Big data and data analytics have been adopted across the life sciences and are driving significant industry changes. For clinical research specifically, big data presents possibilities ranging from better patient recruitment and engagement to more efficient trials and better quality results.
Given the many sources from which this data is generated, it’s inevitable that trial sponsors and managers will need to partner with third parties to access this vital resource. This post explores the opportunities big data presents, as well as the considerations and possibilities involved for clinical trial operations when third party data is used.
The Possibilities of Big Data Trials
Big data holds many promises for improved clinical trials. Indeed, it could earn the U.S. healthcare system around $100 billion in value a year through improved decision-making, write Jamie Cattell, Sastry Chilukuri, and Michael Levy at McKinsey.
Using big data strategies could also build better tools for doctors, medical insurance companies and consumers. The McKinsey trio argues that the future of big data in clinical trials will lead to better predictive modeling of biological processes and drugs through molecular and clinical data.
It will also lead to improved recruitment rates as patients will be selected from multiple data sources. Accurate patient enrollment from stronger real-time monitoring of trials will improve safety and operational procedures. Finally, they say, data will break down the silos between functions, making trials more efficient.
Of course, the difference between what’s possible in the future and what’s happening in the present can be significant. Pharma needs to do more to investigate big data capabilities, the authors argue.
Two Types of Big Data
Sources that contribute to big data in healthcare include EHRs, EMRs, insurance claims, government registries, and other databases. This data is routinely collected and often provides better quality information than controlled clinical trials, writes Tai Xie, Ph.D., president and CEO at clinical data solutions provider Brightech.
Data from treatment in real-world settings is far richer, Xie adds, and allows for the application of more statistical methods.
Xie says there are two types of big data: static, which remains the same over time and is often used as a historical reference; and dynamic, which is new data that is constantly produced, through wearable devices, for example. Dynamic data, which depending on the device may require relationships with the third party providers, can be used to enhance clinical trial operations.
In a 2021 study published in Statistics in Biopharmaceutical Research, Xie and fellow researchers explored a method for offering “anytime accessibility” of health data during a clinical trial. They noted that anytime accessibility isn’t the same thing as continuous monitoring, although both can be used together. Rather, dynamic data that is always accessible allows clinical trial teams to make better decisions in the moment for each individual patient.
The Benefits of IoMT
The IoT healthcare market is set to reach $188.2 billion by 2025, according to one study, with potentially billions of devices in use. Both of these facts are important to the development of the Internet of Medical Things (IoMT), which refers to the network of medical devices and applications collecting healthcare data, explains technology advisor Bernard Marr, author of “Big Data in Practice.”
The data-capturing, processing and analyzing of this network can help patients, caregivers, and healthcare providers better manage patient treatment and gain more accurate insights.
Benefits of the IoMT include more targeted and, therefore, more effective recruitment drives, as recruiters can draw from many online medical profiles, writes biotechnology scientist Sarah Iqbal. Retention is also improved through better patient engagement, with one example being wearables that reduce site visits.
Iqbal says the IoMT also means better data capture and improved study site execution and progress.
The Power of Data
In a study by Gareth Davies and fellow researchers, data from an original randomized controlled trial was reexamined at several later dates. The goal was to explore how probiotics used during pregnancy affect childhood asthma and eczema.
The researchers studied children at six months and again at two years old. The follow-up data gathered through electronic databases, compared with a traditional follow up, provided insights otherwise unavailable. These included reduced risk of bias as data was not reliant on patients’ recalling information, and improved understanding of asthma conditions that tend to appear in children older than two.
By relying on available data in addition to clinical observations and patient histories, the researchers were able to gain a clearer understanding of the link between probiotic use in pregnancy and infant or childhood asthma rates.
Third Party Offerings
There’s much to be gained by biotech and pharma companies that use third party data aggregators.
Consider Trials.ai, a portfolio company owned by Dreamit, writes Jack Kaufman. The heath tech startup uses AI to analyze large amounts of genomic data, as well as data from clinical studies, journal articles, and research. The information is then packaged to advise sponsors on trial design, protocol adherence, patient eligibility, visit management and retention, site performance, and adverse event reporting.
Smartphones As Data Sources
Data from smartphones and digital health technologies present scalable opportunities previously impossible or, at the least, cost prohibitive. With human touches strategically placed along the automated data processing journey, Dr. Yvonne Chan, physician scientist, digital health researcher, and founder of Agiels Science Consulting, says trials will be more cost-effective with higher patient retention.
The other benefit of this third-party source of data is the trial staff can engage in real time with patients, sending push notifications and reminders, as well as collecting data and exchanging feedback.
Perhaps even more importantly, digital health data gives researchers a better understanding of patient populations through more comprehensive and often real-world data about the complex conditions and the effects of treatments.
Chan says the ultimate goal, at this point, is for big data to make accurate predictions about the patterns of disease and wellness. By doing so, healthcare providers can work towards a personalized approach to patient treatment.
Data from Wearables
There’s a concern that data from wearables may not be held to the same level of auditing that data in clinical trials is. But this should not be regarded as an impediment to its use, says pharma industry consultant Julianne Hull. Human error affecting the integrity of data from wearables, which will no doubt improve over time, can be accepted in exchange for real-time data straight from patients to researchers, she explains.
Data sourced this way will be cheaper and more efficiently gathered. While data from wearables is currently better suited to NDAs in observational post-marketing trials, Hull is confident this will change in the future to include phase III trials too.
Connection Between Trials and Pharmacy Customers’ Data
Some pharmacies, in exchange for payment from trial sponsors, will send out mailers to their customers whose conditions match research areas of interest. So when someone fills their prescription for a condition on a trial sponsor’s research agenda, they may soon start receiving information about intended trials.
Expanding use of customer data to sell healthcare — from clinical trials to over the counter products — may offer some benefits. Yet it also poses risks. “There is an urgent need for states to recognize the value of health data and use it to advance human rights,” writes Amy Dickens, a policy advisor at London’s Centre for Data Ethics and Innovation. Striking a balance between the uses of data and the privacy rights of patients is a must.
Improved Understanding of Eligibility
Clinical researchers studying big data often find evidence that complements clinical trial findings. The results give physicians “comparative effectiveness data” on which they can make treatment decisions and recommendations. Indeed, these were the findings by cardiologist Peter Noseworthy and Xiaoxi Yao, Ph.D.
The pair analyzed data from 183,760 patients with atrial fibrillation and looked at those treated with ablation between August 2009 and April 2016, the same period during which the CABANA trial enrolled patients. With a patient cohort 84 times larger than in the CABANA trial, the researchers found that 73.8 percent of patients in routine clinical practice met the eligibility criteria for the trial and that ablation, for these patients, led to 30 percent lower risk of stroke, death, major bleeding, or cardiac arrest.
The results were similar to CABANA’s secondary per-protocol analysis — patients that completed assigned treatment, but came from observational methods alone.
Complexity of Big Data Aggregation Through EHRs
Data aggregation and normalization from many sources is complex and requires data operatives with the right skills, experience, and technology to understand it. Some of the immediate challenges, explains Dr. Paul D. Taylor, CMO at Philips, include EHR vendors using too many terminologies — otherwise known as code sets or dictionaries — that are often unique to vendors.
EHR data must be translated and mapped to be useful, Taylor notes. Additionally, vendors need to apply standardize data terminologies and use of EHRs.
Data Science Revolution
The clinical research industry is in the midst of a data science revolution that requires pharma companies to shift focus from pharma companies. Computational neuroscientist Luca Finelli, vice president and head of insights, strategy and design, data science and AI at Novartis, calls his organization a data company.
Big data analytics bring with it productivity gains and transparency, he says, the latter set to be a huge driver of change in the future, especially as pharma looks to become a learning rather than knowledge culture. But transparency would be really useful if machine learning and cognitive computing technologies are used to extract information to make better decisions.
The result would be clinical research teams using this technology and big data analytics to select hospitals that would optimize operations of the designed trial.
The technology behind big data continues to improve over time, enhancing the integrity and quality of the data available to clinical researchers. Understanding the value of third party data will be key for researchers looking to make their clinical trial operations more efficient, cost-effective, and meaningful to patients.