could also lead to more personalized medicine in the future.

AI can also develop bioweapons

The US company Collaborations Pharmaceuticals tried to turn the tables: Instead of drugs, they tried to use their drug AI to develop molecules that were as toxic as possible. Without knowing the substance, the AI ​​developed, among other things, the extremely potent neurotoxin VX. The results of this experiment were subsequently destroyed, but one thing is clear: in the wrong hands, such a tool can also cause damage.

Still music of the future

It is unclear how successful the use of AI in the pharmaceutical industry will actually be. Some active ingredients designed with the help of AI are already being investigated in clinical trials. The British company Exscientia is testing several such substances, for example the cancer drug EXS-21546.

However, it remains to be seen whether these substances are actually better and have fewer side effects than traditionally developed drugs. It is also open whether the development is faster and more cost-effective. The use of AI in the pharmaceutical industry is still in its infancy – and it will still take time before it can keep up with conventional methods.

source site