Perioperative medicine & outcomes
Novel Preoperative Risk Stratification Using Digital Phenotyping Applying a Scalable Machine-Learning Approach
Authors' conclusion
For 3 frequently performed surgeries, we identified 3 digital phenotypes. The typical profiles of each phenotype were described and could be used to anticipate adverse postoperative events.
Study context
- Population
- Patients undergoing laminectomy, colectomy or thoracic surgery over a 9-year period in a large hospital system2,770 in validation dataset
- Participants
- 7,251 patients
- Exposure
- Digital phenotypes alpha, beta, gamma; phenotypes combined with American Society of Anesthesiologists (ASA) score vs ASA score alone
- Outcome
- Association of preoperative digital phenotypes with postoperative adverse events:
- In-hospital and 30-day mortality
- 30-day surgical redo
- Intensive care unit (ICU) admission
- Hospital length of stay (LOS)
Methodological detailsmachine-learning (consensus k-means clustering, random forest), derivation and validation cohorts split chronologically, single hospital system
- Machine-learning (consensus k-means clustering, random forest)
- Derivation and validation cohorts split chronologically
- Single hospital system
Main finding
As reported in the abstract
Postoperative adverse outcomes by phenotype (validation cohort)
30-day mortality
In-hospital mortality
Prolonged hospital LOS
Observational data: associations, not effects.
Further findings
Phenotype frequency (validation cohort)
Discrimination (AUROC): digital phenotypes combined with ASA score vs ASA score alone
- Hospital mortality AUROC, phenotypes + ASA vs ASA alone
- 0.91 vs 0.84
- Prolonged hospitalization AUROC, phenotypes + ASA vs ASA alone
- 0.80 vs 0.71
Parts of this Brief are withheld: the automatic checks could not confirm them against the abstract.
Original publication
Novel Preoperative Risk Stratification Using Digital Phenotyping Applying a Scalable Machine-Learning Approach. Anesth Analg. Published 5 December 2023. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →