Latest news: artificial intelligence in pharma

Amid rising levels of artificial intelligence (AI) use and application, the European Medicines Agency (EMA) has issued a draft paper outlining its view on the use of AI and machine learning (ML) in various stages of a medicine’s life cycle.

The paper, a part of a joint Human Medicines Agency (HMA)-EMA initiative to develop data-driven regulation, highlights the promise that AI/ML capabilities can bring to all steps of a medicine’s life cycle but warns of measures that should be taken by companies to ensure its legal and ethical use. 

The European Union (EU) has already drafted an AI law, in what will be the world’s first comprehensive law for the technology. And while regulations certainly exist for AI use in medical devices, the pharmaceutical industry lies in more of a grey zone.

AI and ML tools can be extremely useful in the medicinal product life cycle. AI platforms can be used in the drug discovery process, and modelling approaches can be employed, which would change the use of animal models in preclinical development. The harnessing of data by AI/ML in clinical trials is already in use, and AI/ML can even be used at the market-authorisation and post-authorisation stages to help with product information compilation and pharmacovigilance activities.

The paper outlines that companies using AI/ML at any stage of a medicine’s life cycle should be wary of existing legal frameworks and consider limitations or challenges that using the technology might have. These include issues around bias, overfitting, and data protection. An overarching theme of the paper is that companies using AI should always interact with regulators and operate within a “risk-based approach”.

The EMA was keen to state that it is not within its remit to regulate AI/ML software used in medical devices. However, it did add that when using CE-marked devices in a clinical trial, additional requirements might need to be checked off to ensure the integrity of data and results, along with the safety of subjects.

The National Cancer Institute (NCI), part of the US National Institutes of Health (NIH), has awarded $2m in funding to Enspectra Health through the Small Business Innovation Research (SBIR) programme.

The grant will fund the development of deep learning algorithms that can predict which pre-cancerous skin lesions are likely to turn into squamous cell carcinoma (SCC).

In a press release, Enspectra CEO Gabriel Sanchez said: “We are thrilled to have been awarded a Direct-to-Phase II SBIR grant from the NCI to support the development of predictive algorithms for skin precancers. This grant will accelerate our vision to noninvasively detect and monitor skin conditions earlier to advance care for the millions of patients with skin conditions.”

Pre-cancerous lesions, also known as actinic keratosis (AK), are caused by long-term exposure to the sun, particularly UV rays, and can progress to SCC. According to the Skin Cancer Foundation, only 5%-10% of AK can progress to cancer. However, the only way to diagnose which of the pre-cancerous skin lesions can progress to SCC is to do a skin biopsy.

Enspectra plans to use an imaging technology that combines reflectance confocal and multiphoton laser scanning microscopy to eliminate the need for biopsy. Enspecta will first create a digital histopathology database of patients with AK. The information collected will include data before any treatment and will follow patients through the treatment with topical therapy. As patients who are unresponsive to treatment are more likely to develop SCC, the algorithm would be trained to predict which patients are likely to be unresponsive to treatment, thus, are likely to develop SCC.

NVIDIA has announced a $50m private investment in public equity in tech-focused Recursion Pharmaceuticals to create artificial intelligence (AI)-assisted drug discovery models, sending Recursion’s stock to skyrocket.

This investment and partnership news was followed by a 116 % rise in Recursion’s stock price when markets opened on 12 July, compared to the previous day. 

The investment is accompanied by plans for collaboration to distribute these using NVIDIA cloud services and follows the strategic acquisition of Cyclica and Valence to enhance Recursion’s machine-learning and AI capabilities.

This year, there has been an increased focus on using AI in drug discovery among other applications. Several companies have touted their reliance on AI, with Insilico Medicine’s AI-discovered drug for idiopathic pulmonary fibrosis starting its Phase II trial. GlobalData analysts have identified more than 250 unique active drugs using terms such as “AI” or “machine learning” in their drug description in 2022, but observe that the use of AI in drug discovery is still in its infancy and is expected to mature in the coming years.

GlobalData is the parent company of Pharmaceutical Technology Focus.

Recursion plans to capitalise on this increased focus on AI and create and commercialise drug discovery models using BioNeMo, NVIDIA’s generative AI in drug discovery cloud service platform. The models will be trained on Recursion proprietary dataset using NVIDIA’s technology and expertise. These models will be used by Recursion internally and will be marketed on BioNeMo.

Multiple partnerships between tech and pharmaceutical companies for using AI in drug development have been announced last year, including a partnership between Novo Nordisk and Microsoft to generate new scientific insights and create forecast models of atherosclerosis development risk.