Only 12 of drugs that undergo clinical trials end up being approved for commercial launch. The average development time of a drug throughout all its cycles lasts eight years on average. This is why the pharmaceutical industry has always been structured along long cycles and why returns on capital employed have been hampered by low success rates during development phases. And yet, we expect artificial intelligence and machine learning to move this industry into a new era.
Three categories of AI/ML benefits have already been found:
1: Identifying innovative therapies
2: Shortening drug development times
3: Raising the likelihood of success of molecules in clinical trial phases.
In particular, machine learning algorithms will provide major benefits 1- in setting the right dosage of the tested drug so as to find the right balance between safety and efficacy, and 2 – in putting together patient cohorts for clinical trials that offer the desired combination of characteristics (in terms of phenotypes and genotypes).
Based on these findings, the world’s 10 largest pharmaceutical groups have all made small bolt on acquisitions of artificial intelligence start ups. These digital nuggets have been incorporated into the R&D workflows that had been built up in recent decades in mode.
Unsurprisingly, integration has had rather uneven successes, as pharma majors have run into the following obstacles 1 a lack of know how in machine learning techniques among senior drug developers and 2 significant losses of speed in the development process, as entire swaths of the chain are still in analogue mode (unlike fully digitalised companies).
As a result, higher success rates from using artificial intelligence in drug development are now found among companies that, from the start, built fully digitalised development models based on artificial intelligence and its derivatives. The most famous example of this strategy is currently Moderna thanks to its breakthrough mRNA 1273 Covid vaccine and the resulting surge in economic and stock market value in just a few years. There are several other companies, both listed and non listed, that have placed artificial intelligence at the heart of their strategy.
These include 1 Recursion Pharmaceuticals, which will apply machine learning algorithms to proprietary series of biological and chemical data in developing new therapies and 2 Kronos Bio, which will use highly elaborate computational models to attack cancerous targets that had until now withstood all forms of treatment.
Moderna offers the most complete example thus far
“We developed a Covid vaccine in two months This was 90 faster than the normal 20 month development time” This is how Stéphane Bancel describes the process that began with identifying a SARS coronavirus and lasted until the start of phase 1 clinical trials in the first quarter of 2020 An accomplishment like this is not due to chance or to the individual skills of ingenious researchers, but rather to Moderna’s development of a suite of software and algorithmic tools that have digitalised the entire drug development value chain (from messenger RNA sequencing to the study of its chemical and geometric properties to its manufacturing).
Through its fully digitalised approach and its 10 years of experience in this technology, Moderna is its flagship representative In addition to its Covid 19 vaccine, Moderna’s pipeline contains various programmes at the clinical or pre-clinical testing stage, like other prophylactic (flu type) vaccines, cancer vaccines, and drugs that fight some rare diseases We are confident that this pipeline of indications that are treatable using messenger RNA will continue to expand in future years.
To accompany its growth, Moderna possesses production capacities in Massachusetts, all of which are also fully digitalised and headed by the former head of manufacturing at Novartis. The road ahead will nonetheless be long and bumpy. While messenger RNA offers precious advantages (such as being a programmable, i e sequencing therapy), its potential therapeutic use is constrained by 1 how long its effect lasts (for messenger RNA to lead to a drug it must generate proteins, and this effect may last only a short time) 2 the immunity system’s reaction to the insertion of messenger RNA into a cell is not always what is desired by researchers working in these areas and 3 the mode of delivery and genetic modification produced by messenger RNA is lipide nanoparticles ( which have the advantage of non-toxicity (unlike other approaches) but also the drawback of less effective transport faculties of the transforming gene than other transport technologies (such as virus based ones).