Within 10 years 60% of drugs will be developed with artificial intelligence

Within ten years, 60% of new drugs could be designed with artificial intelligence. This is the prediction contained in the Capgemini Research Institute's new report on the digital transformation of the biopharmaceutical sector, which captures a picture of an industry grappling with rising research and development costs, long lead times and declining productivity. According to the survey of 500 executives from pharmaceutical and biotech companies in Europe, the US and Asia, generative Ai and machine learning are already accelerating the discovery of new therapeutic targets, compound design and clinical trial management. The goal is to reduce failures and bring efficiency back into the pipeline. 'For the first time in 20 years we are seeing a technology that can really impact research productivity,' notes a biotech executive involved in the survey.

The economic picture explains the urgency: bringing a drug to market can cost over two and a half billion dollars and take up to 10 years, while nine out of ten molecules fail in the clinical phases. For many managers, artificial intelligence represents the only way to reverse the so-called Eroom's law, which describes the opposite trend to Moore's law: instead of increasing, scientific productivity tends to decrease over time. It is therefore not surprising that 82% of the sample expects a radical transformation of R&D within five years.

"Ia is not a fad: it is the only way to keep costs and time compatible with the needs of patients and regulators," says an R&D manager of a large multinational drug company. The adoption of Ia involves the entire value chain. In the pre-clinical phase, models analyse large omics datasets to identify targets and predict toxicity; in clinical trials they help select centres, identify eligible patients and estimate the likelihood of adverse events, improving recruitment. "The most critical part is not the trial itself, but the recruitment of patients. This is where Ia is already making a difference,' says a clinical trials executive.

In regulatory affairs, Ia automates the preparation of dossiers and anticipates the authorities' requests, while in production it optimises parameters and reduces waste. The report shows that companies are moving towards a hybrid development model, combining in-house expertise and collaborations with start-ups and universities.

"Digital platforms are becoming strategic infrastructures as much as production facilities," says an industrial manager of a large pharmaceutical group. Fifty-two per cent of the sample already have strategic agreements in place, while a third consider possible acquisitions to accelerate the integration of digital platforms. The most mature use cases are concentrated in discovery, but those related to data management, pharmacovigilance and demand forecasting are growing.