Prevailing thought for the past few years has been that with enough data, we’d know most of the answers we have been looking for or at least find it useful for asking the right questions.
In clinical research, access to data — and the software to make sense of it — means researchers are better equipped to define their research questions and align various study requirements to optimize results. By providing improved site selection and recruitment, along with greater insight into patient populations and their shared traits, AI and machine learning are turning data into a valuable tool to define clinical study questions.
Statistics show this big data and machine learning are making enormous inroads into the pharmaceutical and medical research markets. In fact, these tech resources are estimated to deliver around $100 billion each year through improved decision-making, innovation, efficiency of trials, and new tools for physicians, consumers, insurers, and regulators, Jamie Cattell, Sastry Chilukuri, and Michael Levy at management consulting firm McKinsey & Company write.
That’s not surprising considering that by 2021, AI in healthcare should have achieved a compound annual growth adoption rate 42 percent, according to the team at procurement and supply chain consultancy, GEP.
Let’s take a look at how AI-powered algorithms are gathering and making sense of valuable data and how this is shaping researchers’ areas of study.
The first step in the digital revolution in healthcare was creating Electronic Health Records (EHR), but having all that data only gets researchers so far. There needs to be the capabilities to analyze it. This is what software company CBInsights posits as an ideal and achievable AI solution. Sophisticated software scans and gathers relevant information from a patient’s EHR, and matches it against ongoing and upcoming trials.
One example of a company using AI to improve trial research is Antidote, which functions in the clinical trial space the same way traveling aggregating sites like Kayak does. Antidote uses machine learning to translate inclusion/exclusion criteria trial jargon on ClinicalTrials.Gov into language a search engine can understand.
The result for patients is that, after entering their condition along with their age, sex and geographical location, Antidote returns a list of trials that meet those basic criteria, along with further prompts to determine eligibility for existing trials.
There aren’t enough doctors worldwide, and the ones we have are aging. Add the higher demand for chronic care, and Bertalan Meskó, M.D., Ph.D. says the healthcare workforce is facing a crisis. The Medical Futurist director and lead author of the essay at BMC Health Services Research, Meskó explains that artificial intelligence will ease this crisis on multiple levels, from big data analytics to AI-powered cognitive assistants that will help clinical researchers do their jobs better.
Consider IBM’s Medical Sieve project. Its aim is to build an assistant with reasoning capabilities and clinical knowledge that can analyze radiology images to detect medical issues. Then there’s Deep Genomics, an AI-powered discovery platform that trawls large data sets of medical records to determine links between diseases and genetic information.
Atomwise uses supercomputers in pharmaceutical research to find new therapies. The company found two drugs in less than a day that may lower Ebola infectivity rates. A traditional approach to research would have taken years to find similar results.
Meskó says AI-based services could assist with diagnostics, decision-making and administration, but cautions that limitations exist. “Every single element and parameter cannot be translated into a programming language. Moreover, there is no clinical trial or peer-reviewed data about the data points that contribute to a medical decision,” he writes.
When it comes to making informed clinical research decisions, data scientist Kevin Rooney, Ph.D., at Wavelength calls AI the “next revolution of clinical trial development,” comparing it to the “automobile replacing the horse of decision making.”
However, he prefers talking about machine learning (ML) as the focus should be less on AI’s impact but rather on the “waterfall of change management that needs to happen in response to AI-based decisions.” ML and AI are closely related, Rooney explains: ML is what people are already achieving in AI; AI more broadly speaks of aspirational developments.
Currently, Rooney’s team at PRA Health Sciences uses ML to collate medical data in the US from sources as diverse as EHRs, insurance claims and prescriptions to compare patient populations with patients enrolled in current clinical trials. Not only does this help early identifications of at-risk patients, but the technology could also help define future research questions based on accumulation and analysis of the data.
It will also mean making decisions with greater confidence as ML can highlight information you might otherwise not have noticed as influential variables, Rooney adds. Rather than rely on one or two distinct traditional variables, ML can help point to a pattern of variables.
With artificial intelligence, machine learning, big data and even social media being used in clinical research, Dekel Taliaz, Ph.D., cofounder of AI health-analytics company Taliaz, calls this a “new era clinical research.” Specifically, AI will enable researchers to tap into “previously unavailable genomic, clinical and environmental Big Data” allowing AI-personalized, real-world studies.
The big shift will be how research questions are formulated. Traditionally, Taliaz says, this was done by asking a specific question and testing it against a hypothesis in a controlled environment. AI will grant researchers the capability of asking (and answering) additional questions by analyzing real-world data from many thousands of patients recruitable through tech such as Apple ResearchKit.
This might include data from EHRs, registries, hospital records and health insurance data alongside biobank, genomic and digital phenotyping information and wearables. This is a boon for recruitment as millions of patients’ data could be available without patients having to submit that information themselves as trial applicants.
With this approach, research questions can be changed and improved upon as more data is collected and analyzed with a flexibility impossible with a traditional study. This is what happened with the trial by Pfizer and 23andMe, providing access to huge data sets through the latter’s genetic database. Results included linking “15 genome sites to depression using the traditional genome-wide association approach,” Taliaz writes.
While social networks and internet search engines are proving useful places to tailor clinical trial recruitment advertisements, what’s more exciting is a machine-learning algorithm being tested at Cincinnati Children’s Hospital Medical Center.
The aim is to analyze why patients choose whether or not to enroll in relevant studies, Dan Stempel at MD Connect explains. The algorithm analyzes data to “predict specific patient decision making” by working out patients’ personal biases to tailor an even more personalized recruitment strategy. Its success rate so far shows a 12 percent increase to 72 percent compared with traditional recruitment methods.
Deep 6 AI founder and CEO Wout Brusselaers is also using AI to improve recruitment. Deep 6 AI’s software finds patients for trials within minutes and once validated 58 eligible patients in ten minutes — more than twice the number achieved by traditional methods over the space of six months.
The software uses neuro-linguistic programming (NLP) to read doctors’ notes, medical reports and diagnoses. It also detects “hard-to-find lifestyle data” to match patients with trial criteria and maps shared or common traits in patients’ symptoms, disease progression and outcome.
So researchers could select a number of characteristics shared by Alzheimer’s patients and use the software to compare them with undiagnosed patients possessing the same traits. They might also determine whether these traits were predictive of the disease.
Being able to target patient populations with greater accuracy and recruit them with better effect, could well give researchers confidence when defining their areas of research. This is something Daniel Faggella, founder and CEO at market research platform Emerj, highlights. He argues that ML can help “shape and direct clinical trial research” by using predictive analytics to identify the best candidates for clinical trials.
Taking the uses of AI-powered algorithms further, AI expert Anirudh Kala says a site-scoring algorithm will help researchers select the best trial site based on their budget and study requirements.
As chief data scientist healthcare IT platform Vitrana, Kala explains that the scoring algorithm is able to align sites with study protocol, help select investigators, improve patient recruitment and retention through predictive analysis, and measure probabilities of the trial’s success. The algorithm functions by accessing online public datasets and using advanced neural networks and NLP techniques to make sense of the data’s “semantic relationships.”
An example of a semantic relationship is asking the system for all the sites at which gastric bypass surgery was performed. The relevant sites will be gathered and ranked according to parameters such as enrollment rate, site experience and activation time. The value in this approach is making more accurate predictions about the site’s match for the research question.
Images by: sdecoret/©123RF Stock Photo, callipso/©123RF Stock Photo, jirsak/©123RF Stock Photo