Learning Lab: Intelligent or Just Artificial? – The Emerging Quality Challenge of Dealing with Our Smart Machines-Intelligent or Just Artificial mp4

Video Transcription

Video Summary

NAHQ hosted a Learning Lab on the emerging quality challenges of “smart machines” in healthcare, presented by quality consultant Ken Rohde. He explained how clinical decision-making has evolved from individual mental models to written algorithms, to software-based decision support, and now to machine learning systems that continuously adapt based on feedback. While algorithms can improve repeatability and safety, Rohde warned that software-driven systems can fail in high-stakes ways, citing the Therac-25 radiation overdoses and the Boeing 737 MAX crashes as examples of design, testing, training, and reporting breakdowns.<br /><br />Rohde emphasized that healthcare quality professionals don’t need to be software experts, but must provide oversight—especially through validation (ensuring a system meets user needs safely), incident reporting, and tools like FMEA and root cause analysis. He recommended asking whether vendors have structured design, integration/testing, and formal validation processes, even when products are “black boxes.” For machine learning, he highlighted risks such as poor data quality, model drift, and bias, and advised strengthening statistical literacy and partnerships with IT/biomed. In Q&A, he noted automation may replace some routine tasks, but increases the need for quality oversight.

Keywords

smart machines in healthcare

machine learning clinical decision support

healthcare quality oversight

software validation and verification

incident reporting systems

failure mode and effects analysis (FMEA)

root cause analysis (RCA)

model drift and algorithmic bias

Therac-25 and Boeing 737 MAX safety failures