September 14, 2020
How life sciences companies are reimagining trials during the biggest health crisis in a century
Whatever the world was like on March 15, it’s not like that now — and it probably won’t be for months or years. Everything from buying groceries to renewing a driver’s license is completely different from the way we did things just a few months ago. Clinical trials, like most medical activities, have been significantly affected as governments and pharmaceutical companies have pivoted to a single focus: combatting the novel coronavirus. At the same time, the rush to develop cures and vaccines for COVID-19 is condensing the review and approval process from years to months. This is where artificial intelligence (AI) can play a vital role in changing how clinical trials are conducted and how therapies are evaluated and tested.
AI and machine learning are no longer an emerging trend in the healthcare and life sciences industry: in 2019 there was an 88 percent increase in the number of healthcare professionals who said their organizations were implementing an AI strategy. Nevertheless, the reality is that life sciences companies have traditionally been slow to adopt new technologies and currently many are in the process of working through digital transformation strategies to enable AI capabilities. The COVID-19 pandemic is acting as an external disruptive force to the industry that is accelerating innovation around the traditional clinical development process. With the urgent global need for virus prevention and treatment, the industry can no longer wait to innovate, modernize or adopt advanced technologies that positively impact the timelines and costs associated with bringing new treatments to market.
Life sciences organizations need to leverage AI and advanced data science techniques to run clinical trials more efficiently with less cost and time in order to bring important new therapies, including a COVID vaccine, to market faster. With the AI healthcare market projected to reach $6.6 billion by 2021, it is important that organizations know what to expect from an AI-powered future, where to invest to prepare for that future, and how they can leverage these technologies today. It’s easy to throw around terms like AI and machine learning, but it’s a lot harder to operationalize them for use in the real world. Let’s look at a few practical applications of advanced techniques to expedite research.
Clinical trial protocol complexity is driving higher volume and more diverse data collection from numerous systems and sources, including longitudinal data from devices. Efficient data collection and analysis strategies are key for quicker approvals and speed to market. To get the most value and insights out of clinical data, data must be collected across diverse systems and mapped or standardized. This process can be time consuming for organizations, and when it’s done manually, can run the risk of human error. It relies on identifying and mapping tables, variables, derivations and code-lists to apply proper transformations and to ensure full transparency of that data for data lineage.
According to the 2019 Tufts-eClinical Solutions Data Strategies & Transformation Study, having a formalized data strategy in place helps alleviate some of the difficulties associated with data mapping. By implementing AI and machine learning, data mapping or standardization can be significantly less onerous, and completed quickly with an auto-mapping system to scan through source data, identify and apply the required transformations. It could also enable guided-mapping, which can accelerate complex transformations. These applications of AI can yield faster analysis and decision-making for clinical trials.
Inefficient patient selection and recruiting techniques are a primary cause of trial delay and failures. In fact, 85 percent of clinical trials fail to retain enough patients and the average dropout rate for clinical trials is 30 percent. 80 percent of patients typically come from 20 percent of clinical sites, resulting in significant opportunity cost associated with the tremendous resources needed to activate, train and monitor sites that are unable to enroll patients.
Many patients are unaware of clinical research opportunities that exist for their condition. This also means that finding the appropriate trials nearby can be complicated and time consuming. As the complexity of studies increases, so in turn does the inclusion and exclusion criteria, which compounds the patient recruitment challenges. AI and machine learning strategies can be employed to help with enrollment and retention. Algorithms use de-identified healthcare, observational, safety and historical data to identify the right patient population that meets inclusion and exclusion criteria, as well as other factors that affect the structure of the study. Advanced systems can also be developed to notify patients about trials that they are eligible to participate in, taking the burden of participation off of the doctors and patients.
Setting Up a Data Strategy to Manage AI
In the 2019 Tufts-eClinical Solutions Data Strategies & Transformation Study, it was revealed that sponsors who had a formalized data strategy were significantly more mature in both analytics competencies and applying AI and machine learning capabilities to clinical development processes. Without both a comprehensive data strategy for R&D and a defined and automated data pipeline and technology infrastructure to centralize data across the clinical development enterprise, it is difficult to build a sustainable and scalable data environment that is needed for advanced learning technologies to succeed. AI and ML models and algorithms require representative data sets to work well and automated data flows to learn and increase precision over time. Having a successful and well-managed digital data strategy will set the foundation for the organization’s AI strategy.
The same study found that respondents who had both a data strategy and a centralized data hub rated their analytics competencies as more mature across the board, and were more prepared to take advantage of advanced AI capabilities including predictive and prescriptive analytics across both clinical and operational data. These results are further evidence that both a data strategy and a clinical data platform used for centralizing, transforming and publishing data are essential, foundational building blocks to creating AI capabilities across clinical development.
The scope of the foundation for AI and machine learning goes beyond the technological requirements. Employing the right professionals to help facilitate this change is also necessary. There is a broad consensus across the life sciences industry that employees will require knowledge upskilling to work with AI. Existing employees will need to be trained to work with AI technologies in order to keep up with the pace of transformation. The skill shifts are significant, but they also present tremendous opportunity.
Prepare for AI
The COVID-19 pandemic is an example of a very significant external market force with many contributing factors. These factors are usually discussed during strategic planning sessions, and external forces are the ones that wield the greatest influences over the opportunities and threats of an industry, market or business. The pandemic is threatening the traditional research model by pausing 60 percent of trials in the U.S., dramatically reducing the number of sites that can perform research due to surge capacity and limiting access to patients for any non-COVID related study. It also presents several important opportunities. Technology can be used to reach patients directly, and it can better leverage the ever-increasing, new digital data streams that exist across both healthcare and research. New technologies can also take the work, time and cost out of the clinical development process to get COVID prevention and treatment to market — because the urgency has never been greater.
AI can be applied to many of the opportunities that are present in the post-COVID pandemic environment. This includes using AI models to mine healthcare data to identify the many different patient segments that must be studied for the prevention and treatment of the virus. These models can also speed the protocol finalization and review process across stakeholders. AI can help automate data review processes and highlight risk for patients and sites to focus resources where they will have the highest impact. AI can also standardize data so that it is available for analysis faster, and predict the most promising treatments and targets to reduce opportunity costs. Across the board, AI can be used to take work genuinely out of the system and deliver an improved experience to researchers, clinicians, patients and health authorities united in the goal of reinventing how therapies are developed. The advantages of this transformation are numerous, including quicker and more accurate clinical decision making. And right now, this is the most important thing for the industry.