Opinion: Fast tracking drug discovery in Australia with GenAI

Australia is at the forefront of medical research, with a thriving sector that underpins the nation’s health.

But translating drug discoveries into commercially viable candidates remains a challenge, with the preclinical stage alone taking up to six years and potentially costing billions of dollars, according to the UNSW Sydney.

Despite these significant investments in time and money, a large majority of drug candidates fail to pass clinical development and regulatory approval, so any technology that can reduce this will quickly provide a return on investment (ROI).

This is where AI, particularly generative AI is changing the game in early stages of drug discovery.

Early analysis of AI-discovered drugs in clinical trials showed an 80-90 percent success rate in Phase I, when the drug is tested with a small group of people for the first time, according to ScienceDirect. This is substantially higher than historic industry averages.

Recent high-profile failures in clinical trials, however, have brought some needed balance to the hype surrounding this topic.

Despite this, the clear potential outweighs the initial failures and is where Australia’s pharmaceutical and biotech industry should focus investment, especially given the Australian Government’s mission to deliver a future made in Australia.

With $1.5 billion earmarked in the recent Federal Budget to boost medical science manufacturing, there’s a great opportunity for investment in AI-enabled drug discovery, as well as in new skills required to achieve this.

There are already great instances of this happening, with the CSIRO a perfect example. A cross-disciplinary team is currently developing new AI tools to accelerate the drug discovery process for emerging infectious diseases.

The CSIRO “envisions a future where data science is deeply embedded into the drug discovery process, giving us rapid and cost-effective access to AI-derived next generation medicines for a range of currently untreatable diseases.”

How generative AI is being used

Generative AI is primarily being used in preclinical drug development to analyse large datasets and identify potential drug targets; identify molecules that may bind to a target protein; predict properties of lead compounds; and improve drug solubility, stability and bioavailability.

Training generative AI models on large datasets of biochemical structures and their associated properties allows the models to learn patterns and generate new molecules with desired characteristics.

This not only accelerates the drug discovery process but potentially leads to the development of breakthrough therapies that address unmet medical needs.

Gartner predicts at least one generative AI-discovered drug candidate will reach Phase III clinical stage by 2025, which is when the safety and efficacy of the drug is tested on thousands of participants.

Overcoming data bottlenecks

Current shortcomings in the quantity, quality and accessibility of data are hindering efforts to operationalise generative AI drug discovery at scale for many organisations.

Generative AI systems need to generate or access large amounts of labelled, curated, high-quality, contextually rich data from individual experiments.

But the reality for most biopharma is that data is far from that, and no amount of technical progress in AI can compensate for poor data foundations.

Addressing the AI-readiness of data in drug development should be the focus of Australia’s investment in early R&D.

Otherwise, the ROI will not be improved despite the efficiency gains in drug candidate creation.

For the greatest business impact, organisations need IT to strengthen the underlying data structure as existing systems tend to be fragmented, unable to be harnessed at an enterprise scale because data is trapped in functional silos.

Furthermore, complex privacy controls, unstructured reports, limitations on storage and compute, and lack of robust metadata create obstacles to reliably train advanced generative AI models.

Gartner’s 2024 Gartner CIO and Technology Executive Survey highlighted that this isn’t being overlooked, with 26 percent of life science CIOs expecting research/scientific solutions to be a top priority for new or additional funding this year.

In addition, 62 percent will increase funding in cloud platforms.

To fully embrace next-generation computational approaches, particularly generative AI, organisations must focus on using findable, accessible, interoperable and reusable data principles.

It’s important they make strategic investments to curate quality datasets, migrate core data to cloud-ready data lakes, create data catalogues and institute data governance.

By 2026, Gartner expects drug discovery will reach an inflection point where in silico (computer designed) first strategies overtake traditional laboratory research at major pharmaceutical companies.

Skills development

The successful development, implementation and governance of generative AI drug discovery initiatives and projects is going to take teams from across the organisation to provide the necessary expertise, guidance and support.

Organisations can move from ad hoc to strategic generative AI deployment by establishing an AI centre of excellence that consists of a broad range of stakeholders including R&D, business, risk, IT and analytics.

One of the main responsibilities should be to promote collaboration within the organisation and effectively manage strategic skills.

Existing workforces will also need to be reskilled to ensure they have both the traditional research understanding and computational skills required for the future requirements of generative AI-enabled drug discovery and development.

Ultimately, for generative AI in drug discovery to deliver on its promise, data must be the operational focus, supported by a greater investment in skills and cross-functional collaboration.

About the author

Reuben Harwood is a director analyst at Gartner, focused on the role of AI in drug discovery, next-generation therapeutics and precision medicine. The latest AI trends will be discussed during the Gartner Data & Analytics Summit in Sydney (29-30 July).