To understand how modern data analysis supports faster clinical research, it is important to consider how far we have come. In previous decades, paper-based data was the standard and the time taken to produce these records meant that there were inevitable delays in the analysis of data and subsequent formal assessment on the Investigational Medicinal Product (IMP).

Capturing data electronically and in real-time results in fewer errors, increased speed and quality of data capture and enables data to be available immediately. With the advent of increasingly sophisticated ways to present and segment this data, scientists can utilise platforms that present information visually, enabling a real-time assessment of the safety of patients and the outcomes of trials. This aids medical decision-making and ensures more efficient studies, with researchers able to make faster assessments of the efficacy of therapies and maintain a holistic view of the studies.

For example, my team at Richmond Pharmacology recently re-analysed large historic data on blood sample parameters and re-adjusted normal ranges as per the manufacturer recommendations. This enabled the widening of the inclusion criteria of eligible subjects, and subsequently resulted in the faster completion of the clinical trial. Therefore, upgraded data analytics can help make innovative medications available to the public sooner, by safely and comprehensively addressing hurdles to recruitment, monitoring or follow-up faster than ever before.

Artificial Intelligence facilitates safer, more efficient and more accurate clinical trials and the collection of real-world data (RWD) has significantly increased in recent years. However, the true game-changer is how to utilise the vast amounts of RWD effectively. AI, defined as the ability for models to ‘learn’ from data and continuously improve, offers cost-effective ways to usefully monitor vast amounts of RWD, as well as offer new insights not previously obvious to the human eye. This can include predictive modelling for drug events, occult medical conditions (eg, non-alcoholic fatty liver disease), to diagnostic algorithm detection for safety events, including advanced data visualisation over a whole study.

These advances go hand-in-hand with adaptive trial protocols that are increasingly the favoured method of trial design by CROs and the pharmaceutical industry. The key advantage of adaptive trials is they provide the ability to operate flexibly, enabling the modification of a trial’s course in accordance with pre-specified rules set out in the original protocol, with the power of AI ensuring these adaptive decisions are made accurately and early, avoiding harm.

Compared to traditional rigid protocol trials, this agile approach saves time and resources. It enables important assessments to be made earlier, meaning the cost of bringing a drug to market is lower and eventually making the medication more affordable.

Adaptive trial protocols are facilitated by researchers having data available in real time, empowering them to take the decisions and trial deviations that are supported by a more flexible protocol. Therefore, effective data science systems that may incorporate data visualisation, big data analysis and AI predictions are viewed as a key criterion for sponsors when choosing a CRO to work with. In my current work, our academic research organisation, Richmond Research Institute, is aiming to develop AI systems to predict occult fatty liver disease, an important confounder and contributor to adverse liver reactions in new medicines, and predicting other outcomes, from commonly collected RWD such as ECGs and changes in patient body weight.

We have developed a number of data monitoring and visualisation tools to increase the efficiency of running clinical trials. These advances support improved drug development and treatments being accepted into the market faster, ultimately benefitting patients.

It is welcome that regulators across the world have been able to set updated and clear criteria for the type of evidence they will accept during early-phase clinical research. This assists CROs to build evidence bases that are based on AI interpretations, safe in the knowledge that this is understood and accepted.

In the US, the Food and Drug Administration (FDA) is increasingly accepting drug submissions based on RWD – patients’ health status that is routinely collected from a variety of sources – and real-world evidence (RWE) – the insights gained through analysis of RWD. More recently this evidence is being used to assess new indications for an IMP, without having to conduct clinical trials. For example, the FDA approved an immunosuppressant usually used for organ transplant, tacrolimus, for use in combination with other immunosuppressant drugs to prevent organ rejection, following successful observational studies based on RWE. The regulator highlighted how this approval reflected how a study relying on RWD when compared with a suitable control can be considered sufficient.

Something similar is also occurring in the UK. The Medicines and Healthcare products Regulatory Agency (MHRA) has recently provided guidance on the use of RWD in clinical studies to support regulatory decisions. The overall objective of this is to speed development programmes at reduced cost. The MHRA also suggest that RWD is more representative of the true effects of a treatment in a community, and more generally applicable than data collected from the standardised setting of a traditional clinical trial.

More broadly it is important that the UK government has committed to a ten-year National AI Strategy that plans for and invests in the long-term needs of the AI ecosystem for the benefit of all sectors and regions of the country. Interconnected is the NHS AI Lab, created by the Department of Health & Social Care to bring together different stakeholders for programmes which address significant barriers to development and deployment of AI systems in health and care – for example accelerating disease detection and maintaining a safe and ethical regulatory system for the development and deployment of AI technologies.