Faculty Opinions recommendation of A clinically applicable approach to continuous prediction of future acute kidney injury.
Problem
About 11% of hospital deaths follow failure to promptly recognize deteriorating patients; continuous, accurate, individual-level risk prediction is needed, using acute kidney injury as an exemplar.
Why it matters
Delayed recognition of life-threatening deterioration such as acute kidney injury can be fatal.
What the AI did
A deep learning model continuously predicts future patient deterioration from longitudinal electronic health records across diverse clinical environments.
Results
Predicts 55.8% of all inpatient acute kidney injury episodes; developed on 703,782 adult patients across 172 inpatient and 1,062 outpatient sites (abstract truncated)
Impact
Faculty Opinions post-publication peer review endorsed the underlying study; clinical adoption not stated in source.
Record
| Venue | Faculty Opinions – Post-Publication Peer Review of the Biomedical Literature |
|---|---|
| Published | 2019 |
| Work type | peer-review |
| Authors | Kevin Yuk-Lap Yip |
| Citations | 0 (OpenAlex, at import time) |
| References | 67 |
| Model or method | not stated in source |
| Organization | not stated in source |
| Primary source | https://doi.org/10.3410/f.736293810.793564412 |
| Open access | green |
| Keywords | acute kidney injury, deep learning, electronic health records, risk prediction |
Importance rating
significant — thousands to millions affected
Abstract states an estimated 11% of hospital deaths follow failure to promptly recognize deteriorating patients; model developed on 703,782 patients across 1,234 sites, indicating large-scale clinical relevance.