What question did this study set out to answer?

To identify and evaluate the technical and operational barriers preventing the practical implementation of machine learning models in intensive care units.

December 21, 2022Communications MedicineOpen Access

Identifying and evaluating barriers for the implementation of machine learning in the intensive care unit

View Full Paper

Ask AI

Bookmark

Authors

EDEllie D’HondtMercy Health TAThomas J. AshbyIMEC ICImen ChakrounIMEC

Discussion

Member takes

Implication

Mixed-methods study reveals substantial predictive model performance loss under real-world clinical shifts in intensive care units, highlighting critical barriers to practical AI adoption.

Key Points

To identify and evaluate the technical and operational barriers preventing the practical implementation of machine learning models in intensive care units.
Conducted a qualitative study comprising 29 interviews with 40 staff members from intensive care units, hospitals, and MedTech companies.
Performed quantitative experiments using two clinically relevant predictive models to measure performance degradation caused by data drift over time, altered patient features, data scarcity, and out-of-context deployment.
Qualitative findings confirmed widespread acknowledgment of AI potential alongside pervasive technical barriers that obstruct practical clinical integration.
Quantitative experiments demonstrated that data drift, feature variation, scarce training data, and cross-context deployment each induce substantial loss of predictive performance depending on the model.

Cite This Study

D’Hondt et al. (2022) studied this question.

synapsesocial.com/papers/6a1734c9d17772c818bb88d0 https://doi.org/10.1038/s43856-022-00225-1

View Full Paper

Ask AI

Bookmark

Implication

Mixed-methods study reveals substantial predictive model performance loss under real-world clinical shifts in intensive care units, highlighting critical barriers to practical AI adoption.

Identifying and evaluating barriers for the implementation of machine learning in the intensive care unit

Discussion

Member takes

Implication

Key Points

Cite This Study

Identifying and evaluating barriers for the implementation of machine learning in the intensive care unit

Discussion

Member takes

Implication

Key Points

Cite This Study

Also Consider

Also Consider