How Watson for Clinical Trial Matching is Accelerating the Screening Process

Share this post:

Innovation, by its nature is new, and having never been done before, is hard.  To do something for the first time takes much more effort and concentration than when there is muscle memory, a process, precedent to follow.

To do something innovative with multi stakeholder input – provider, pharma, technology – adds another dimension of challenge. It is hard, really hard, to keep all parties focused and to keep removing unforeseen barriers, be they regulatory, technical, or environmental.

Last week the IBM Watson Health, Novartis and Highlands Oncology team were announced as winners in the Best Sponsor Focused Technological Development Category at the Clinical & Research Excellence Awards.

I could not be more proud of the team and am so happy that the teams’ efforts were recognized. This was a partnership not just with man and machine (Watson) but a mutually beneficial voyage of discovery with Novartis, IBM Watson Health and Highlands Oncology, a community cancer center. Together we showed that screening workload can be dramatically reduced with the help of Watson for Clinical Trial Matching. We also proved that just because it is serious doesn’t mean it can’t be fun.

Advances towards more personalized cancer treatments increases the complexity of clinical trial recruitment, and despite NCCN guidelines stating that the best management for any cancer patient is a clinical trial  [1]. The identification of potentially eligible patients is a highly manual, lengthy process subject to error.

Today, IBM Watson for Clinical Trial Matching is a commercially available solution designed to help Oncologists match a patient to a clinical trial. Watson is integrated into the EMR system and uses structured and unstructured (pathology reports, lab reports, physician notes) patient data to quickly check eligibility across potentially relevant clinical trials from CT.gov or investigator trials.

Physicians are provided with eligible trial considerations ranked by relevance, in real time, during the patient consultation. This project was initiated to demonstrate that Watson could be trained in the full inclusion / exclusion criteria and evaluate thousands of patients against one protocol (rather than start with one patient and look across many protocols).

The team demonstrated technical feasibility in the real world community setting. Metrics indicated overall satisfaction with the process, reduced site screening time / productivity enhancement and a five-fold increase in evaluable patient criteria over that which is posted in CT.gov. Digital protocol assessment showed several areas for Novartis to improve protocol authoring and write clear non-ambiguous protocols that both Watson and Oncologists can understand.

This was a proof of technology, a first step, the first flex of our muscles in automated trial screening with real world real time data.  We will advance the process of innovation, we will continue to learn, we will expand use cases and we will scale, globally.  As I reflect on where we have been it makes me even more excited for the future, I know there will be challenges but I also know we have the best team working together, to make a difference.

______________________________________

This story first appeared on the IBM Watson Health blog, Perspectives.

[1] https://www.ncbi.nlm.nih.gov/books/NBK220370/