Data, AI And Machine Learning Are Propelling Clinical Trials Data Analysis To New Heights

The quality of clinical trials and data analysis have come a long way in the past 10 years – and even in the past three, says Christian Hebenstreit of Medidata, the cloud-based platform that supports pharma and medtech companies through clinical studies ahead of their commercial uptake. The buzzwords of the fairly recent past are now being applied as real tools in the clinical trials arena, and look set to transform the sector.

The London arm of Medidata NEXT, the seven-venue awareness event for clinical trials professionals around the globe, is humming with industry players keen to hear how they can use Medidata Solutions Inc.'s Clinical Cloud tool, which combines science and the operation of clinical trials. At the podium, US product development executive Dave Metz is explaining the MEDS (Medidata Enterprise Data Store) element of the offering: the repository of clinical and operational data that are the basis of clinical process decision-making.

Metz's tone is irrepressibly upbeat: Medidata's service offerings to clinical clients are increasing in speed and quality across the industry. And investors are similarly optimistic: the group has something of a financial tailwind, with Q1 2018 revenues rising 17% to over $149 million and shares not far off their all-time high of 10 months ago. Moreover, the newest agreement by Medidata means that all of the top five CROs are now using Clinical Cloud, whereas another top 15 pharma company signed up to an enterprise agreement in Q1, and 17 new imaging customers were added.

Christian Hebenstreit, Medidata general manager and SVP for the EMEA region, told In Vivo during a break in presentations that the growth has been built on a platform of stability. The group's co-founders, President Glen de Vries and Chairman and CEO Tarek Sherif, are still in place 19 years on. Eighteen of the top 25 pharmas use Medidata as their platform to conduct clinical studies, and 250 of the group's 1,000 customers globally are based in Europe, where Hebenstreit, who joined two years ago from salesforce.com, is based (at the London, UK offices).

Medidata's cloud and analytics expertise is seen as a tool to connect all parties involved in clinical trials: doctors, nurses, sites and companies – mainly pharma, but device and diagnostics firms too. "We help clients capture the data, and there are different ways to do it now as almost every trial uses mobile data, or data from sensors or wearable devices, for instance," Hebenstreit notes. The amount of data gathered through a clinical trial is increasing massively, and the number of trials is doubling every year, due principally to the capabilities that big data offers.

Having conducted more than 14,000 studies over the past 10 to 15 years involving over 4 million trial patients, Medidata has amassed billions of data points. It is now starting to bring together the experiences gathered over its historical studies, says Hebenstreit. In some areas, studies have already been checked by other companies. "We are trying now to package it all up and build tools to simulate control groups," he says, stressing "We are reaching the point where we can create synthetic control groups through data."

Medidata can do this with its own data and with data where it has anonymization rights. Most of its clients consent to the use of their anonymized data. Hebenstreit is encouraged that parts of the business are now making huge inroads on the use of big data. Machine learning and AI capabilities can be used even for ostensibly easy tasks, he notes. For example, one aspect that has always led to problems is the coding process, and Hebenstreit observes that there is a multitude of terms for "headache," say, but they all mean the same thing.

"Overall, our industry is now really starting to make use of all these capabilities – it's a fascinating time." Medidata's Christian Hebenstreit

Pharma companies involved in trials may have many hundreds of staff just doing this type of coding to define a single data point. "We can do this with machine learning capabilities, and it's not always about the 'big bang' effect of these advanced technologies: it's often a case of just making small but significant steps with ML and AI. But overall, our industry is now really starting to make use of all these capabilities – it's a fascinating time."

Medidata is developing proprietary tools in many growth areas, such as in mobile health technology, allowing the group to read mobile health-generated data; in payment solutions, enabling patients and sites to be paid on time – not always an easy task in the past, given the international nature of trial programs; and in managing the risk, verifying that the data are good enough, for which the group has its own tools to double- and triple-check data.

The Challenge Of Unstructured Data

The ability of trials to improve the quality of insights is a major differentiator, says Hebenstreit, recalling the difference in trial quality in 2018 compared with, say, 2005. In a crowded environment, Medidata uses internal and external data, refined by a sizeable data science team established three years ago.  "The big task now is to make unstructured data usable in a CT. Our own data are already structured and can be used easily," Hebenstreit points out. But making unstructured data from external sources work – qualifying and gathering insights from it – is a major task for the 60+ data experts at Medidata's New York and London offices.

"Synthetic controls are a big thing for patients, and both economically and ethically represent a major shift within life sciences and R&D."

Last year, Medidata became the first non-life science company to be selected to present at the American Society of Clinical Oncology (ASCO) annual meeting. In two abstracts, Medidata reviewed its scientific findings from a synthetic control arm (SCA) which was created in acute myeloid leukemia; and presented a clinical trial genomics (CTG) algorithm that generated a scientific finding in breast cancer. University of Texas MD Anderson Cancer Center professor Don Berry, PhD, said at the time that synthetic control arms, created out of large pools of patient data, offer the possibility of early insight during clinical development programs.

Hebenstreit adds that synthetic controls are cheaper and faster. Not only are they hugely interesting from an ethical standpoint, but also more trial participants will be the recipients of newer, better treatments, with fewer getting the placebo or the control. "Synthetic controls are a big thing for patients, and both economically and ethically represent a major shift within life sciences and R&D."

No Longer Mere Buzzwords

It is clear that real shifts are happening in the field of clinical trials generally. "We've been talking about big data and data science for many years now, they were already buzzwords 10 years ago, and yet so few people still know what this is all about," says Hebenstreit. In fact, few companies were actually ready for it10 years ago, but with all the quantum leaps in technology – the cloud, AI, machine learning – this has changed definitively. The future looks distinctly alluring, given that the capabilities needed to really engage with data science have only been in place for some two to three years. "We are seeing a major change," he declares.

But patient confidentiality and data privacy standards are not about to change, and will remain high. Operationally, Medidata cannot engage with patients and neither do the pharma companies; only the participating sites and site investigators can. Medidata’s experts work closely with the FDA. "We know how to handle data – personal data, ID data, basically, patient data," Hebenstreit says. This is the most sensitive topic. The data are pseudo-anonymized, and on top of that, all the clinical patient data are encrypted.  As a further layer of protection in the EU, compliance with the new GDPR (General Data Protection Regulation) will be enforced after May 25, 2018. And in 2019 the EU's new Clinical Trials Regulation will define the specific processes that companies engaged in clinical trials are required to adopt.

"We are giving GDPR a lot of attention. Every company that in any way deals with human beings or patients need to respect GDPR – not just physicians or people working on products." The subject's "right to be forgotten" is a hot topic, and remains in discussion in the life sciences arena, but the issue for this industry is that it is difficult for a patient who has given consent to participate in a clinical trial in the past to withdraw it. As a result, the overall R&D space has been carved out of the regular right to be forgotten concept.

Patients will still need to give written consent that they are willing to participate in a trial and that their data can be used to improve the quality of information of the trial. There are various ways to do this, such as paper-based informed consent; or the e-consent solution, via iPad or another device type that allows patients to really screen through the process of the clinical trial and investigators to better account for the consent process. This also makes it easier for those conducting the trial to explain to patients what is happening over the coming months and years.

More And Bigger Trials

The amount of data being gathered is proof that pharma companies are doing more and bigger trials. Medtech is a relatively small part of the business for Medidata, and the needs are different. Trials for, say, Siemens AG or Novartis AG would be completely different.

Today, there is a lot of attention on cancer, diabetes, and Alzheimer's research, but Medidata is also working with clients on rare disease areas, which are smaller studies in terms of patient participation.

As to geographic regions, Europe is one of Medidata’s fastest growing markets, as well as Asia. These two areas will drive a lot of life sciences business activity over the coming year or two, Hebenstreit believes.

The UK, Switzerland, Germany, France and Denmark are strong national markets within Medidata's EMEA sector. The UK is growing quite substantially, says Hebenstreit, who believes that even with Brexit, and the European Medicines Agency (EMA) moving to Amsterdam, the Netherlands, this won't change. Israel and its "hundreds of biotech companies" is another important business area for the EMEA business sector.

"As a cloud provider, we have always worked in many different regulatory environments and under different laws. That's one of the beauties cloud computing: all you need is an internet connection and a browser," Hebenstreit continued.

Near-Term Business Drivers

He also observes that the regulators are increasingly interested in making the running: "They are really pressing for more use of new technologies such as data science and synthetic control groups. It is not a case of getting pushback from the regulators, rather the EMA and FDA are actually doing a lot of the asking. It's quite surprising." They are doubtless mindful of their cost-effectiveness, and of the advantages to be gained from an ethical standpoint.

The key consideration is that the technology adds real value and is not "hype." Machine learning algorithms and AI applied to certain areas can provide real benefit in the clinical trials arena, says Hebenstreit. By contrast, can the same be said for, say, Blockchain used by supply chain partners, or is the buzz around this particular cloud tool merely a passing phase? It is actually a self-fulfilling prophecy in the end – if a technology provides real help, it will be adopted.

More generally, the volume of clinical studies will increase in the future, there will be greater attention to rare diseases, and trials will become much more granular. Medidata plans to keep developing its position by continuing to invest a quarter of its revenue into building new technology solutions and making new releases that help clients. Rivals might choose to spend on marketing and sales, whereas Medidata invests in R&D, Hebenstreit observes. And the group is profitable, he adds, which is not always the case in the business of cloud computing.