Predicting sepsis onset in ICU using machine learning models: a systematic review and meta-analysis
Problem
Sepsis in ICU patients is often recognized too late, and early recognition is crucial for effective treatment. The study systematically evaluated how well machine learning models predict sepsis onset.
Why it matters
Sepsis has a high mortality rate; earlier prediction could enable timely treatment and save lives.
What the AI did
A PRISMA systematic review/meta-analysis of 23 studies (4,314,145 patients) covering 26 machine learning models for sepsis onset prediction, assessed with C-index, accuracy, and PROBAST risk-of-bias tool.
Results
not stated in source (abstract truncated before C-index/accuracy values)
Impact
Synthesizes evidence that ML models predict sepsis onset, but no deployed validated tool is reported.
Record
| Venue | BMC Infectious Diseases |
|---|---|
| Published | 2023 |
| Work type | review |
| Authors | Zhenyu Yang, Xiaoju Cui, Zhe Song |
| Citations | 68 (OpenAlex, at import time) |
| References | 32 |
| Model or method | random forest and extreme gradient boosting most frequent; model list truncated in abstract |
| Organization | not stated in source |
| Primary source | https://doi.org/10.1186/s12879-023-08614-0 |
| Open access | gold |
| Keywords | sepsis, machine learning, ICU, prediction, meta-analysis |
Importance rating
major — millions of lives, an industry or a field
Abstract states sepsis imposes significant worldwide health and economic burden with high mortality; early recognition is crucial for treatment.