The Impact of AI, Precision Medicine, Mobile Health Technology on Streamlined Clinical Trials

January 17, 2019
Mark Lambrecht

Applied Clinical Trials

Today’s therapies are developed to be more personalized, precise, and real-time data-driven than ever, yet the clinical trials in which they are tested deliver results based on the average response from a given trial’s many participants. Further, new therapies are tested in clinical practice by physicians caring for patients that might never have been included in such trials. These patients are different from their experimental cohorts in myriad ways: they have different genetic backgrounds, may suffer from additional diseases and likely have a history of taking different medications.

This implementation gap between clinical research and practice explains why promising new medicines often do not find uptake in clinical practice. To bridge this gap, clinical trials will make increasing use of mobile health technology and data that is specific to the individual patient. Indeed, having been digitized over the past decade, clinical trials are now bound to become more pragmatic, incorporating patient-specific data from IoT-enabled medical devices, digital health apps, and even digital biomarkers.

Patient-specific data

The rising use of patient-specific data will lead to streamlined clinical trials. It will also narrow the implementation gap of new therapies by helping physicians, health tech companies, reimbursement agencies, and the public at large to understand the clinical, economic, societal, and scientific value of such therapies. The challenge of these new type of trials, however, goes beyond the expensive technology. It also lies in the complex processes needed to capture high-quality data and the required statistical design that amplifies process complexity. 

Fortunately, mobile health technology is evolving quickly, becoming less expensive even as we engineer better, smarter, smaller sensors for measuring clinical parameters. Couple such advances with enormous leaps forward in statistics, AI, machine learning and real-time analysis, it’s easy to see how a growing number of trials in diverse therapeutic areas will benefit from this streamlined research model.

From paper to digital

Clinical trials are by necessity run in a very sequential, risk-averse, rigorous, and statistically controlled manner. Producing evidence in a statistically robust and scientifically validated manner ensures that the medicines and therapies brought to market are both safe and efficacious. 

By transitioning from paper-based clinical trials to digital systems in recent years, pharmaceutical sponsors have been able to better plan and conduct trials, recruit patients and analyze results more rapidly, and improve standardization. Pharmaceutical companies and CROs have deployed this digital layer in support of the standard clinical trial, but many have not changed the traditional, phased process. 

Now digitally-enhanced clinical trials are being overhauled in favor of more profound streamlining that incorporates into the development phase real-world patient information from a clinical setting (as opposed to the experimental trial setting). This next wave of clinical trial design will combine the best of both worlds: the gold standard randomized controlled trials (RCT) and the ability to adapt the design based on real world data. It will require under-used trial design methodologies such as N-of-1 trials and adaptive design techniques to ensure analysis results are robust and statistically reliable.

Real world data used in the context of clinical trials

Real world data (RWD) is all data that is not captured as part of the RCT but in the practice of clinical care. Beyond health records, RWD can include both information derived from insurance claims or patient registry retrospectively and/or capture new data prospectively during clinical development or after the therapy has been approved.

Real world evidence (RWE) is clinical evidence derived from analyzing RWD. In complement to RCTs, such analyses can reveal potential short- or long-term risks and benefits of drugs or therapies. For example, the STRIDE-PD study found an unexpected incidence of prostate cancer in the group taking Parkinson’s medication entacapone, compared to the control group. At the FDA’s request, the drug’s manufacturer conducted an observational cohort study using Finnish patient registries and ultimately concluded that patients taking entacapone were at no additional risk of prostate cancer. 

Global regulatory agencies such as the FDA and the EMA are setting the tone. Both recently issued policy texts indicating a readiness to accept RWD to better assess the safety, effectiveness and, in some cases, efficacy of new therapeutics. The FDA-issued a framework for a real world evidence (RWE) program allows the use of RWE for decisions of effectiveness but only in limited conditions. The EMA has likewise indicated that RWE can be used for complementing RCT decisions in carefully defined settings.

Project Data Sphere is an example of the potential of crowdsourced historical clinical trial data. The cancer research initiative allows researchers to submit, collect, and analyze such data for the benefit of research in its online analytics environment. Used in this way, historical trial data can guide better trial development and act as comparator arms.

RWD is not a magic formula that solves all the information gaps of clinical trials, of course. But when collected to gather more meaningful and observational data about the trial recruits, it can help advance understanding of a therapy. Digital technology such as wearables or adherence-measuring techniques might tell us whether the patient is dosed correctly, for example, or perhaps provide more confidence about optimal dose or shed light on possible adverse events and interactions with concomitant medications.

Pragmatic trials 

Clinical trials are optimized to capture very specific information about treated patients with the goal of demonstrating the efficacy of a therapy under optimal study conditions. Pragmatic trials, or effectiveness trials, expand to include information from interventions and interactions with many healthcare professionals even outside the context of the experiment trial setting. They also involve patients that would otherwise be excluded from a RCT, but they still include prospectively enrolled subjects that are randomized for the treatment under investigation. 

Pragmatic trials require an information system that links up electronic health records (EHRs) across primary care physicians, pharmacies, and hospitals. The world’s first large scale pragmatic trial and well published example is the Salford lung study

Pragmatic trials are gaining in popularity for a couple of reasons:

· Firstly, the generated data helps ensure that the trials are relevant and include the right patients and that the evidence is gathered not just about clinical aspects but also patient outcomes, economic relevance and impact on quality of life parameters. 

· Secondly, the RWE captured during a pragmatic trial contains a treasure grove of insights about the subjects, which can aid in closing the implementation gap. 

In contrast, RWE such as that derived from claims and EHR data is known to be useful for epidemiology and in addressing health economics questions but often falls short of advancing clinical information about individual patients because of lack of EHR information for those same patients. For some disease areas, claims databases might be more useful than other sources of patient data, because they contain more information captured by primary care physicians sought after pragmatic trials, too.

Pragmatic trials are more expensive as they involve digital health apps, integrated health systems (such as linking pharmacy prescriptions with clinical records) and better data and analytics platforms. 

Mobile apps and mobile health technology 

Digitally-enhanced clinical trials include additional sources of high quality or regulatory-grade information such as data derived from patient wearables or patient-reported outcomes. Using these digital health initiatives will greatly enrich the data available to help determine the optimal dosing scheme of a new drug, for example. 

Patient wearables such as those used in diabetes or cardiovascular trials are creating an abundance of data points, but finding meaning in this data is not easy. For wearables data to be useful in a RCT, it needs to be meaningful both clinically and scientifically. And, importantly, the apps or wearables must be approved by the regulators. The FDA has issued guidance around “software as a medical device,” and the European new Medical Device Regulation offers a robust framework to allow algorithms and apps to be approved as clinical-grade devices. 

The ability to capture data from patients in real time or near real time will not merely digitally enhance RCTs. It will change them profoundly.

Scientific advances require new trial designs

At the same time as RCT design and the use of RWE are being investigated, trials are becoming more personalized using gene therapy, gene transfer, or immunotherapy. The enormous biological complexity and tremendous amounts of data generated by such trials require new trial designs and the use of mobile health technology. Further, the related analytical systems must be able to rapidly test hypotheses, find biomarkers and deal with genomics, and monitor trial results as they become available.

The quantified self-impact of AI, engineering and technology on trials

The ability to capture clinical and adherence information in near real-time from patients using mobile health technology and apps is exploding. Deep learning AI technology that can generate biomarkers that have relevance as endpoints for an intervention’s safety or efficacy is under development. 

Multi-sponsor environments can accommodate pragmatic and personalized trials across different trials as the complexity and expense of these trials will be shared across sponsors. It is likely that these systems will not be the classic EHR systems that are used today. 

New statistical designs for future clinical trials

In 2019, more statistical designs will be used that can accommodate streamlined clinical trials. Going from a classically three-phased trial approach to one longer trial that constantly integrates latest insights from the patients, so called cohort expansion trials, is what the FDA has issued in a draft guidance.

Interestingly, the industry association Biotechnology Innovation Organization (BIO) and the Drug Information Association’s Adaptive Designs Scientific Working Group submitted multiple comments to the guidance. The latter group wrote, “While the draft guidance recognizes the desirability of adaptive, learn-as-you-go, seamless designs operating between trial participants, it ignores the real possibility of adaptively individualizing dose within participants on an N-of-1 basis,” as reported by the Regulatory Affairs Professionals Society.

Basket trials test the effect of one drug on a single mutation in a variety of tumor types at the same time. Umbrella trials have many different treatment arms in a single trial, but all tested for one disease. Different treatment baskets group patients that share certain genetic anomalies in their tumors. These studies also have the potential to greatly increase the number of patients who are eligible to receive certain drugs relative to other trials designs. (https://www.asco.org/research-progress/clinical-trials/clinical-trial-resources/clinical-trial-design-and-methodology)

Adaptive trials aim to evolve the treatment given to patients in the trial based on the results during the trial. N-of-1 trials take the personalized treatment even further by considering the treatment of the individual as one experiment by itself. Physicians and patients are blinded for the therapy given, and appropriate controls, such as periods of standard care followed by wash-out periods, are given. They can be properly statistically designed and controlled. The ability to personalize treatment using omics assays, the available of inexpensive and regulatory-grade devices and wearables, and the right regulatory environment will increase the interest in these types of trials. 

Advances in scientific and clinical practice, RWE, the ability to monitor patients with regulatory-grade devices, and the digital health ecosystem will lead to more complex trials that ultimately lead to better therapies. The first step is to embrace the use of real world data with randomized controlled trials to deliver the find the right therapy, for the right patient, at the right time.  

 

Mark Lambrecht, PhD, Director of the Health and Life Sciences Global Practice at SAS

Related Content:

News