2438. Artificial Intelligence Model Using Deep Learning for Operating Room Time Prediction
Invited abstract in session TC-13: AI in healthcare, stream OR in Healthcare (ORAHS).
Tuesday, 12:30-14:00Room: Clarendon SR 1.01
Authors (first author is the speaker)
| 1. | Luís Lapão
|
| Intelligent Decision Support Systems Laboratory, Universidade Nova de Lisboa |
Abstract
Introduction: Efficient management of surgical centers requires optimizing operating room time while minimizing team idle time. This study presents a deep learning-based tool designed to predict operating room time and patient destination, initially implemented for scheduling coordinators at ICHC-FMUSP (Central Institute of the Hospital das Clínicas, Faculty of Medicine, University of São Paulo) to enhance scheduling efficiency and reduce idle time.
Methods: A non-parametric deep learning model was developed using a multilayer perceptron approach (using Python libraries), with anonymized variables for privacy purpose and addressing GDPR. Surgical data from 2016 to 2021 was used. After exploratory analysis and the exclusion of combined procedures and non-elective surgeries, a total of 31,090 elective surgeries (70% of the dataset) were used for model training.
Results: The weighted average between the current OR time prediction method and the deep learning model resulted in an algorithm with an accuracy of 0.78, according to Lin’s concordance correlation coefficient. Predicted and actual times showed no significant difference (Hosmer-Lemeshow test: χ² = 12.45, p = 0.052).
Conclusion: The artificial intelligence (deep-learning) model improved operating room time prediction accuracy, demonstrating potential benefits in optimizing surgical scheduling, resource utilization and reducing inefficiencies.
Keywords
- Health Care
- Scheduling
- Artificial Intelligence
Status: accepted
Back to the list of papers