According to the US Food & Drug Administration (FDA), digital technology is driving a revolution in healthcare.
The lines between healthcare, drug discovery, clinical research and commercialisation are blurring, as the patient becomes a key partner and focus. We are seeing a rapid expansion in the use of mobile and patient-centric devices, exponential growth in the volume and diversity of life sciences data and acceleration in the use of data-dependent computation to gain insight and automate – loosely called artificial intelligence (AI). These digital health trends are naturally combining to transform the patient experience and the application of new scientific ideas and breakthroughs.
These advances are permeating all aspects of clinical research but the recent acceleration of decentralised and hybrid clinical trials (DCTs) illustrates how far reaching digitalisation is becoming. Patient-centricity has been driving the decentralisation of clinical trials for some time but the rapid emergence of the COVID-19 pandemic required the pharmaceutical industry to pivot operations and accelerate its DCT programmes. This response required a corresponding ramp-up in mobile technology, data management and AI.
AI can provide insight into protocol complexity and contribute to protocol designs better adapted to DCTs, including the creation of virtual control arms. The intelligent use of data to include historical data as well as the data collected during a clinical trial can optimise the number and diversity of patients needed to reach the desired endpoints and give the patients who do participate a higher value experience.
Using health insurance claims data, social intelligence data and other sources of real- world evidence, AI can foster a ‘right patients, right time’ philosophy for each trial. Where the human mind finds it impossible to make connections, AI algorithms can combine the meaning of a health insurance code with a social media event, alongside lots of other data points, to establish a robust country/site mix that drives the overall strategy for the trial. Identifying the right patients ensures better diversity and inclusion and a better patient experience, as well as scientific success. Diversity objectives range from reaching all patients who can benefit from participation in a clinical trial, to ensuring that the trial collects the maximum range of data within the scientific inclusion/ exclusion criteria. Identifying the right patients also means you identify the right investigators and expedite study start-up activities. The new era of DCT, combined with the advance of AI, expands the horizon for patient identification beyond traditional site-based models.
DCT opens up many more ways of collecting data directly from patients and this march of digitisation, aided by the increasing sophistication and connectedness of mobile devices, creates a positive virtual cycle with AI. More data means more opportunity to develop smart risk monitoring processes that use AI to detect emerging patient risk early and recommend mitigation. Even more exciting is the prospect of combining data collected by remote devices with electronic clinical outcomes assessment data collected through questionnaires to develop new digital endpoints or predict placebo effects. What was once a topic for blue sky innovation is becoming a reality as more and more data is collected and increasing numbers of data science professionals apply AI to those data sets. It will require effort and courage to maintain momentum in the development and validation of these challenging initiatives and not revert to a pre-COVID-19 mindset.
Even the relatively repetitive tasks of clinical data management, where clinical trial databases must be carefully curated, cleaned and standardised for statistical analysis, are coming under the eye of data scientists. The diversification of data sources drives the need for smarter, more efficient ways of conducting data management and this need is matched by the increased focus on AI, as a means of optimising all aspects of the clinical research value chain.
An important side narrative to this DCT revolution is the nature of AI itself. There are plenty of definitions of AI and many are quite specific. However, it is the more vague, general sense of a machine being able to do something previously thought to be an exclusively human activity that is gaining practical acceptance by business leaders. It doesn’t matter if the solution is an advanced deep learning model or a traditional piece of procedural programming if the result produces better insights or more optimised processes.
The digitisation of all aspects of life is driving business leaders to restructure processes around new digital technologies and data scientists are listening to those business leaders to develop AI solutions. This is fundamentally an activity driven by data availability and data literacy. A key sustainer of the relationship between data scientist and business leader is model transparency and direct end user involvement in any AI solution. Transparency and the dynamic injection of human decision-making into AI models not only supports real-world validation and accountability, it generates a key synergy between AI and HI (human insight).
Clinical research, digitalisation and AI are at an important juncture as we emerge from the COVID-19 pandemic. There is much to glean from the experiences of the past two years and even more to gain in the future as the industry continues to transform the clinical trial into a more patient-centric, decentralised, dynamic, data-centric, AI-powered activity. Clinical research can increasingly become a healthcare option and the best practices from DCT and the accompanying technology solutions will find their way into mainstream healthcare practices, decreasing the need for expensive and inconvenient in-patient care. The benefits will be felt by scientists, clinicians, commercial interests and patients alike.
Michael Phillips is Director of Innovation and Informatics at ICON