September 2026 Session: Encoding Equity: A Practical Toolkit for AI Implementation-Encoding Equity: A Practical Toolkit for AI Implementation

Video Transcription

Video Summary

The webinar introduced the Encoding Equity AI Implementation Toolkit, a practical framework for evaluating and governing AI throughout its use in healthcare. Speakers emphasized that the toolkit is a starting point—not a checklist or certification of safety—and that organizations must adapt it to their patients and communities.<br /><br />The framework covers six connected areas: data equity and representation; algorithmic accountability; causal reasoning; model transparency and explainability; inclusive governance and stakeholder engagement; and ethical review, monitoring, remediation, or de-implementation. A sepsis-risk example illustrated how teams can examine training data, question model inputs, assess performance across groups, involve affected communities, and pause deployment when risks emerge.<br /><br />Panelists called for continuous, lifecycle oversight rather than one-time approval, with clear responsibility for detecting problems and acting on them. They urged organizations to make questioning AI a normal part of responsible practice and to give patients and communities meaningful influence—not merely a seat at the table. Governance should distinguish regulatory compliance from broader responsibility, consider the whole clinical system, and keep learning after deployment. Completing the toolkit does not prove a model is safe; ongoing evidence is needed to show it benefits patients equitably.

Keywords

AI implementation toolkit

healthcare AI governance

data equity

algorithmic accountability

model explainability

inclusive stakeholder engagement

continuous monitoring

ethical review

equitable patient care