What happens after a model is built and deployed and is in use by clinical stakeholders? Changes in patient data and their relationships are ineluctable, and models must be maintained, surveilled, and adapted accordingly to continue performing well. This is the world of Machine Learning Operations.

George Mastorakos believes combining data types into what are called "multimodal models" may be the key to moving clinical artificial intelligence into the next phase of better performance and broader applicability of clinical decision-making.

Trust is at the center of all healthy relationships, including patient-clinician and clinician-machine ones. If trust is not at the foundation of the clinician-machine relationship, useful AI algorithms will not see the light of day. The way to build this trust is through AI healthcare teams showing patients how it works.

The ubiquitous adoption of modern electronic health records has fostered the storage, management, and sharing of virtually all of an individual’s protected health information. This includes everything from a patient’s address to their heart rate variability during a morning run to X-ray images. However, without proper security measures, these capabilities are merely digital castles made of sand.