Artificial intelligence model predicting postoperative pain using facial expressions: a pilot study
Authors' conclusion
ML models using facial expressions can accurately predict the presence of significant postoperative pain
Study context
- Population
- Patients after gastric cancer surgeryfacial expressions analysed
- Participants
- 155 patients
- Comparison
- AU17 and AU20 vs All action units (AUs)
- Outcome
- Prediction of significant postoperative pain (NRS ≥ 7 vs NRS < 7), AUROC
Methodological detailspilot study, machine learning
- Pilot study
- Machine learning
Performance
As reported in the abstract
Prediction of NRS ≥ 7 (AUROC)
| Test or model | AUC |
|---|---|
| AU17 and AU20 | 0.62 |
| All AUs | 0.81 |
| Head position and facial landmarks | 0.85-0.96 |
| Merged | 0.90 |
AUC
AU17 and AU200.62
All AUs0.81
Head position and facial landmarks0.85-0.96
Merged0.90
- AU17 and AU20 vs All AUs: P = 0.006 — AU17/AU20 model inferior
Further findings
- AU17 (chin raising) and AU20 (lip stretching) associated with NRS ≥ 7
- P ≤ 0.004
- AUs known to be associated with pain in awake patients not associated with pain in postoperative patients
Original publication
Artificial intelligence model predicting postoperative pain using facial expressions: a pilot study. J Clin Monit Comput. Published 27 December 2023. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →