Clinical trials are lengthy and expensive processes. While the exact figures vary, research suggests that the whole process takes an average of 7.5 years, and costs anywhere between $161 million and $2 billion per drug.
In recent years, data has been touted as pharma’s answer to this issue and heralded as a magic weapon that can accelerate study start up, improve quality, reduce costs, and deliver better therapy.
While it’s true that the industry produces mammoth amounts of data, the ability to harness this information to glean useful insights is a challenge.
This is not helped by the outdated methods of data collection and analysis techniques still in place at many organizations, some even dating back to 1946.
Despite these barriers to adoption, sponsors are realizing the value that new data tools and technologies – specifically around artificial intelligence (AI) – can bring.
What Is AI?
AI involves the creation of machines that work in an intelligent way, learning from experiences and adjusting their approach in a manner similar to that of humans. AI has given rise to other buzzwords, which are often used interchangeably, such as:
While AI is still in its infancy, 40% of pharmaceutical and life sciences organizations report that they have already deployed AI and that it is working as expected.
A report from Accenture described the potential of AI for the pharma industry,
“Unlike legacy technologies that are only algorithms / tools that complement a human, health AI today can truly augment human activity—taking over tasks that range from medical imaging to risk analysis to diagnosing health conditions.”
Applications of AI for Clinical Trials
AI could have a potentially transformative effect on many different aspects of the drug development process, from patient recruitment to study design and adherence, as well as electronic health record (EHR) processing and site optimization.
Patient enrollment is one of the most time-consuming, ineffective and expensive aspects of the clinical trial process. In fact, patient recruitment is estimated to take up to 30% of the clinical timeline.
While it has not been proven, many are looking at how AI can not only help sponsors with patient enrollment, but how it might help patients find clinical trials themselves rather than relying on their doctors for information, or searching clinical trial databases such as ClinicalTrials.gov.
Deep6 offers an AI-based tool that helps trials find more patients, as well as aids patients in finding appropriate trials by themselves. To oversimplify, the software works by turning clinical trial information, such as symptoms, diagnoses, and treatments from unstructured text into easily searchable data that physicians and researchers can use to more quickly and accurately identify suitable patients.
One challenge that pharma companies face when designing trials is making use of the increasing amount of data available to them from different sources.
By using AI-powered predictive algorithms that can quickly ingest and analyze large amounts of data from multiple sources, sponsors can obtain more accurate answers to questions such as whether patients will remain in the trial and if the trial is likely to be successful. This insight can help identify problems in advance and allow for much earlier protocol amendments.
CROs such as Medidata and ICON are currently working on this, along with digital health companies such as Trial AI. The company’s software uses NLP to analyze protocol text for risk factors and barriers to a successful trial. It then provides recommendations on how to improve the protocol to overcome these issues.
Trying to mine data from EHRs can be like finding a needle in a haystack. Yet, as sponsors increasingly turn their back on a one-size-fits-all approach to developing new drugs, AI can play a major role in helping them embrace precision medicine through improved data discovery and extraction capabilities.
Linguamatics’s I2E software, which is already available today, leverages NLP to help customers in pharmaceutical and biotech industries, payers/health insurers, hospitals, academic medical centers, and cancer research organizations find important information in EHRs and other unstructured text.
Optimizing Site Performance
The smooth running of clinical trials depends on how well equipped the site is, and how well the site personnel can do their jobs. AI-powered workforce and e-learning platforms can assess the behavior of site staff and provide actionable insights about what a site needs to perform at an optimal level.
ArcheMedX’s digital platform Ready delivers this kind of insight and allows CROs and sponsors “to identify risks sooner, ensure resources are more effectively deployed, and enhance staff and site performance.”
Proceed with Caution
While AI has great potential, it’s not yet full steam ahead when it comes to its deployment in clinical trials. Pharma organizations first need to ensure they have a data strategy in place that allows for better capture, integration, and quality. There is also a lack of regulatory frameworks in place to govern the use of AI technology and applications, which suggests that pharma should proceed with caution.
That’s why pharma companies need to invest in not only new data management technologies, but also the necessary skills to ensure digital transformation—the integration of digital technology into all areas of a business—before they can reap the benefits of AI.
As Stefan Harrer, a researcher at IBM Research-Australia, puts it,
“Further work is necessary before the AI demonstrated in pilot studies can be integrated in clinical trial design. Any breach of research protocol or premature setting of unreasonable expectations may lead to an undermining of trust—and ultimately the success—of AI in the clinical sector."