einstein (São Paulo). 15/ago/2023;21(Suppl 1):EISIC_MV0020.
Dilemmas and possibilities in the development of Science of Improvement projects involving the use of Artificial Intelligence in Healthcare
DOI: 10.31744/einstein_journal/2023ABS_EISIC_MV0020
III Einstein International Symposium on Intensive Care and the XXX International Symposium on Mechanical Ventilation. Aug 16-18, 2023.
Category: Safety / Quality / Management
Introduction:
In recent years, there has been great interest in Artificial Intelligence (AI) applications in healthcare. However, the issues related to AI’s implementation, ethical use, and safety are not consensual and have not yet been adequately addressed.()
Objective:
Report the experience of the Einstein Network for Critically Ill Patients, focusing on the dilemmas and possible paths for the use of Science of Improvement (SI) tools in the study of potential applications of AI in healthcare (specifically, in the management of severe patients or those at risk of clinical deterioration).
Methods:
Report of the experience and presentation of insights based on a prospective observational study (Proof of Concept) conducted from February 2022 to June 2023 in the Semi-Intensive Unit of the Hospital Israelita Albert Einstein Hospital (HIAE), as part of the Improvement Advisor Course from the Institute for Healthcare Improvement (IHI).
Results:
In phase 1, data was collected from the monitoring center, while in parallel, an AI algorithm, previously trained with retrospective data, was also fed (without contact with the healthcare team). After assessing the non- inferiority of the algorithm compared to the monitoring center, phase 2 (POC), which involved creating a user interface for physicians and actual use, was authorized.
Conclusion:
The experience report showed that using SI tools in a project integrating AI to redesign an early detection process for clinical deterioration in the Semi-Intensive Unit is feasible and beneficial.
98
