Cognitive computing in healthcare
A collaboration between IBM and Boston Children’s Hospital to use Watson’s AI as a service for healthcare around the world.
CarePlex used IBM’s Watson predictive analytics technology to create a dynamic medical record for patients. Targeting clinicians, patients, and caregivers, the Watson technology provided teams with a platform for supporting the holistic needs of individuals. It used a rich store of medical knowledge in combination with patients’ medical histories, aggregated knowledge, and personal medical observations to suggest diagnoses and recommend treatments.
As the lead product designer for the project, my task was to prototype how Watson could be applied in professional medical practice.
Bringing value to physicians, patients, and caregivers
In order to be valuable in a clinical environment, a system has to be able to distill the huge corpus of information that comprises a particular patient’s complex clinical situation (including symptoms, condition, drugs, therapies, and treatments). The challenge was to present this clinical data in a way that provided insight into the underlying logic behind Watson’s recommendations while ensuring that the most salient points remain clear and accessible.
To make this happen, I designed an experience mediated through natural language and validated through transparency, with direct access to external resources to verify Watson’s diagnoses and recommendations.
Natural Language
My first decision was to leverage Watson’s abilities for more than just medical diagnoses. The primary form of interaction relied on natural language input and output. This saved caregivers’ cognitive overhead for critical tasks in emergency situations, letting them focus all of their attention on the matter at hand.
A Living Medical History
An electronic medical record incorporating hospital records, medical reports, and real-time data from connected sensors.
Transparency
Visualization of Watson’s predictive analytics provided insight into how it drew its conclusions. Doctors could assess an individual’s needs and select interventions with greater confidence and accuracy.
Verifiability
Automatic recommendations of potential resources to supplement and/or clarify Watson’s recommendations, including journal articles, instructional videos, tests, and connections to known subject matter experts.
Physicians, patients, and caregivers could see a list of experts who could provide a second opinion, and connect with these individuals through the secure CarePlex community.
The Outcome
The greatest potential for CarePlex came from the fact that it was a rapidly deployable, cloud-based SaaS solution that connected stakeholders in communities around the world. Because the interaction was mediated through natural language, the platform was easily adapted to work anywhere without training. The project evolved and today supports a global community of medical professionals who use Watson services as a platform for medical practice.
