The increasing adoption of AI in drug discovery and development will likely reduce the time required to commercialise new treatments. Within this process, AI is being used in areas including target identification, drug design, drug repurposing, and clinical trials. With up to 70% of life sciences companies using AI in research and development according to DLA Piper’s AI Governance Report, this departure from more traditional, resource-intensive approaches to drug discovery and development should accelerate the launch of new treatments onto the market.
The current process for drug discovery and development is widely regarded as expensive, slow and inefficient. On average, it costs approximately USD1.3 bilion and 10 years to bring a new therapeutic drug to market.1 AI can play a role to reduce drug discovery costs and timelines, and increase the probability of success, with potential application at numerous stages of the drug discovery and development process.
However, the implementation and use of AI in drug discovery and development and the life sciences industry more broadly is not without risks and challenges. Issues including privacy and the ethical use of AI are likely to be key concerns for organisations involved in drug discovery and development, regulators and consumers alike. Notwithstanding the considerable benefits that AI will bring to this field, as well as the risks, many regulators around the world have yet to enact specific laws relating to the implementation and use of AI in this area.
AI is shaving years from the drug discovery and development process
Drug discovery and development are being transformed by the life sciences industry’s use of AI. In 2024, a study found that AI technologies such as machine learning algorithms were used to develop 164 investigational drugs and one approved drug, with the most common target treatment types being anticancer and neurological treatments.2
When it comes to target identification and drug design, AI addresses inefficiencies in traditional methods, including reducing costs, streamlining processes and potentially improving success rates. A unique advantage of AI is its ability to quickly analyse large datasets and potentially uncover hidden patterns and relationships, which traditional methods may overlook. With natural language processing algorithms able to scan and analyse millions of documents (such as research papers and patents), the equivalent of innumerable years of desk-based research can be carried out relatively quickly and easily and connections can be made which humans would have no hope of finding on their own, and may even not have thought to look for.
For example, a company focusing on precision oncology reportedly used AI models to find novel oncology biomarkers and targets, reducing the timeline for drug design from 4-7 years to just 3 years.3 A team at MIT recently revealed that, using AI, researchers have designed novel antibiotics that can combat two hard-to-treat infections: drug-resistant Neisseria gonorrhoeae and multi-drug-resistant Staphylococcus aureus (MRSA).4 The candidates are structurally distinct from any existing antibiotics and show that, by using AI to generate hypothetically possible molecules that don’t exist or haven’t been discovered, it should be possible to explore a much greater diversity of potential drug compounds.
Following target identification, AI can also be used to assess possible drug-drug interactions and predict the efficacy and toxicity of new compounds, freeing research from reliance on the use of animal models as a proxy for experimentation in humans. The adoption of these capabilities in the pre-clinical stages could help reduce the number of drug candidates being discarded due to unexpected adverse effects in clinical trials.5
In clinical trials, life sciences companies could leverage AI to streamline recruitment and optimise trial design and data management. For instance, AI can help identify suitable participants by analysing large volumes of electronic health data and optimise trial designs by digitally simulating test scenarios and predicting outcomes. In turn, AI can help determine appropriate dosages and treatment combinations for each participant group.6 Using AI to analyse and manage trial data could also accelerate the process of preparing submissions for regulatory approval for a new treatment.
A number of companies, such as Exscientia and BenevolentAI, are seeking to leverage the power of AI in drug discovery and development through innovation in the creation of commercially available drug discovery platforms. Partnerships between these AI-focussed technology companies and major pharmaceutical companies are on the increase and yielding promising candidates in diverse therapeutic areas. Some of these AI platforms offer features to streamline every aspect of the process – from target identification, through optimisation, to clinical trial matching.
As AI models in the life sciences industry become more sophisticated, the timeline for drug discovery and development is shortening significantly and the chances of identifying a drug candidate that will be successfully brought to market are improving. AI is also advancing other areas, including drug repurposing and personalised medicine. In particular, personalised medicine is a growing trend that can be supported by AI’s vast analytical capabilities, including the development of individualised treatment plans that minimise the risk of adverse reactions.
Tackling the risks and challenges
However, as with any new technology, there are clearly risks and challenges that affect reliance on AI in the drug discovery and development process. An added challenge is that many jurisdictions, outside of the EU, including Australia, are yet to enact AI-specific legislation.
Privacy
AI use in the drug discovery and development process raises significant privacy concerns given large datasets are used to train AI models. Existing privacy and data protection laws may go some way toward protecting individuals’ privacy, including in respect of AI use.
In Australia, privacy obligations are set out in the Privacy Act 1988 (Cth). When it comes to AI use, some key considerations highlighted by the Australian privacy regulator are:
- personal information may not automatically be used to train generative AI models;
- privacy obligations apply to any personal information input into an AI system and the system’s output, including any inferred, incorrect or artificially generated information produced (such as hallucinations); and
- developers of in-house AI models should take reasonable steps to ensure accuracy in their generative AI models.
As AI develops globally, companies in the life sciences space should be attuned to regulatory and legislative changes in each market in which they operate.
Fair and ethical use of AI
AI use in drug discovery and development must be fair and ethical given the risks posed by AI’s ability to mine, process and analyse data quickly. For example, bias in AI algorithms such as the identification of inadvertent patterns in toxicity predictions may cause certain groups’ ineligibility for clinical trials, potentially resulting in unequal access to medical treatment. To address fairness and ethics concerns, some jurisdictions like the EU have enacted legislation, while others have issued regulatory guidance. For example, the Australian Government has proposed mandatory guardrails for AI use in high-risk settings (e.g. settings that may adversely impact an individual’s human rights, health or safety).7 In the UK, laws relating to the regulation of the use of AI systems are currently progressing through the legislative system.
Impact on IP Protection
A developing issue that will only become more acute with the increasing use of AI in drug discovery and development and with the increasing sophistication of AI systems, is the extent to which the AI platforms themselves, and the resulting drugs, will be patentable. When it comes to protecting innovation in the AI-driven drug discovery platforms, the obligations arising from privacy and data protection laws may throw a spanner into the works, potentially tying the hands of would-be patentees when it comes to ensuring that sufficient information is provided in the patent specification.
A fundamental principle of patent law is that a patent must disclose the invention in a manner that is clear and complete enough for a person skilled in the relevant technical field to put the invention into practice without undue burden or further inventive step. Depending on the invention, this could require inclusion of underlying algorithms, training steps, and training datasets. Sufficient disclosure may prove impossible where data privacy laws prevent the disclosure of essential information, such as necessary details of training datasets, potentially preventing valuable patent protection.
DLA Piper’s AI Laws of the World
Maximising benefits while mitigating risks (including relating to privacy, fairness and ethics) is the balancing act that governments are grappling around the world when it comes to AI development. Some jurisdictions have implemented AI-specific laws, while others are in the process of drafting regulations. Many are currently relying on existing legal frameworks to address AI-related concerns. DLA Piper’s AI Laws of the World provides an overview of AI laws and proposed regulations across more than 40 countries (including Australia), including key legislative developments, regulations, proposed bills and guidelines issued by governmental bodies. If you have any queries regarding AI and its evolving regulation, or regarding life sciences, please get in touch with us or your regular DLA Piper contact.