Delving into ICON's Use of IBM's Watson

September 21, 2015
Moe Alsumidaie

Lisa Henderson

Lisa Henderson is Editor-in-Chief of Applied Clinical Trials and Pharm Exec. She can be reached at

IBM and ICON plc recently announced a collaboration to use the Watson system to enhance oncology clinical trial feasibility, recruitment and start up via Trial Matching.

IBM and ICON plc recently announced a collaboration to use the Watson system to enhance oncology clinical trial feasibility, recruitment and start up via Trial Matching. Trial matching is a novel concept that emerged a few years ago, and Applied Clinical Trials featured an article on how the method works with aggregated EMR, however, IBM’s approach towards trial matching is a bit different than the methodology outlined in the aforementioned article, as IBM is taking a holistic approach towards patient recruitment and protocol design optimization. 

IBM’s advancement into oncology healthcare research is quite new. News about Watson’s involvement in oncology research emerged around mid-late 2014, as it initiated its Patient Matching proof of concept pilot with The Mayo Clinic. In April 2015, IBM announced that it acquired both Explorys and Phytel, which have offered IBM access to more than 50 million patients at more than 350 hospitals, and tapping hundreds of billions of longitudinal data points, which will likely be used to optimize the Watson engine.

Initially, ICON is applying Watson Clinical Trial Matching to its breast, lung, colon and rectal cancer trials. Cutler told Applied Clinical Trials that the initial pilot with IBM will take place over the next six months, in 25 studies in those four oncologic areas. “Key sites will have access to Watson and Watson has access to the inclusion/exclusion criteria for those studies and access to the records at the sites.” Watson will use the records and inclusion/exclusion criteria to match the up with the EMRs in its database, explained Cutler. Watson has the ability to access and read both these structured data, as well as unstructured data, such as handwritten notes in a patient file.

The pilot is concentrated in the Midwest of the United States, primarily based on the access to the Mayo Clinic and Cleveland Clinic’s records. Cutler expects the company will learn a lot from the pilot, from working through operational procedures to adjusting processes and should have initial feedback on the pilot by end of first-quarter 2016.

IBM’s access to empirical patient data and active patients can translate into significant advancements for cancer clinical research. Below are a few of many potential applications:

  • Protocol Optimization. Sponsors, can leverage the Watson engine to evaluate clinical trial enrollment potential during the protocol design phase; optimizing protocols can maximize patient eligibility pools. 

  • Patient Recruitment and Enrollment.  Study teams can efficiently detect and reach out to patients that closely match a protocol’s inclusion/exclusion criteria in an automated fashion; this process not only expedites the recruitment and enrollment process, but also saves sites a lot of resources and manual effort in medical chart review. 

  • Site Feasibility.  Sponsors and CROs can identify patients at medical institutions before they initiate sites in a study, which can expedite First Patient In (FPI) timelines, and eliminate wasted capital on initiating sites that do not enroll. 

  • Aggregated Control Pools. Accessing an array of empirical patient data enables sponsors to create control pools, which allows for adverse event validations (i.e., is a clinical trial adverse event also commonly observable in patients not on investigational drug), and reduces the total number of required patients in control groups; this means faster clinical trials and less patients needed to fulfill a study’s endpoints.

Cutler spoke to these core capabilities. “The IBM Watson matching gives ICON’s sites the ability to make the process of every patient a candidate for a trial. It is very rapid so that patients and doctors can have that conversation about the right trial for them. The patient gets good information and the physician is saved work,” explained Cutler.

For Watson’s benefit, ICON will be able to add value to the process. “We run a lot of clinical trials, and it’s not as simple as matching inclusion/exclusion criteria. There is informed consent, which we have electronically through our Firecrest eConsent, so we can work that into the process to help with the recruitment,” said Cutler. Further, the experience ICON has in keeping patients compliant and staying in the trial is important.

What’s different about Watson is its cognitive computing abilities, and in clinical trials, Watson seems to be capable of making recommendations to assist study teams with fitting their studies to current patient data. For example, Watson can not only validate protocol enrollment potential against current patient populations, but also recommend changes to optimize inclusion/exclusion criteria and enhance compatibility with patient populations through statistical algorithms. As to protocol optimization, Cutler said, “If a protocol has a screen failure rate of 90%, that means only one in 10 potential patients make it into the trial. And you have to ask yourself how valid is that data and how valid is the protocol? Feasibility of the protocol is a very important part of Watson.”

Watson's approach can translate into notable advances in oncology research, as sponsors can not only recruit very specific patient populations, but also efficiently model studies to fit oncology populations without having to worry much about enrollment. This will create a fertile environment that promotes breakthrough innovations in targeted therapy development.

But There’s Much More to Watson

IBM’s business objectives are focused on fully penetrating and connecting patients, healthcare enterprises, researchers and the biopharmaceutical industry. To elaborate, IBM’s vision is to connect data from clinical trials, patient medical records, research databases, mHealth devices, biosensors and mobile apps, and then advise physicians on the best treatment options. 

In clinical trials, this platform can mean not only recruiting patients through Trial Matching, but, also collecting patient data via mHealth, EMR-connected EDC, and adverse event signal detection and analysis, to name a few.  

IBM is positioned to be a comprehensive beginning to end solution to optimize study design, enhance trial execution, and most importantly, improve the efficiency of the clinical trial platform, bringing targeted cancer therapies to patients at rates that we cannot fathom in today’s environment. Nonetheless, the big question remains: how long will it take for the biopharmaceutical and healthcare industries to adopt a technology as disruptive as Watson in their clinical trials?

Related Content:

Online Extras