The Institution

Clinical AI deployment has outpaced the curriculum that prepares clinicians to use it. The Epic Sepsis Model, deployed across hundreds of hospitals, was found in a Michigan Medicine external validation of 38,455 hospitalizations to have an AUC of 0.63 versus the developer reported range of 0.76 to 0.83. It missed two thirds of sepsis cases while generating alerts on 18 percent of all hospitalizations. The model performed as designed. The training around the model was absent.

The Sororium teaches the clinical judgment required to work with AI. Our fellowships train clinicians to read the test characteristics of AI systems, to distinguish discrimination from calibration, to recognize dataset shift after deployment, to construct clinical actions from probabilistic output, to triage alert fatigue, and to document override reasoning under medico legal scrutiny.

Two populations. Two structurally different AI workflows. Doctors operate AI at decision points. They prescribe, they interpret images, they choose treatments. The training a physician needs covers evaluative literacy and prescribing accountability. Nurses operate AI continuously. They assess, they document, they escalate. The training a nurse needs covers workflow integration, alert triage, and protocol activation discipline. A single curriculum cannot serve both professions well. The Sororium runs separate fellowship tracks for each profession under the same institutional standard.

The Sororium is one of four initiatives under Aethryva Deeptech. PriyAI Sentinel is the bedside intelligence layer. tachyDx is the physician clinical reasoning platform. MOISSCode is the open research layer. The Sororium is the workforce pillar. We train the clinicians who will operate the systems the rest of the stack builds.