Healthcare is a natural fit for Cyc’s differentiators: transparency, context-sensitivity, user modelling, and value partnering.
- Transparency: The stakes are high. People’s lives and quality of life are at stake, which means that healthcare providers face scrutiny from their patients and regulators in addition to needing to deliver a high standard of care. Head nurses need explainable AI, rather than just some black box that says that the Emergency Department is going to be short-staffed in a few hours. Patients need to be reassured that AI understands them as individuals.
- Context Sensitivity: Healthcare is too complex to resolve with a set of one-size-fits-all decision trees. Hospitals, for instance, are constantly experiencing disruptive events: surges, equipment failures, staffing issues, full ICUs…the list goes on. And each of these events has consequences that ripple throughout the facility. Cyc has the awareness to see how all of the pieces fit together and can therefore stay ahead of the pitfalls that will sideline more brittle technologies.
- User Modelling: Cyc excels here for two reasons. First, the expressivity of the language allows Cyc to appreciate the nuances of individuals that other systems are forced to gloss over. Second, taking advantage of a user model requires a great deal of common sense. For example, suppose that we know a patient is a tennis player. Then we’d expect treatments or medications that have an impact on, say, shoulder mobility, to be especially relevant to their life. Similarly, the religious affiliation of a patient conveys a wealth of information about how a patient may prefer to be treated, as long as the AI is smart enough to understand the content and consequences of religious beliefs.
- Hospital Throughput and Productivity Advisor: Cyc improves patient care and throughput by monitoring data feeds from across the facility, including real-time patient data, to build models of individual patients and track them throughout their care progression.
Healthcare Research Clinical Trial
Goal: For Cleveland Clinic to improve its efficiency in identifying cohorts across their Epic Enterprise for clinical trials and in reporting procedures and outcome data.
Solution: Cycorp’s Semantic Research Assistant software enabled CCF medical personnel to easily and rapidly specify ad hoc queries to support their own clinical research and to support CCF’s quarterly external reporting requirements.
Results: Answers to researchers’ clinical queries can be obtained in minutes rather than the weeks (or even months) required when the process was manual. This allows greater pre-trial exploration during cohort formation: Ten queries could be formulated and asked and answered in one afternoon instead of one year.
Building on the same Cyc-based knowledge models, an automated data reporting capability was implemented and was certified by both the Society of Thoracic Surgeons and the American College of Cardiology. Report generation times were reduced by 35% and many errors in reports were caught and corrected for the first time.