A Machine Learning Model for Predicting Postoperative Oxygen Requirement Among Surgical Patients in Rwanda Using Routinely Collected Patient Data

Background Medical oxygen is crucial for safe postoperative care, yet data on its requirements post-surgery is scarce, particularly in resource-limited settings. Objectives The objective for this study were to identify the key predictors of postoperative oxygen requirement, develop and validate a machine learning model for predicting postoperative oxygen requirements among surgical patients. Methods A retrospective cross-sectional analytical study. Key predictors were identified using descriptive statistics, chi-square tests, and multivariable logistic regression. Nine machine learning algorithms were used to train and validate the model. Results Postoperative oxygen was required by 1,045 patients (28.0%). Key predictors included respiratory disease, higher ASA class, general anesthesia, prolonged surgery, anemia, blood loss, sleep apnea, pediatric or elderly status, and low baseline SpO₂. COPD/asthma was the strongest predictor. LightGBM achieved the highest accuracy (78.88%) and F1-score (77.94%), while XGBoost had the highest AUC (0.8163). Conclusion Machine learning showed potential for predicting postoperative oxygen requirements from routinely collected clinical data. LightGBM, XGBoost, and MLP demonstrated strengths across different performance measures. External validation is needed before clinical implementation.

Authors

Institutions

Publication Details

Journal
Research Square
Published
2026-11-18
DOI
https://doi.org/10.21203/rs.3.rs-11525699/v1
Primary Topic
Artificial Intelligence in Healthcare
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

A Machine Learning Model for Predicting Postoperative Oxygen Requirement Among Surgical Patients in Rwanda Using Routinely Collected Patient Data

Jean Marie Vianney Hope, Christophe Mpirimbanyi, Emmy Mugisha, Melissa Uwase et al.
6 links
Research Square
Artificial Intelligence in Healthcare
preprint

A Machine Learning Model for Predicting Postoperative Oxygen Requirement Among Surgical Patients in Rwanda Using Routinely Collected Patient Data

Jean Marie Vianney Hope, Christophe Mpirimbanyi, Emmy Mugisha, Melissa Uwase, Emmanuel Kalisa, Pasteur Dushimimana, Theogene Sengarambe, Noa Irabizi, Goreth Mukakayindo, Fatima Hammad Mohamed Adam, Bertin Irakoze Ringuyeneza, Norbert Rukundo Izere, Joel Munyaneza
preprint en

Abstract

Background Medical oxygen is crucial for safe postoperative care, yet data on its requirements post-surgery is scarce, particularly in resource-limited settings. Objectives The objective for this study were to identify the key predictors of postoperative oxygen requirement, develop and validate a machine learning model for predicting postoperative oxygen requirements among surgical patients. Methods A retrospective cross-sectional analytical study. Key predictors were identified using descriptive statistics, chi-square tests, and multivariable logistic regression. Nine machine learning algorithms were used to train and validate the model. Results Postoperative oxygen was required by 1,045 patients (28.0%). Key predictors included respiratory disease, higher ASA class, general anesthesia, prolonged surgery, anemia, blood loss, sleep apnea, pediatric or elderly status, and low baseline SpO₂. COPD/asthma was the strongest predictor. LightGBM achieved the highest accuracy (78.88%) and F1-score (77.94%), while XGBoost had the highest AUC (0.8163). Conclusion Machine learning showed potential for predicting postoperative oxygen requirements from routinely collected clinical data. LightGBM, XGBoost, and MLP demonstrated strengths across different performance measures. External validation is needed before clinical implementation.

Research Square
University of Rwanda (RW), University of Global Health Equity (RW)
Good health and well-being
Artificial Intelligence in Healthcare
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon