Accelerating vaccine development with AI

Despite the tremendous advances in medical science in recent decades, there is still so much about the human body that we’re yet to discover, reports Billy Sisk, Life Sciences Industry Manager, EMEA, Rockwell Automation

The mission of the life sciences industry, therefore, is to continually probe at the next frontier and expand our collective understanding. The response to infectious diseases is just one aspect of this; yet, it presents a hugely significant area in improving global health and life expectancy.

What typically separates acute diseases from chronic conditions is the sense of urgency. When a new strain of disease is identified, its viral nature — combined with the interconnectedness of modern societies — can quickly lead to an exponential rise in cases requiring treatment. This places immense pressure on governments and healthcare infrastructure as action is taken to restrict the spread and apply treatment so that normal social and economic life can be resumed.

Life sciences organisations have an essential role to play in this regard and, through the use of artificial intelligence (AI) and other advanced analytic technologies, can help to rapidly accelerate the path towards the development and dissemination of such treatments.

What’s involved in developing a vaccine?

The sense of urgency that comes with infectious diseases can itself fast-track progress towards arriving at a vaccine. The need for immediate action unites and galvanises a range of bodies — from researchers and clinicians through to regulatory bodies and manufacturers — in the quest to get an effective treatment into the hands of healthcare professionals as quickly as possible.

Even with these groups working tirelessly and with a determined focus, the end-to-end process can still take years. There are several stages that come before a treatment is ready to be put into the market. These include

  • exploration: working through thousands of potential compounds to shortlist vaccine candidates and investigate the immune response
  • preclinical stage: lab analysis to identify relevant antigens to arrive at a vaccine concept and design
  • clinical development: trialling the vaccine on test groups of varying characteristics
  • regulatory review and approval: verification of vaccine safety and compliance with health regulations
  • manufacturing and quality control (QC): the development of medicines in preparation for mass distribution.

Each of these steps is essential in ensuring that the vaccine produced is effective and safe, that any side-effects are properly understood and that it can be produced at scale on a consistent basis until the threat of illness has been sufficiently minimised.

Historically, the complexity, regulation and cost involved in each of these stages has slowed down the response to emerging health issues. Now, owing to advances in AI-related technologies, we have an opportunity to rapidly accelerate the process through which we can deploy treatments into the field.

How can AI assist in this process?

Although we can never expect overnight success when dealing with something as complex as vaccine development, we can act to remove some of the constraints and bottlenecks that may hamper progress.

Advances in automating data analysis and improving that visualisation of what is happening at each step of the discovery stage can address some of these inefficiencies, helping to accelerate the process of vaccine development and streamline operations to scale-up production.

Here are some of the roles AI can play at each stage:

Exploration/preclinical: The initial phases of drug discovery often involve a filtering process to narrow down vaccine candidates based on prior studies and treatments. Researchers can use AI to process vast digital libraries of data (such as analysing the properties of thousands of pharmaceutical compounds) with significantly more accuracy than manual processing, to arrive at potential treatment candidates.

AI can also be used in these stages for DNA sequencing based on complex human data, allowing clinicians to conduct genetic matching and immunity response tests.

Clinical development and trials: Once suitable compounds have been identified, the process moves towards live testing. Different patients will react differently to treatments based on factors such as age and prior medical history. The tests therefore need to be comprehensive enough to include marginal cases whereby a patient may react badly to treatment.

By training deep learning algorithms, researchers can conduct these tests at a previously unimaginable scale, even before physically administering the vaccine candidate to test patients.

These algorithms can be used to identify and sample antibodies to fight infectious diseases with drastic improvements in speed and cost.

Advanced analytics and data visualisation of the human response to potential vaccines can then be used to assist with rapid testing, allowing for more intricate analysis and lower error rates.

Manufacturing and QC: Upon regulatory approval, the race is on to develop and distribute the vaccine across a vast network of hospitals and clinics. This has significant operational implications for the manufacturers making the products, requiring rapid decision making in terms of factors such as their output capacity, the quality of the product and optimum packaging solutions.

Combining AI and sensor-based technologies, manufacturers can harness granular data to bring greater supply chain efficiencies. This helps to avoid demand–supply misalignments in their production processes and minimises the risk of products being spoiled in distribution.

Faster treatments in times of need

A viral outbreak can bring unforeseen challenges for those involved in public health management, from policy makers and health authorities through to clinicians and manufacturers. Although the former can take prompt action to test for infection and put containment measures in place, the latter are often under renewed pressure to deliver treatments at speed.

Being able to find new efficiencies in the development of vaccines can make a considerable difference to treating identified cases, relieving pressures on healthcare infrastructures and contributing to better recovery rates.

AI capabilities allow those involved in the development process to act faster under pressure. Techniques such as deep learning and advanced data visualisation allow researchers to lean on the body of existing research to tackle the complexities involved in discovering suitable treatments for novel viruses.

The utility of AI extends through to production and distribution, wherein manufacturers play a powerful role in getting these medicines into the field at rapid speed and amid conditions of great uncertainty.

Companies