[Image courtesy of Adobe Stock]
This translational research challenge continues to be so significant that Harvard University recently launched a pilot program to explore overcoming the “multiple roadblocks” that “limit the breadth and impact of research on human health.”
Roadblocks that Harvard has identified include patient recruitment and data management. Unfortunately, data linking patients who would benefit from participating in clinical trials to emerging therapeutics in the development pipeline remains siloed to the detriment of all stakeholders. In oncology, in particular, this lack of awareness about available clinical trials offering novel therapies and the inability to identify eligible patients can unintentionally prevent access to potentially life-extending or quality of life enhancing treatment options.
Artificial Intelligence, however, may help researchers and providers overcome these “bench-to-bedside” obstacles. With AI-enhanced data entry, analysis and integration, clinical trial tools are already helping healthcare systems and pharmaceutical companies collaborate to identify qualified clinical trial participants more quickly and to fulfill the promise of research as care to their patients.
Delays impact costs and patients
Between 2009 and 2018, the median R&D cost for each novel pharmaceutical drug brought to market was $1.1 billion, which includes losses spent on failed formulations. Furthermore, only 18% of drugs advance from preclinical research to clinical trials, taking an average of nearly 14.5 years to make that transition.
There are numerous reasons for these costs and time delays. Notably, identifying potential clinical trial sites and recruiting a sufficient cohort of patients to participate is time-consuming. Contracting between pharmaceutical companies and academic medical centers or contract research organizations that recruit participants and manage trials is also slow and bureaucratic.
Yet, many pharmaceutical sponsors find that limiting their clinical trial collaboration to only well-known large academic medical centers and CROs is part of the problem. While experienced and skilled in research, these institutions may only have access to data on a limited pool of potential participants within their catchment area. However, pharmaceutical companies who may be eager to broaden their network of research partners could discover that midsize health systems may lack the staff and/or technology resources to support and scale a trial.
Expanding trial and drug access
Technological limitations, however, are by no means limited to midsize and smaller healthcare organizations. Patient data in electronic health records (EHRs) must be transformed into a compatible format for clinical trials. The process for identifying patients that match trial protocols is still done manually, even by the most prominent institutions, and requires extensive human clinical validation to ensure the patient record is abstracted and matched to the protocols that are interpreted correctly.
Recognizing these obstacles, pharmaceutical companies have launched programs to improve data management efficiency and expand access to novel therapies to more patients. For example, a large pharmaceutical company recently provided trial sites with user-license credits to access an AI-powered platform that rapidly and accurately pre-screens patients for the sponsor’s high-priority, interventional oncology drug trial.
Identifying, recruiting, and enrolling oncology patients in such trials is typically challenging due to the complexity and time-sensitivity of identifying the right patient at the right moment for the most appropriate trial. Oncology clinical trial inclusion and exclusion criteria are very complex – often with over 50 specific requirements for participation. Beyond the clinical complexity and specificity, the time and travel constraints faced by patients and their families can be a barrier to participation. Expanding site locations efficiently is essential to reach more patients and advance research. In this case, the pharmaceutical sponsor and the health-tech company identified ideal pilot sites with potential trial participants who would likely benefit from the novel drug.
Removing data obstacles
The AI in the clinical trial platform sponsored by the pharmaceutical company identifies and screens patients using a validated combination of clinical knowledge, large-language models and natural language processing to analyze both structured and unstructured EHR data, including scanned documents. A structured patient summary is used to screen against the trial’s complex inclusion/exclusion criteria, significantly reducing the manual time and effort required for patient screening and enabling existing clinical research staff to screen >20X more patients with no additional time effort.
Similarly, AI tools used in clinical trials can be used for other purposes including, feasibility analysis, cohort discovery for research and automatically perform the lengthy electronic data capture (EDC) process from the EHR, saving time while ensuring that required data elements are captured and entered in a standardized clinical language. Throughout, a clinician experienced and skilled human-in-the-loop process supervises the AI’s progress to ensure accuracy. This application of AI is already used in numerous health systems to support data abstracting from the EHR system to clinical data registries, typically eliminating a months-long backlog of data entry in less than a week.
Making trials ubiquitous
Streamlining clinical trial recruitment and clinical data management could shorten drug development timelines and make bench-to-bedside collaboration more feasible for sponsors and researchers. However, more importantly, helping identify eligible trial participants across a wider geographic area can enable more patients to receive potentially life-extending novel therapies sooner without the expense and disruption of travel.
Expanding care access this way may not be the primary goal of clinical translational research. Nonetheless, clinical research as a care option is a worthy objective that can further pursue discoveries that will help even more people achieve better outcomes
Aaron Brauser
Aaron Brauser is president and co-founder of Realyze Intelligence, a Carta Healthcare company that delivers AI tools to accelerate clinical trial matching and abstraction of Real-World Data for NCI designated oncology centers, Integrated health networks and life science companies. Prior to founding Realyze Intelligence, Aaron was the VP of Solution Management at M*Modal responsible for the commercialization and delivery of M*Modal’s solutions which was acquired by 3M in 2018. Aaron has spent the last 25 years developing and delivering software for both startups and fortune 500 businesses such as 3M, M*Modal, Nokia, CoManage and US Steel. Aaron earned both his MBA and BS in Computer Science from the University of Pittsburgh.
Filed Under: clinical trials, Drug Discovery, machine learning and AI
