Perioperative medicine & outcomes
Digitization of Symbol-Denoted Blood Pressure Data From Intraoperative Paper Health Records in a Low-Middle-Income Country Using Deep Image Segmentation and Associated Postoperative Outcomes: A Feasibility Study
Authors' conclusion
In this study, we report our experience with a deep image segmentation model for digitization of symbol-denoted blood pressure from intraoperative anesthesia PHRs.
Study context
- Population
- Intraoperative paper health records (PHRs) in a LMICintraoperative PHRs imaged
- Participants
- 350 participants
- Comparison
- Systolic blood pressure (SBP) vs diastolic blood pressure (DBP)
- Outcome
- Detection performance and prediction error of U-Net digitization of hand-drawn BP symbols vs ground truth
Methodological detailsdeep image segmentation (U-Net), smartphone mHealth platform
- Deep image segmentation (U-Net)
- Smartphone mHealth platform
Main finding
As reported in the abstract
BP symbol detection and prediction error
| Group | Value | Overall accuracy |
|---|---|---|
| SBP | mean 2.1 (SD 4.9) mm Hg | 0.839 |
| DBP | mean −0.8 (SD 3.9) mm Hg | 0.911 |
Value
SBPmean 2.1 (SD 4.9) mm Hg
DBPmean −0.8 (SD 3.9) mm Hg
Overall accuracy
SBP0.839
DBP0.911
Further findings
In-hospital mortality by minutes with DBP <50 mm Hg
| OR (CI) | P value | ||
|---|---|---|---|
| Minutes DBP <50 mm HgMultivariate regression | 1.03 (1.01–1.05) | P = 0.003 | |
| 0.512Odds ratio (log scale) |
LOS by cumulative minutes with SBP 80–90 mm Hg
| IRR (CI) | P value | ||
|---|---|---|---|
| Minutes SBP 80–90 mm HgLonger LOS | 1.02 (1.0–1.03) | P < 0.05 | |
| 0.512Incidence rate ratio (log scale) |
| Outcome | IRR (CI) | P value |
|---|---|---|
| LOS by cumulative minutes with SBP 140–160 mm HgShorter LOS | 0.9 (0.96–0.99) | P < 0.05 |
Original publication
Digitization of Symbol-Denoted Blood Pressure Data From Intraoperative Paper Health Records in a Low-Middle-Income Country Using Deep Image Segmentation and Associated Postoperative Outcomes: A Feasibility Study. Anesth Analg. Published 25 August 2022. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →