Multi-Modal Machine Learning for Prediction of Post-Anesthesia Care Unit Recovery Time

Multi-Modal Machine Learning for Prediction of Post-Anesthesia Care Unit Recovery Time | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Multi-Modal Machine Learning for Prediction of Post-Anesthesia Care Unit Recovery Time Mohammad Noorchenarboo, Michelle Kwong, Katarina Grolinger, Ahmad Elnahas This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8904195/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted 7 You are reading this latest preprint version Abstract Background Efficient recovery management in the Post-Anesthesia Care Unit (PACU) is important for patient safety and care coordination, yet estimating recovery duration remains challenging. Traditional regression models and discharge scoring systems have shown limited accuracy and reliability across patient populations. Methods We developed and evaluated machine learning models for PACU recovery prediction using an institution-wide dataset of 203,393 surgical cases collected between January 2012 and July 2025. Structured variables (demographics, surgical metrics, provider identifiers, timing features) were integrated with unstructured operative narratives transformed into biomedical SBERT embeddings, yielding a multi-modal dataset. Model performance was evaluated using robust validation procedures to ensure findings were not dependent on a single patient group or time period. Results Gradient-boosted tree ensembles demonstrated the best performance, with XGBoost achieving a mean absolute error of approximately 42 minutes and explaining over one-third of the variance in recovery time. Nearly 70% of predictions fell within 60 minutes of observed recovery duration. Statistical testing confirmed significant improvements over linear baselines, conventional ensembles, and neural networks. Explainability analyses revealed that operative narratives provided the greatest predictive signal, while provider identifiers and surgical location also contributed, reflecting both patient-level complexity and system-level workflow factors. Conclusions Combining structured perioperative variables with embeddings of operative narratives enables accurate and interpretable PACU recovery prediction at scale. These findings demonstrate the feasibility of multimodal prediction of PACU discharge-ready time and provide a foundation for future evaluation of operational and clinical applications. Future work should include external validation and integration into real-time perioperative systems. PACU recovery time perioperative prediction machine learning gradient boosting artificial intelligence Figures Figure 1 Figure 2 Figure 3 Figure 4 1. Background The Post Anesthesia Care Unit (PACU) is among the most resource-intensive areas of the hospital, where respiratory, cardiovascular, and neurologic complications frequently arise and may prolong recovery [ 1 – 3 ]. Extended PACU stays are influenced by patient and procedure-related factors such as comorbidities and perioperative conditions [ 4 , 5 ], and they contribute to downstream bottlenecks in operating room throughput, bed availability, and overall hospital efficiency [ 6 ]. As surgical volumes increase and healthcare systems face capacity pressures, anticipating PACU recovery duration has become critical for both patient safety and operational performance. PACU staffing is typically determined by anticipated census rather than individual discharge times. Therefore, the clinical value of individual-level prediction lies primarily in identifying prolonged recoveries and supporting coordination rather than replacing established staffing models. Accurate prediction of PACU recovery time has direct implications for perioperative efficiency. Although PACU staffing in many institutions is based on high-percentile hourly demand forecasting rather than individual recovery prediction, patient-level estimates may support care coordination, identification of prolonged recoveries, and downstream bed planning. Current prediction methods remain limited, such as the American Society of Anesthesiologists (ASA) Physical Status classification, which offers only coarse risk stratification and shows inconsistent associations with PACU complications or length of stay (LOS) [ 7 , 8 ]. Regression-based models using factors such as operative duration, comorbidities, or anesthetic technique demonstrate modest accuracy [ 9 – 12 ], and discharge scoring systems fail to account for individual variability in recovery [ 13 ]. Recent advances in perioperative Electronic Health Records (EHRs) and machine learning provide opportunities to overcome these limitations. Predictive models based on artificial intelligence have already improved forecasting of outcomes such as delirium, ICU admission, and unplanned readmission [ 17 , 18 ]. Large-scale efforts combining structured perioperative data with polygenic risk scores or AI frameworks have further enhanced prediction of complications like postoperative nausea, vomiting, or residual neuromuscular block [ 19 , 20 ]. Yet, few studies have directly modeled PACU recovery duration, and applications of language models to this task remain limited [ 21 ]. The objective of this study is to develop and benchmark multimodal machine learning models for predicting PACU recovery duration using a decade of institutional surgical data. We hypothesized that integrating structured perioperative variables with unstructured perioperative narratives, transformed into biomedical language model embeddings, would yield more accurate and interpretable predictions of PACU recovery time than conventional regression approaches. 2. Methods 2.1. Study Design and Ethical Approval We conducted a retrospective cohort study using de-identified electronic medical record data from the London Health Sciences Centre, a multi-site academic tertiary care hospital system in Ontario, Canada. The dataset included surgical encounters from January 2012 to July 2025. Ethical approval was obtained from the Research Ethics Board of Western University (London, Ontario, Canada). A data analysis and statistical plan was written and filed with the Research Ethics Board before data were accessed. This study adhered to the STROBE initiative and followed the Transparent Reporting of a Multivariable Prediction Model for Individual Prognosis or Diagnosis (TRIPOD-AI) guidelines [ 22 ]. 2.2. Patient Population The initial cohort included 227,910 surgical encounters across multiple specialties. We included adult patients (≥ 18 years) undergoing non-cardiac ambulatory and inpatient surgery under general anesthesia with documented PACU recovery. Pediatric cases and cardiac surgeries were excluded. Other exclusions applied were: 125 cases with rare demographic categories, 11 with rare surgical locations, 762 with invalid preoperative classifications, 19,242 without recorded PACU recovery duration, 4,232 with implausible durations, and 145 missing anesthesiologist identifiers. The final analytic cohort consisted of 203,393 valid encounters (Fig. 1 ). 2.3. Measurements and Variables Structured variables included demographics such as age, sex, and Body Mass Index (BMI); comorbidities captured by the ASA Physical Status classification; and operative and recovery metrics including operative room duration, PACU delay in minutes, and surgical recovery duration. PACU delay was defined as the time between documented discharge-ready status and physical transfer out of PACU. Postoperative pain assessments included the Average Postoperative Pain Score (AEPS) as well as binary indicators for pain present (yes/no). Scheduling descriptors included case day of week, case day of month, case day of month, case month, and a weekend indicator. Surgical location was represented through categorical variables identifying the operative site. Provider identifiers for both surgeons and anesthesiologists were consolidated and one-hot encoded, yielding 142 surgeons and 134 anesthesiologist variables. Unstructured operative narratives were extracted from the finalized operative documentation associated with each case. Structured procedural codes were analyzed separately from free-text operative narratives. The performed procedure name was treated as a structured categorical variable, while narrative operative documentation was transformed into embeddings. Because these fields closely parallel the scheduled procedure descriptions available preoperatively, the same modeling framework can be adapted for predictive use before surgery. The narratives were transformed into 768-dimensional biomedical Sentence-BERT embeddings for each encounter. Feature engineering encompassed both structured and unstructured data transformations. For structured variables, this included data cleaning, normalization, extraction of time-based fields (day, month, shift indicators), and one-hot encoding of categorical variables such as surgical location and provider identifiers. For unstructured operative narratives, feature engineering referred to converting free-text inputs into numerical representations using the pretrained biomedical Sentence-BERT model. Missing values, primarily in continuous variables such as BMI, were imputed using the Multiple Imputation by Chained Equations (MICE) technique, with XGBoost regressors serving as the predictive model in each iteration. Because this study aimed to benchmark the predictive capacity of multimodal models rather than deploy a preoperative forecasting system, we included all relevant perioperative variables. This allowed assessment of their relative contributions to recovery prediction. Variables such as postoperative pain and anesthetic duration are not available before surgery and would be excluded in future real-time or preoperative implementations. Notably, feature attribution analysis indicated that these features contributed minimally to predictive accuracy. 2.4. Outcomes The primary endpoint was PACU recovery duration, defined as the time from documented PACU arrival to discharge ready time. This endpoint reflects clinical recovery rather than physical departure from PACU, which may be influenced by ward bed availability. The distribution of recovery duration was examined and found to be moderately right-skewed. To evaluate the effect of this skewness, sensitivity analyses were conducted using log-, square-root-, and Box–Cox–transformed targets. Model performance and relative rankings remained stable across all transformations, indicating that distributional normalization did not materially influence results. Secondary endpoints included model performance metrics (Mean Absolute Error (MAE), Mean Squared Error (MSE), R 2 , Mean Absolute Percentage Error (MAPE), and Symmetric MAPE (SMAPE)) as well as comparative accuracy between models at encounter level. The statistical methods used to assess these comparisons are described in the Statistical Analysis section. 2.5. Model Development We evaluated four model families: Linear baselines including linear regression, ridge, and lasso Tree-based methods including decision tree, random forest, and extra trees Gradient-boosting ensembles including XGBoost, LightGBM, and histogram-based boosting Deep learning architectures including feedforward neural networks. Models were trained using nested cross-validation, with inner folds used for hyperparameter optimization via Optuna and outer folds reserved for unbiased performance estimation (Fig. 2 ). In addition to nested cross-validation, we performed temporal validation by training models on encounters from January 2012 through December 2023 and testing cases from January 2024 through July 2025 to approximate prospective performance. The final multimodal dataset included 1,063 predictor dimensions. Of these, 295 were structured features derived from demographic, surgical, timing, and provider variables, and 768 were continuous features generated from Sentence-BERT embeddings of operative narratives. The outcome variable, PACU recovery duration (minutes), was modeled separately and excluded from the predictor set. Tree-based and boosting models handled the full feature space natively, while linear and neural models employed regularization to control overfitting. 2.6. Statistical Analysis Model performance was first summarized using average error measures across cross-validation folds. To make direct comparisons between models, we calculated the difference in prediction error for the same patient encounters. To estimate the uncertainty around these differences, we used bootstrap resampling, which repeatedly re-samples encounters and recalculates results to generate 95% confidence intervals. We chose this method because prediction error distributions in clinical data are often skewed and do not reliably follow a normal distribution, making bootstrap a more robust approach than traditional parametric methods. To test whether one model was consistently more accurate than another, we applied the Wilcoxon signed-rank test. This non-parametric test is well suited for paired data, such as the same encounter predicted by two models, and it avoids the assumption of normally distributed errors. This makes it appropriate for our project, where individual recovery times and prediction errors vary widely. Because we compared multiple models, we used the Holm adjustment to reduce the likelihood of false positive results. All comparisons were performed on the same set of encounters to ensure fairness between models. 3. Results 3.1. Baseline Characteristics Characteristics for the final study population are summarized in Table 1 . The mean PACU recovery time was 143.5 ± 72.50 minutes. Analysis based on time of day showed that PACU recovery duration during day shifts was marginally longer compared to night shifts (143.6 ± 130.0 minutes vs. 142.2 ± 132.0 minutes). SD: standard deviation; BMI: body mass index; ASA: American Society of Anesthesiologists; PACU: post-anesthesia care unit; AEPS: average post-operative pain score; LOS: length of stay. 3.2. Model Performance Predictive performance is summarized in Table 2 in panel A. Linear models (ordinary least squares, ridge, lasso) achieved limited explanatory power ( R 2 = 0.16–0.17) with MSE values of approximately 4,380–4,405. Treebased ensembles improved performance, with random forests ( R 2 = 0.26) and extra trees ( R 2 = 0.21). Boosting models demonstrated the highest accuracy. XGBoost achieved the best overall results ( R 2 = 0.36, MAE = 42.0 minutes, SMAPE = 29.8%), followed closely by histogram-based gradient boosting and LightGBM. Deep learning models yielded mixed results: a feedforward neural network outperformed linear baselines but underperformed boosting models. With over 200,000 encounters, training was computationally intensive, and performance is highly sensitive to hyperparameter tuning. All models were tuned using the Optuna optimization framework under a defined and literature-supported hyperparameter search space. Neural and transformer architectures were optimized across key parameters including layer depth, hidden-unit size, learning rate, dropout, and attention-head configuration. The goal was to ensure fair and reproducible model comparison rather than exhaustive architecture-specific optimization. The reported results therefore represent robust performance within a systematic and transparent tuning protocol. Table 2. Model performance and statistical comparisons (A) Comparative performance of predictive models for PACU recovery duration. Values represent mean ± standard deviation across outer cross-validation folds. Bold values indicate best performance per metric. Model MSE MAE R² sMAPE Linear models Linear 4384.25 ± 53.76 49.55 ± 0.22 0.17 ± 0.00 35.15 ± 0.15 Ridge 4385.96 ± 51.78 49.56 ± 0.21 0.17 ± 0.00 35.15 ± 0.14 Lasso 4405.14 ± 54.58 49.71 ± 0.22 0.16 ± 0.00 35.24 ± 0.12 Tree-based models Decision Tree 4906.55 ± 83.02 52.87 ± 0.44 0.07 ± 0.01 37.54 ± 0.29 Random Forest 3878.28 ± 36.23 46.07 ± 0.18 0.26 ± 0.01 32.71 ± 0.13 Extra Trees 4140.62 ± 57.83 48.04 ± 0.27 0.21 ± 0.00 34.28 ± 0.15 Boosting models Gradient Boosting 3485.25 ± 48.93 43.16 ± 0.25 0.34 ± 0.01 30.70 ± 0.19 Hist. Gradient Boosting 3445.28 ± 37.27 42.83 ± 0.17 0.34 ± 0.01 30.53 ± 0.13 XGBoost 3376.16 ± 35.16 42.04 ± 0.17 0.36 ± 0.00 29.83 ± 0.11 LightGBM 3391.31 ± 27.12 42.32 ± 0.28 0.35 ± 0.01 30.13 ± 0.28 Deep learning model Neural Network 4190.48 ± 356.1 48.04 ± 2.33 0.20 ± 0.06 34.10 ± 1.51 (B) Pairwise comparison against XGBoost on aligned encounters. ∆MAE is the mean difference in absolute error (Other−XGB) in minutes; positive values indicate higher error for the comparator. Bootstrap 95% CIs are shown for the mean difference. Model Other MAE (min) ΔMAE (mean, min) 95% CI (min) Win rate (%) Decision Tree 52.52 10.59 [10.26, 10.92] 62.35 Neural Network 49.76 7.83 [7.54, 8.13] 60.04 Lasso 49.69 7.76 [7.47, 8.04] 60.22 Linear 49.51 7.58 [7.30, 7.88] 60.01 Ridge 49.51 7.58 [7.30, 7.88] 59.99 Extra Trees 48.15 6.22 [5.97, 6.47] 59.44 Random Forest 45.93 4.00 [3.78, 4.22] 57.20 Hist. Gradient Boosting 42.61 0.68 [0.50, 0.85] 52.36 LightGBM 42.21 0.28 [0.12, 0.44] 51.26 Pairwise comparisons were conducted between XGBoost and all other models (Table 2 on panel B). XGBoost consistently outperformed all alternatives, with all adjusted p-values < 0.001 after Holm correction. Relative to LightGBM, the performance difference was minimal (∆MAE = 0.3 minutes; win rate 51.3%), indicating near-equivalent accuracy. Compared with random forests, XGBoost predictions were 4 minutes closer on average (win rate 57.2%). Against decision trees and neural networks, differences were larger (∆MAE 7.8–10.6 minutes). The greatest margin was observed against the TabTransformer (∆MAE = 13.3 minutes; win rate 63.8%). 3.3. Agreement and Clinical Reliability Agreement analysis for XGBoost predictions is shown in Fig. 3 . Panel A illustrates predicted versus actual PACU recovery times with error bands of ± 30 minutes and ± 60 minutes. Panel B summarizes the proportion of excellent ( 60 min) predictions across recovery-time ranges. Predictions were most accurate for recovery times ≤ 180 minutes, where the majority fell within ± 30 minutes of the true value. Accuracy declined for longer recoveries (> 300 minutes), which were frequently underestimated, while very short stays (< 60 minutes) also showed larger relative errors. Across the entire cohort, 40.3% of encounters were predicted within ± 30 minutes of the actual PACU recovery time, and another 29.0% were within ± 60 minutes. Thus, nearly 70% of cases were predicted within one hour, while the remaining 30.7% exceeded a 60-minute margin, representing complex encounters that accounted for most residual error. Panel C shows the overall distribution of error categories, and Panel D displays individual encounters ranked by absolute error. 3.4. Model Explainability and Feature Contributions Feature attribution analysis (Fig. 4 ) showed that narrative clinical context extracted from operative notes contributed the largest share of predictive signal (78.8%). These text embeddings captured detailed information about the procedure performed, patient comorbidities, and intraoperative events that are often not represented in structured data. Provider identifiers together explained 13.4% of variance (surgeon 9.6%; anesthesiologist 3.8%), reflecting differences in case mix and practice patterns. Surgical location (5.5%) also contributed, likely capturing site-specific resources and workflow factors that influence recovery. Postoperative pain assessments, such as average pain scores and pain documentation, accounted for a small share (1.7%). Similarly, anesthetic duration contributed little to predictive accuracy (0.1%). Demographic factors (age, sex, body mass index), ASA classification, anesthesia duration, and scheduling variables (weekday versus weekend) contributed minimally. Overall, these results highlight that unstructured operative documentation provides the richest predictive information, with provider- and system-level variables adding further context, while traditional demographic and timing variables offered limited additional value. A preoperative-only model excluding intraoperative duration and postoperative pain variables achieved an MAE of X minutes compared with 42 minutes for the full multimodal model. Relative performance rankings remained unchanged, suggesting that much of the predictive signal is available before surgery. 4. Discussion This study demonstrates that multimodal machine learning models can predict PACU discharge-ready time with moderate accuracy in a large institutional dataset. The best-performing model achieved a mean absolute error of 42 minutes and explained 36% of the variance in recovery time. Given the inherent variability of postoperative recovery and unmeasured intraoperative factors, this level of performance is consistent with prior perioperative prediction studies. This novel patient-level prediction may support the identification of prolonged recoveries, downstream bed coordination, and quality improvement initiatives targeting extended PACU stays. Boosting-based approaches outperformed linear regression, traditional ensembles, and neural networks. Notably, unstructured narrative features derived from operative reports contributed most of the predictive power. Quantitative feature-attribution analysis showed that operative narratives accounted for 79% of overall model importance; excluding these features reduced R² from 0.36 to 0.22, indicating that narrative documentation captured most of the actionable signal. These narratives encode rich procedural details, intraoperative events, and patient-specific nuances not reflected in structured variables. While operative narratives provided the largest predictive contribution, comparable structured procedure descriptors may allow adaptation to preoperative forecasting frameworks. Further work is required to confirm their real-time applicability. Our findings align with and extend previous work on PACU recovery prediction. Earlier studies identified comorbidities, procedure duration, and intraoperative complications as predictors of prolonged stay [ 11 , 12 , 23 ]. Kim et al. [ 24 ] demonstrated the potential of nonlinear models, while Truong et al. [ 13 ] highlighted the role of structured discharge scoring. More recent research has incorporated predictive analytics, including nomograms [ 11 ], outpatient models [ 10 ], and ensemble or machine learning-based approaches [ 25 – 28 ]. Consistent with studies reporting ensemble superiority, our results suggest that gradient boosting offers the best balance of accuracy and interpretability, while differing from deep learning–focused work likely due to variations in dataset size, tuning, and data modality [ 27 – 28 ]. By highlighting the predictive strength of narrative data, our study expands on earlier efforts that relied solely on structured variables [ 23 , 25 , 26 ]. Feature-importance analysis also provided new clinical insights. Factors traditionally linked with prolonged PACU stay, such as postoperative pain [ 29 – 31 ] and ASA score [ 29 , 31 ], had minimal predictive weight (1.7% and 0.1%, respectively). Anesthetic duration, seldom examined in prior research, also contributed little to model performance. While these features retain explanatory value, they are recorded postoperatively and thus unsuitable for preoperative forecasting. Their limited contribution suggests that models based solely on preoperative variables could achieve comparable accuracy, supporting future efforts toward deployable, real-time prediction systems that inform scheduling and staffing before surgery. Strengths of this study include the use of over 200,000 surgical encounters, multimodal data integration, and rigorous benchmarking through nested cross-validation. Limitations include the single-center design and lack of external validation. In this study, surgeon and anesthesiologist identifiers contributed meaningfully to predictions, likely reflecting practice variation, documentation practices and case mix. This suggests a limitation in the model’s generalizability and the need to recalibrate in external settings. Important contributors such as intraoperative findings, nursing workload, and social determinants were unavailable, and associations derived from feature importance should not be interpreted causally. Future models containing only preoperative features and excluding any intraoperative and postoperative variables could also explore predictions at the time of scheduling. In conclusion, gradient-boosted tree models enriched with operative narrative embeddings demonstrated accurate, interpretable prediction of PACU recovery duration at scale. These findings highlight the predictive value of multimodal electronic health record data, particularly unstructured text, for modeling PACU time. External validation and prospective evaluation are required before operational deployment. Abbreviations PACU – Post Anesthesia Care Unit ASA – American Society of Anesthesiologists LOS – Length of Stay BMI – Body Mass Index EHR – Electronic Health Record ICU – Intensive Care Unit STROBE – Strengthening the Reporting of Observational Studies in Epidemiology TRIPOD-AI – Transparent Reporting of a Multivariable Prediction Model for Individual Prognosis or Diagnosis – Artificial Intelligence extension MAE – Mean Absolute Error MSE – Mean Squared Error R² / R2 – Coefficient of Determination MAPE – Mean Absolute Percentage Error SMAPE – Symmetric Mean Absolute Percentage Error AI – Artificial Intelligence MICE – Multiple Imputation by Chained Equations XGBoost – Extreme Gradient Boosting LightGBM – Light Gradient Boosting Machine Declarations Ethics approval and consent to participate The study adhered to the Declaration of Helsinki. Ethical approval for this study was obtained from the Western University Research Ethics Board (London, Ontario, Canada). Given the retrospective nature of the study and the use of fully de-identified data, the requirement for informed consent to participate was waived by the Research Ethics Board in accordance with applicable institutional policies and national regulations. Consent for publication Not Applicable. Availability of data and materials The de-identified dataset used in this study is not publicly available due to hospital privacy and confidentiality regulations. Analytical code and preprocessing scripts are available from the corresponding author upon reasonable request to support reproducibility. Competing interests The authors declare that they have no competing financial or personal interests that could have influenced the work reported in this manuscript. Funding The Academic Medical Organization of Southwestern Ontario (AMOSO)’s Innovation Fund (Grant Number INN23-015) and Canada Research Chairs Program [CRC-2022-00078] was used for support of this study. Author’s contributions Study conception and design: Noorchenarboo, Kwong, Grolinger, Elnahas Acquisition of data: Noorchenarboo, Elnahas Analysis and interpretation of data: Noorchenarboo, Kwong, Grolinger, Elnahas Drafting of manuscript: Noorchenarboo, Elnahas Critical revision: Noorchenarboo, Kwong, Grolinger, Elnahas Acknowledgements During the preparation of this manuscript, the authors used ChatGPT (OpenAI) and Claude (Anthropic) to support text editing (readability and language fluency) and coding tasks (data preprocessing, model development, and visualization). All outputs were reviewed, validated, and edited by the authors, who take full responsibility for the content and integrity of the publication. References Aihua Liu and Yun Shi. Analysis of adverse events in the postanesthesia unit at a tertiary pediatric hospital. Journal of PeriAnesthesia Nursing , 39(5):750–756, October 2024. Olubukola O. Nafiu and Vivian Onyewuche. Association of abdominal obesity in children with perioperative respiratory adverse events. Journal of PeriAnesthesia Nursing , 29(2):84–93, April 2014. Olubukola O Nafiu, Constance C Burke, Wilson T Chimbira, Ray Ackwerh, Paul I Reynolds, and Shobha Malviya. Prevalence of habitual snoring in children and occurrence of peri-operative adverse events. European Journal of Anaesthesiology , 28(5):340–345, May 2011. Anwar Ul Huda and Nasrullah Sheikh. Evaluation of risk factors for prolonged stay in post anaesthesia care unit (pacu) in a tertiary care hospital of saudia arabia. Sri Lankan Journal of Anaesthesiology , 29(1):29–33, April 2021. T.W. Painter and G.L. Ludbrook. Predicting patients at risk of early postoperative adverse. Anaesthesia and Intensive Care , 42(5):649 – 656, 2014. Cited by: 20; All Open Access, Bronze Open Access. Abdulrahman Almansouri, Angela Yang, Anissa Djedid, Ashraf Emara, Javad Nadaf, and Kevin Petrecca. Adverse events within 24 hours after 1070 adult brain tumor surgeries recovered in a neurocritical care unit or a postanesthesia care unit. Neurosurgery Practice , 6(1), 2025. Cited by: 0. Omer Ozbudak, Hulya Dirol, Ilker Onguc, and Hulya Kahraman. Is asa classification useful in risk stratification for ebus-tbna? Turkish Thoracic Journal , 22(5):364–368, October 2021. Margaret R. Flanigan, Sarah G. Bell, Heidi S. Donovan, Jian Zhao, Jennifer M. Holder-Murray, Stephen A. Esper, Gabriella Ficerai-Garland, and Sarah E. Taylor. Variables impacting prolonged post-anesthesia care unit length of stay in gynecologic cancer patients in the era of same day minimally invasive hysterectomy. Gynecologic Oncology , 186:211–215, July 2024. Maria Teresita B. Aspi and Evangeline Ko-Villa. The use of determinants of length of stay in the post-anesthesia care unit (pacu) at the philippine general hospital among postoperative patients who underwent elective surgeries to create a predictive model for pacu length of stay. Acta Medica Philippina , 54(5), October 2020. Ahmad Elsharydah, Daren R Walters, Alwin Somasundaram, Trenton D Bryson, Abu Minhajuddin, Rodney A Gabriel, and Gaganpreet K Grewal. A preoperative predictive model for prolonged post-anaesthesia care unit stay after outpatient surgeries. Journal of Perioperative Practice , 30(4):91–96, May 2019. Fuquan Fang, Tiantian Liu, Jun Li, Yanchang Yang, Wenxin Hang, Dandan Yan, Sujuan Ye, Pin Wu, Yuhan Hu, and Zhiyong Hu. A novel nomogram for predicting the prolonged length of stay in post-anesthesia care unit after elective operation. BMC Anesthesiology , 23(1), December 2023. Xi Luo, Yanmei Ying, Lu Yin, and Pan Chang. Analysis of risk factors for hypoxemia in pacu for patients undergoing thoracoscopic lung cancer resection based on logistic regression model. BMC Anesthesiology , 25(1), April 2025. L. Truong, J. L. Moran, and P. Blum. Post anaesthesia care unit discharge: A clinical scoring system versus traditional time-based criteria. Anaesthesia and Intensive Care , 32(1):33–42, February 2004. Stacy N. Carr, Regina M. Reinsvold, Thomas E. Heering, and Virginia C. Muckler. Integrating the stop-bang questionnaire into the preanesthetic assessment at a military hospital. Journal of PeriAnesthesia Nursing , 35(4):368–373, August 2020. Mei Liu and Lei Qi. The related factors and countermeasures of hypothermia in patients during the anesthesia recovery period. American journal of translational research , 13(4):3459, 2021. Cyrus Motamed and Jean Louis Bourgain. Comparison of the time to extubation and length of stay in the pacu after sugammadex and neostigmine use in two types of surgery: A monocentric retrospective analysis. Journal of Clinical Medicine , 10(4):815, February 2021. Andrew Bishara, Catherine Chiu, Elizabeth L. Whitlock, Vanja C. Douglas, Sei Lee, Atul J. Butte, Jacqueline M. Leung, and Anne L. Donovan. Postoperative delirium prediction using machine learning models and preoperative electronic health record data. BMC Anesthesiology , 22(1), January 2022. Tyler J. Loftus, Matthew M. Ruppert, Tezcan Ozrazgat-Baslanti, Jeremy A. Balch, Benjamin Shickel, Die Hu, Philip A. Efron, Patrick J. Tighe, William R. Hogan, Parisa Rashidi, Gilbert R. Upchurch, and Azra Bihorac. Postoperative overtriage to an intensive care unit is associated with low value of care. Annals of Surgery , 277(2):179–185, July 2022. Nicholas J. Douville, Lisa Bastarache, Jing He, Kuan-Han H. Wu, Brett Vanderwerff, Emily Bertucci-Richter, Whitney E. Hornsby, Adam Lewis, Elizabeth S. Jewell, Sachin Kheterpal, Nirav Shah, Michael Mathis, Milo C. Engoren, Christopher B. Douville, Ida Surakka, Cristen Willer, and Miklos D. Kertai. Polygenic score for the prediction of postoperative nausea and vomiting: A retrospective derivation and validation cohort study. Anesthesiology , 142(1):52–71, September 2024. Ma´ıra I. Rudolph, Pauline Y. Ng, Hao Deng, Flora T. Scheffenbichler, Stephanie D. Grabitz, Jonathan P. Wanderer, Timothy T. Houle, and Matthias Eikermann. Comparison of a novel clinical score to estimate the risk of residual neuromuscular block prediction score and the last train-of-four count documented in the electronic anaesthesia record: A retrospective cohort study of electronic data on file. European Journal of Anaesthesiology , 35(11):883 – 892, 2018. Cited by: 12; All Open Access, Bronze Open Access. Philip Chung, Christine T. Fong, Andrew M. Walters, Nima Aghaeepour, Meliha Yetisgen, and Vikas N. O’Reilly-Shah. Large language model capabilities in perioperative risk prediction and prognostication. JAMA Surgery , 159(8):928, August 2024. BMJ , page q902, April 2024. Samuel Rupp, Elena Ahrens, Maira I. Rudolph, Omid Azimaraghi, Maximilian S. Schaefer, Philipp Fassbender, Carina P. Himes, Preeti Anand, Parsa Mirhaji, Richard Smith, Jeffrey Freda, Matthias Eikermann, and Karuna Wongtangman. Development and validation of an instrument to predict prolonged length of stay in the postanesthesia care unit following ambulatory surgery; [mise au point et validation d’un instrument de pr´ediction d’une prolongation de la dur´ee de s´ejour en salle de r´eveil apr`es chirurgie ambulatoire]. Canadian Journal of Anesthesia , 70(12):1939 – 1949, 2023. Cited by: 7. Won Oak Kim, Hae Keum Kil, Jung Wan Kang, and Hong Ro Park. Prediction on lengths of stay in the postanesthesia care unit following general anesthesia: preliminary study of the neural network and logistic regression delling. Journal of Korean Medical Science , 15(1):25, 2000. Jeffrey L. Tully, William Zhong, Sierra Simpson, Brian P. Curran, Alvaro A. Macias, Ruth S. Waterman, and Rodney A. Gabriel. Machine learning prediction models to reduce length of stay at ambulatory surgery centers through case resequencing. Journal of Medical Systems , 47(1), July 2023. Guang-Hong Xie, Jun Shen, Fan Li, Huan-Huan Yan, and Ying Qian. Development and validation of a clinical model for predicting delay in postoperative transfer out of the post-anesthesia care unit: A retrospective cohort study. Journal of Multidisciplinary Healthcare , Volume 17:2535–2550, May 2024. Rodney A. Gabriel, Bhavya Harjai, Sierra Simpson, Nicole Goldhaber, Brian P. Curran, and Ruth S. Waterman. Machine learning-based models predicting outpatient surgery end time and recovery room discharge at an ambulatory surgery center. Anesthesia amp; Analgesia , 135(1):159–169, April 2022. Shahnam Sedigh Maroufi, Maryam Soleimani Movahed, Azar Ejmalian, Maryam Sarkhosh, and Ali Behmanesh. Postanesthesia care unit (pacu) readiness predictions using machine learning: a comparative study of algorithms. BMC Medical Informatics and Decision Making , 25(1), March 2025. Leen Alghamdi, Razan Filfilan, Arwa Alghamidi, Roza Alharbi and Haifaa Kayal. Factors associated with prolonged-stay patients within the post-anesthesia care unit: a cohort retrospective study. Cureus ., 16(5):e60092, May 2024. Michael Ganter, Stephan Blumenthal, Seraina Dubendorfer, Simone Brunnschweiler, Tim Hofer, Richard Klaghofer, Andreas Zollinger and Christoph Hofer. The length of stay in the post-anesthesia care unit correlates with pain intensity, nausea and vomiting on arrival. Periop Med , 3.10, November 2024. Qintong Zhang, Feng Xu, Dongsheng Xuan, Li Huang, Min Shi, Zichuan Yue, Dongxue Luo and Manlin Duan. Risk factors for delayed recovery in postanesthesia care unit after surgery: a large and retrospective cohort study. International Journal of Surgery 109(5):p 1281-1290, May 2023. Additional Declarations No competing interests reported. Cite Share Download PDF Status: Under Review Version 1 posted Reviewers agreed at journal 11 May, 2026 Reviewers agreed at journal 22 Apr, 2026 Reviewers invited by journal 18 Mar, 2026 Editor assigned by journal 25 Feb, 2026 Editor invited by journal 24 Feb, 2026 Submission checks completed at journal 20 Feb, 2026 First submitted to journal 20 Feb, 2026 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-8904195","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":609099824,"identity":"958d9e8c-eaf7-4447-9703-c89dc15353b4","order_by":0,"name":"Mohammad Noorchenarboo","email":"","orcid":"","institution":"Western University","correspondingAuthor":false,"prefix":"","firstName":"Mohammad","middleName":"","lastName":"Noorchenarboo","suffix":""},{"id":609099828,"identity":"06dc7bc2-afc7-40e1-873f-1fa4147b0eaa","order_by":1,"name":"Michelle Kwong","email":"","orcid":"","institution":"University of Alberta","correspondingAuthor":false,"prefix":"","firstName":"Michelle","middleName":"","lastName":"Kwong","suffix":""},{"id":609099831,"identity":"7cdcf31d-de94-407a-aa19-eb5b87f7a5d8","order_by":2,"name":"Katarina Grolinger","email":"","orcid":"","institution":"Western University","correspondingAuthor":false,"prefix":"","firstName":"Katarina","middleName":"","lastName":"Grolinger","suffix":""},{"id":609099834,"identity":"1b5a92d2-8a9b-4a5b-8ce2-42cc096c7e3a","order_by":3,"name":"Ahmad Elnahas","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAABAUlEQVRIiWNgGAWjYDACCSDmYWCQAZMMFRYJRGuBIIYzEqRqYWwjQgv/7O7ED28YbHjM2c8e/PhznkSefPsBxscFDHbyOC25c3az5ByGNB7Lnrxkad5tEsUGZxKYjWcwJBs24NJzI3eDNA/DYR6DAzkG0ozbJBI3SDCwAUUOMOLSIn8jd/NvHob/PAbn3xj//DlHInH+DIgWe1xaDG7kbgMp4DG4kWMmwdsgkdhwA6IlEZcWQ6AWyzkGyUAtb8yseY4BHXYmsdmYxyA5GZcWOaDDbrypsJMzOJ9jfPNHjU3i/PbDBx/zVNjZ4vQ+xHkoPJDHDbArHAWjYBSMglFAHAAA2c9RWMTQSxYAAAAASUVORK5CYII=","orcid":"","institution":"Western University","correspondingAuthor":true,"prefix":"","firstName":"Ahmad","middleName":"","lastName":"Elnahas","suffix":""}],"badges":[],"createdAt":"2026-02-17 20:38:28","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-8904195/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8904195/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":105727934,"identity":"5118e6fb-7a44-4c6f-aa01-ea3288edcded","added_by":"auto","created_at":"2026-03-30 11:05:56","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":140277,"visible":true,"origin":"","legend":"\u003cp\u003eCONSORT-style cohort selection flowchart showing exclusions during data cleaning and the final analytic cohort. Initial cohort: \u003cem\u003eN\u003c/em\u003e=227\u003cem\u003e,\u003c/em\u003e910; excluded: \u003cem\u003eN\u003c/em\u003e=24\u003cem\u003e,\u003c/em\u003e517; final analytic cohort: \u003cem\u003eN\u003c/em\u003e=203\u003cem\u003e,\u003c/em\u003e393.\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-8904195/v1/00b277b49f7ccd42c9b0979a.png"},{"id":105282597,"identity":"52203123-5800-4d38-a3c5-d346869e5dbd","added_by":"auto","created_at":"2026-03-24 10:28:35","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":200242,"visible":true,"origin":"","legend":"\u003cp\u003eOverview of the methodological pipeline: preprocessing, model training, and evaluation.\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-8904195/v1/00710ad5392e06ffb76e4722.png"},{"id":105282600,"identity":"db085778-c86a-4685-99cc-a5b602f79a59","added_by":"auto","created_at":"2026-03-24 10:28:35","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":254880,"visible":true,"origin":"","legend":"\u003cp\u003eAgreement and clinical reliability of XGBoost predictions for PACU recovery times.\u003c/p\u003e","description":"","filename":"3.png","url":"https://assets-eu.researchsquare.com/files/rs-8904195/v1/43c116b2c85e322941b07920.png"},{"id":105728305,"identity":"962d97f9-b097-44b4-a90a-e5a14d15073f","added_by":"auto","created_at":"2026-03-30 11:11:21","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":17885,"visible":true,"origin":"","legend":"\u003cp\u003eRelative contribution of feature families to PACU recovery prediction\u003c/p\u003e","description":"","filename":"4.png","url":"https://assets-eu.researchsquare.com/files/rs-8904195/v1/9e280e901c54c405bd877af2.png"},{"id":105731396,"identity":"7593e3fd-a9d8-45d0-bfdb-5875b969dfe0","added_by":"auto","created_at":"2026-03-30 11:30:44","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1393104,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8904195/v1/07233ea4-c751-4cae-be8f-3ac5f23415a2.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Multi-Modal Machine Learning for Prediction of Post-Anesthesia Care Unit Recovery Time","fulltext":[{"header":"1. Background","content":"\u003cp\u003eThe Post Anesthesia Care Unit (PACU) is among the most resource-intensive areas of the hospital, where respiratory, cardiovascular, and neurologic complications frequently arise and may prolong recovery [\u003cspan additionalcitationids=\"CR2\" citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e]. Extended PACU stays are influenced by patient and procedure-related factors such as comorbidities and perioperative conditions [\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e, \u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e], and they contribute to downstream bottlenecks in operating room throughput, bed availability, and overall hospital efficiency [\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e]. As surgical volumes increase and healthcare systems face capacity pressures, anticipating PACU recovery duration has become critical for both patient safety and operational performance. PACU staffing is typically determined by anticipated census rather than individual discharge times. Therefore, the clinical value of individual-level prediction lies primarily in identifying prolonged recoveries and supporting coordination rather than replacing established staffing models.\u003c/p\u003e \u003cp\u003eAccurate prediction of PACU recovery time has direct implications for perioperative efficiency. Although PACU staffing in many institutions is based on high-percentile hourly demand forecasting rather than individual recovery prediction, patient-level estimates may support care coordination, identification of prolonged recoveries, and downstream bed planning. Current prediction methods remain limited, such as the American Society of Anesthesiologists (ASA) Physical Status classification, which offers only coarse risk stratification and shows inconsistent associations with PACU complications or length of stay (LOS) [\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e, \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e]. Regression-based models using factors such as operative duration, comorbidities, or anesthetic technique demonstrate modest accuracy [\u003cspan additionalcitationids=\"CR10 CR11\" citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e], and discharge scoring systems fail to account for individual variability in recovery [\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eRecent advances in perioperative Electronic Health Records (EHRs) and machine learning provide opportunities to overcome these limitations. Predictive models based on artificial intelligence have already improved forecasting of outcomes such as delirium, ICU admission, and unplanned readmission [\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e, \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e]. Large-scale efforts combining structured perioperative data with polygenic risk scores or AI frameworks have further enhanced prediction of complications like postoperative nausea, vomiting, or residual neuromuscular block [\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e, \u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e]. Yet, few studies have directly modeled PACU recovery duration, and applications of language models to this task remain limited [\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e]. The objective of this study is to develop and benchmark multimodal machine learning models for predicting PACU recovery duration using a decade of institutional surgical data. We hypothesized that integrating structured perioperative variables with unstructured perioperative narratives, transformed into biomedical language model embeddings, would yield more accurate and interpretable predictions of PACU recovery time than conventional regression approaches.\u003c/p\u003e"},{"header":"2. Methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003e2.1. Study Design and Ethical Approval\u003c/h2\u003e \u003cp\u003e We conducted a retrospective cohort study using de-identified electronic medical record data from the London Health Sciences Centre, a multi-site academic tertiary care hospital system in Ontario, Canada. The dataset included surgical encounters from January 2012 to July 2025. Ethical approval was obtained from the Research Ethics Board of Western University (London, Ontario, Canada). A data analysis and statistical plan was written and filed with the Research Ethics Board before data were accessed. This study adhered to the STROBE initiative and followed the Transparent Reporting of a Multivariable Prediction Model for Individual Prognosis or Diagnosis (TRIPOD-AI) guidelines [\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e].\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec4\" class=\"Section2\"\u003e \u003ch2\u003e2.2. Patient Population\u003c/h2\u003e \u003cp\u003eThe initial cohort included 227,910 surgical encounters across multiple specialties. We included adult patients (\u0026ge;\u0026thinsp;18 years) undergoing non-cardiac ambulatory and inpatient surgery under general anesthesia with documented PACU recovery.\u003c/p\u003e \u003cp\u003ePediatric cases and cardiac surgeries were excluded. Other exclusions applied were: 125 cases with rare demographic categories, 11 with rare surgical locations, 762 with invalid preoperative classifications, 19,242 without recorded PACU recovery duration, 4,232 with implausible durations, and 145 missing anesthesiologist identifiers. The final analytic cohort consisted of 203,393 valid encounters (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec5\" class=\"Section2\"\u003e \u003ch2\u003e2.3. Measurements and Variables\u003c/h2\u003e \u003cp\u003eStructured variables included demographics such as age, sex, and Body Mass Index (BMI); comorbidities captured by the ASA Physical Status classification; and operative and recovery metrics including operative room duration, PACU delay in minutes, and surgical recovery duration. PACU delay was defined as the time between documented discharge-ready status and physical transfer out of PACU. Postoperative pain assessments included the Average Postoperative Pain Score (AEPS) as well as binary indicators for pain present (yes/no). Scheduling descriptors included case day of week, case day of month, case day of month, case month, and a weekend indicator. Surgical location was represented through categorical variables identifying the operative site. Provider identifiers for both surgeons and anesthesiologists were consolidated and one-hot encoded, yielding 142 surgeons and 134 anesthesiologist variables.\u003c/p\u003e \u003cp\u003eUnstructured operative narratives were extracted from the finalized operative documentation associated with each case. Structured procedural codes were analyzed separately from free-text operative narratives. The performed procedure name was treated as a structured categorical variable, while narrative operative documentation was transformed into embeddings. Because these fields closely parallel the scheduled procedure descriptions available preoperatively, the same modeling framework can be adapted for predictive use before surgery. The narratives were transformed into 768-dimensional biomedical Sentence-BERT embeddings for each encounter. Feature engineering encompassed both structured and unstructured data transformations. For structured variables, this included data cleaning, normalization, extraction of time-based fields (day, month, shift indicators), and one-hot encoding of categorical variables such as surgical location and provider identifiers. For unstructured operative narratives, feature engineering referred to converting free-text inputs into numerical representations using the pretrained biomedical Sentence-BERT model.\u003c/p\u003e \u003cp\u003eMissing values, primarily in continuous variables such as BMI, were imputed using the Multiple Imputation by Chained Equations (MICE) technique, with XGBoost regressors serving as the predictive model in each iteration. Because this study aimed to benchmark the predictive capacity of multimodal models rather than deploy a preoperative forecasting system, we included all relevant perioperative variables. This allowed assessment of their relative contributions to recovery prediction. Variables such as postoperative pain and anesthetic duration are not available before surgery and would be excluded in future real-time or preoperative implementations. Notably, feature attribution analysis indicated that these features contributed minimally to predictive accuracy.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec6\" class=\"Section2\"\u003e \u003ch2\u003e2.4. Outcomes\u003c/h2\u003e \u003cp\u003eThe primary endpoint was PACU recovery duration, defined as the time from documented PACU arrival to discharge ready time. This endpoint reflects clinical recovery rather than physical departure from PACU, which may be influenced by ward bed availability. The distribution of recovery duration was examined and found to be moderately right-skewed. To evaluate the effect of this skewness, sensitivity analyses were conducted using log-, square-root-, and Box\u0026ndash;Cox\u0026ndash;transformed targets. Model performance and relative rankings remained stable across all transformations, indicating that distributional normalization did not materially influence results. Secondary endpoints included model performance metrics (Mean Absolute Error (MAE), Mean Squared Error (MSE), \u003cem\u003eR\u003c/em\u003e\u003csup\u003e2\u003c/sup\u003e, Mean Absolute Percentage Error (MAPE), and Symmetric MAPE (SMAPE)) as well as comparative accuracy between models at encounter level. The statistical methods used to assess these comparisons are described in the Statistical Analysis section.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec7\" class=\"Section2\"\u003e \u003ch2\u003e2.5. Model Development\u003c/h2\u003e \u003cp\u003eWe evaluated four model families:\u003c/p\u003e \u003cp\u003e \u003col\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eLinear baselines including linear regression, ridge, and lasso\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eTree-based methods including decision tree, random forest, and extra trees\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eGradient-boosting ensembles including XGBoost, LightGBM, and histogram-based boosting\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eDeep learning architectures including feedforward neural networks.\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003c/ol\u003e \u003c/p\u003e \u003cp\u003eModels were trained using nested cross-validation, with inner folds used for hyperparameter optimization via Optuna and outer folds reserved for unbiased performance estimation (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e). In addition to nested cross-validation, we performed temporal validation by training models on encounters from January 2012 through December 2023 and testing cases from January 2024 through July 2025 to approximate prospective performance. The final multimodal dataset included 1,063 predictor dimensions. Of these, 295 were structured features derived from demographic, surgical, timing, and provider variables, and 768 were continuous features generated from Sentence-BERT embeddings of operative narratives. The outcome variable, PACU recovery duration (minutes), was modeled separately and excluded from the predictor set. Tree-based and boosting models handled the full feature space natively, while linear and neural models employed regularization to control overfitting.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003e2.6. Statistical Analysis\u003c/h2\u003e \u003cp\u003eModel performance was first summarized using average error measures across cross-validation folds. To make direct comparisons between models, we calculated the difference in prediction error for the same patient encounters. To estimate the uncertainty around these differences, we used bootstrap resampling, which repeatedly re-samples encounters and recalculates results to generate 95% confidence intervals. We chose this method because prediction error distributions in clinical data are often skewed and do not reliably follow a normal distribution, making bootstrap a more robust approach than traditional parametric methods.\u003c/p\u003e \u003cp\u003eTo test whether one model was consistently more accurate than another, we applied the Wilcoxon signed-rank test. This non-parametric test is well suited for paired data, such as the same encounter predicted by two models, and it avoids the assumption of normally distributed errors. This makes it appropriate for our project, where individual recovery times and prediction errors vary widely. Because we compared multiple models, we used the Holm adjustment to reduce the likelihood of false positive results. All comparisons were performed on the same set of encounters to ensure fairness between models.\u003c/p\u003e \u003c/div\u003e"},{"header":"3. Results","content":"\u003cdiv id=\"Sec10\" class=\"Section2\"\u003e\n \u003ch2\u003e3.1. Baseline Characteristics\u003c/h2\u003e\n \u003cp\u003eCharacteristics for the final study population are summarized in Table \u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e. The mean PACU recovery time was 143.5\u0026thinsp;\u0026plusmn;\u0026thinsp;72.50 minutes. Analysis based on time of day showed that PACU recovery duration during day shifts was marginally longer compared to night shifts (143.6\u0026thinsp;\u0026plusmn;\u0026thinsp;130.0 minutes vs. 142.2\u0026thinsp;\u0026plusmn;\u0026thinsp;132.0 minutes).\u003c/p\u003e\n \u003cp\u003e\u003cimg src=\"data:image/png;base64,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\" width=\"606\" height=\"512\"\u003e\u003c/p\u003e\n\u003c/div\u003e\n\u003cp\u003eSD: standard deviation; BMI: body mass index; ASA: American Society of Anesthesiologists; PACU: post-anesthesia care unit; AEPS: average post-operative pain score; LOS: length of stay.\u003c/p\u003e\n\u003cdiv id=\"Sec11\" class=\"Section2\"\u003e\n \u003ch2\u003e3.2. Model Performance\u003c/h2\u003e\n \u003cp\u003ePredictive performance is summarized in Table \u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e in panel A. Linear models (ordinary least squares, ridge, lasso) achieved limited explanatory power (\u003cem\u003eR\u003c/em\u003e\u003csup\u003e2\u003c/sup\u003e\u0026thinsp;=\u0026thinsp;0.16\u0026ndash;0.17) with MSE values of approximately 4,380\u0026ndash;4,405. Treebased ensembles improved performance, with random forests (\u003cem\u003eR\u003c/em\u003e\u003csup\u003e2\u003c/sup\u003e\u0026thinsp;=\u0026thinsp;0.26) and extra trees (\u003cem\u003eR\u003c/em\u003e\u003csup\u003e2\u003c/sup\u003e\u0026thinsp;=\u0026thinsp;0.21). Boosting models demonstrated the highest accuracy. XGBoost achieved the best overall results (\u003cem\u003eR\u003c/em\u003e\u003csup\u003e2\u003c/sup\u003e\u0026thinsp;=\u0026thinsp;0.36, MAE\u0026thinsp;=\u0026thinsp;42.0 minutes, SMAPE\u0026thinsp;=\u0026thinsp;29.8%), followed closely by histogram-based gradient boosting and LightGBM. Deep learning models yielded mixed results: a feedforward neural network outperformed linear baselines but underperformed boosting models. With over 200,000 encounters, training was computationally intensive, and performance is highly sensitive to hyperparameter tuning. All models were tuned using the Optuna optimization framework under a defined and literature-supported hyperparameter search space. Neural and transformer architectures were optimized across key parameters including layer depth, hidden-unit size, learning rate, dropout, and attention-head configuration. The goal was to ensure fair and reproducible model comparison rather than exhaustive architecture-specific optimization. The reported results therefore represent robust performance within a systematic and transparent tuning protocol.\u003c/p\u003e\n\u003cp\u003eTable 2. Model performance and statistical comparisons\u003c/p\u003e\n\u003cp\u003e(A) Comparative performance of predictive models for PACU recovery duration. Values represent mean \u0026plusmn; standard deviation across outer cross-validation folds. Bold values indicate best performance per metric.\u003c/p\u003e\n\u003ctable border=\"0\" cellspacing=\"0\" cellpadding=\"0\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eModel\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eMSE\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eMAE\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eR\u0026sup2;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003esMAPE\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eLinear models\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eLinear\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e4384.25 \u0026plusmn; 53.76\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e49.55 \u0026plusmn; 0.22\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.17 \u0026plusmn; 0.00\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e35.15 \u0026plusmn; 0.15\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eRidge\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e4385.96 \u0026plusmn; 51.78\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e49.56 \u0026plusmn; 0.21\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.17 \u0026plusmn; 0.00\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e35.15 \u0026plusmn; 0.14\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eLasso\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e4405.14 \u0026plusmn; 54.58\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e49.71 \u0026plusmn; 0.22\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.16 \u0026plusmn; 0.00\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e35.24 \u0026plusmn; 0.12\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eTree-based models\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eDecision Tree\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e4906.55 \u0026plusmn; 83.02\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e52.87 \u0026plusmn; 0.44\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.07 \u0026plusmn; 0.01\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e37.54 \u0026plusmn; 0.29\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eRandom Forest\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e3878.28 \u0026plusmn; 36.23\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e46.07 \u0026plusmn; 0.18\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.26 \u0026plusmn; 0.01\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e32.71 \u0026plusmn; 0.13\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eExtra Trees\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e4140.62 \u0026plusmn; 57.83\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e48.04 \u0026plusmn; 0.27\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.21 \u0026plusmn; 0.00\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e34.28 \u0026plusmn; 0.15\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eBoosting models\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eGradient Boosting\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e3485.25 \u0026plusmn; 48.93\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e43.16 \u0026plusmn; 0.25\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.34 \u0026plusmn; 0.01\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e30.70 \u0026plusmn; 0.19\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eHist. Gradient Boosting\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e3445.28 \u0026plusmn; 37.27\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e42.83 \u0026plusmn; 0.17\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.34 \u0026plusmn; 0.01\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e30.53 \u0026plusmn; 0.13\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eXGBoost\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e3376.16 \u0026plusmn; 35.16\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e42.04 \u0026plusmn; 0.17\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.36 \u0026plusmn; 0.00\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e29.83 \u0026plusmn; 0.11\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eLightGBM\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e3391.31 \u0026plusmn; 27.12\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e42.32 \u0026plusmn; 0.28\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.35 \u0026plusmn; 0.01\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e30.13 \u0026plusmn; 0.28\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cem\u003eDeep learning model\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eNeural Network\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e4190.48 \u0026plusmn; 356.1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e48.04 \u0026plusmn; 2.33\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e0.20 \u0026plusmn; 0.06\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e34.10 \u0026plusmn; 1.51\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e(B) Pairwise comparison against XGBoost on aligned encounters. ∆MAE is the mean difference in absolute error (Other\u0026minus;XGB) in minutes; positive values indicate higher error for the comparator. Bootstrap 95% CIs are shown for the mean difference.\u003c/p\u003e\n\u003ctable border=\"0\" cellspacing=\"0\" cellpadding=\"0\" class=\"fr-table-selection-hover\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eModel\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eOther MAE (min)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026Delta;MAE (mean, min)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003e95% CI (min)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eWin rate (%)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eDecision Tree\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e52.52\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e10.59\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e[10.26, 10.92]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e62.35\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eNeural Network\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e49.76\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e7.83\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e[7.54, 8.13]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e60.04\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eLasso\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e49.69\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e7.76\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e[7.47, 8.04]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e60.22\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eLinear\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e49.51\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e7.58\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e[7.30, 7.88]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e60.01\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eRidge\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e49.51\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e7.58\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e[7.30, 7.88]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e59.99\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eExtra Trees\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e48.15\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e6.22\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e[5.97, 6.47]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e59.44\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eRandom Forest\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e45.93\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e4.00\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e[3.78, 4.22]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e57.20\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eHist. Gradient Boosting\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e42.61\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.68\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e[0.50, 0.85]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e52.36\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eLightGBM\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e42.21\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e0.28\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e[0.12, 0.44]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e51.26\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n \u003cp\u003ePairwise comparisons were conducted between XGBoost and all other models (Table \u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e on panel B). XGBoost consistently outperformed all alternatives, with all adjusted p-values\u0026thinsp;\u0026lt;\u0026thinsp;0.001 after Holm correction. Relative to LightGBM, the performance difference was minimal (∆MAE\u0026thinsp;=\u0026thinsp;0.3 minutes; win rate 51.3%), indicating near-equivalent accuracy. Compared with random forests, XGBoost predictions were 4 minutes closer on average (win rate 57.2%). Against decision trees and neural networks, differences were larger (∆MAE 7.8\u0026ndash;10.6 minutes). The greatest margin was observed against the TabTransformer (∆MAE\u0026thinsp;=\u0026thinsp;13.3 minutes; win rate 63.8%).\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec12\" class=\"Section2\"\u003e\n \u003ch2\u003e3.3. Agreement and Clinical Reliability\u003c/h2\u003e\n \u003cp\u003eAgreement analysis for XGBoost predictions is shown in Fig. \u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e. Panel A illustrates predicted versus actual PACU recovery times with error bands of \u0026plusmn;\u0026thinsp;30 minutes and \u0026plusmn;\u0026thinsp;60 minutes. Panel B summarizes the proportion of excellent (\u0026lt;\u0026thinsp;30 min), acceptable (30\u0026ndash;60 min), and poor (\u0026gt;\u0026thinsp;60 min) predictions across recovery-time ranges. Predictions were most accurate for recovery times\u0026thinsp;\u0026le;\u0026thinsp;180 minutes, where the majority fell within \u0026plusmn;\u0026thinsp;30 minutes of the true value. Accuracy declined for longer recoveries (\u0026gt;\u0026thinsp;300 minutes), which were frequently underestimated, while very short stays (\u0026lt;\u0026thinsp;60 minutes) also showed larger relative errors.\u003c/p\u003e\n \u003cp\u003eAcross the entire cohort, 40.3% of encounters were predicted within \u0026plusmn;\u0026thinsp;30 minutes of the actual PACU recovery time, and another 29.0% were within \u0026plusmn;\u0026thinsp;60 minutes. Thus, nearly 70% of cases were predicted within one hour, while the remaining 30.7% exceeded a 60-minute margin, representing complex encounters that accounted for most residual error. Panel C shows the overall distribution of error categories, and Panel D displays individual encounters ranked by absolute error.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec13\" class=\"Section2\"\u003e\n \u003ch2\u003e3.4. Model Explainability and Feature Contributions\u003c/h2\u003e\n \u003cp\u003eFeature attribution analysis (Fig. \u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e) showed that narrative clinical context extracted from operative notes contributed the largest share of predictive signal (78.8%). These text embeddings captured detailed information about the procedure performed, patient comorbidities, and intraoperative events that are often not represented in structured data. Provider identifiers together explained 13.4% of variance (surgeon 9.6%; anesthesiologist 3.8%), reflecting differences in case mix and practice patterns. Surgical location (5.5%) also contributed, likely capturing site-specific resources and workflow factors that influence recovery.\u003c/p\u003e\n \u003cp\u003ePostoperative pain assessments, such as average pain scores and pain documentation, accounted for a small share (1.7%). Similarly, anesthetic duration contributed little to predictive accuracy (0.1%). Demographic factors (age, sex, body mass index), ASA classification, anesthesia duration, and scheduling variables (weekday versus weekend) contributed minimally. Overall, these results highlight that unstructured operative documentation provides the richest predictive information, with provider- and system-level variables adding further context, while traditional demographic and timing variables offered limited additional value.\u003c/p\u003e\n \u003cp\u003eA preoperative-only model excluding intraoperative duration and postoperative pain variables achieved an MAE of X minutes compared with 42 minutes for the full multimodal model. Relative performance rankings remained unchanged, suggesting that much of the predictive signal is available before surgery.\u003c/p\u003e\n\u003c/div\u003e"},{"header":"4. Discussion","content":"\u003cp\u003eThis study demonstrates that multimodal machine learning models can predict PACU discharge-ready time with moderate accuracy in a large institutional dataset. The best-performing model achieved a mean absolute error of 42 minutes and explained 36% of the variance in recovery time. Given the inherent variability of postoperative recovery and unmeasured intraoperative factors, this level of performance is consistent with prior perioperative prediction studies. This novel patient-level prediction may support the identification of prolonged recoveries, downstream bed coordination, and quality improvement initiatives targeting extended PACU stays.\u003c/p\u003e \u003cp\u003eBoosting-based approaches outperformed linear regression, traditional ensembles, and neural networks. Notably, unstructured narrative features derived from operative reports contributed most of the predictive power. Quantitative feature-attribution analysis showed that operative narratives accounted for 79% of overall model importance; excluding these features reduced R\u0026sup2; from 0.36 to 0.22, indicating that narrative documentation captured most of the actionable signal. These narratives encode rich procedural details, intraoperative events, and patient-specific nuances not reflected in structured variables. While operative narratives provided the largest predictive contribution, comparable structured procedure descriptors may allow adaptation to preoperative forecasting frameworks. Further work is required to confirm their real-time applicability.\u003c/p\u003e \u003cp\u003eOur findings align with and extend previous work on PACU recovery prediction. Earlier studies identified comorbidities, procedure duration, and intraoperative complications as predictors of prolonged stay [\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e, \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e, \u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e]. Kim et al. [\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e] demonstrated the potential of nonlinear models, while Truong et al. [\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e] highlighted the role of structured discharge scoring. More recent research has incorporated predictive analytics, including nomograms [\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e], outpatient models [\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e], and ensemble or machine learning-based approaches [\u003cspan additionalcitationids=\"CR26 CR27\" citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e]. Consistent with studies reporting ensemble superiority, our results suggest that gradient boosting offers the best balance of accuracy and interpretability, while differing from deep learning\u0026ndash;focused work likely due to variations in dataset size, tuning, and data modality [\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e]. By highlighting the predictive strength of narrative data, our study expands on earlier efforts that relied solely on structured variables [\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e, \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e, \u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eFeature-importance analysis also provided new clinical insights. Factors traditionally linked with prolonged PACU stay, such as postoperative pain [\u003cspan additionalcitationids=\"CR30\" citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e] and ASA score [\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e, \u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e], had minimal predictive weight (1.7% and 0.1%, respectively). Anesthetic duration, seldom examined in prior research, also contributed little to model performance. While these features retain explanatory value, they are recorded postoperatively and thus unsuitable for preoperative forecasting. Their limited contribution suggests that models based solely on preoperative variables could achieve comparable accuracy, supporting future efforts toward deployable, real-time prediction systems that inform scheduling and staffing before surgery.\u003c/p\u003e \u003cp\u003eStrengths of this study include the use of over 200,000 surgical encounters, multimodal data integration, and rigorous benchmarking through nested cross-validation. Limitations include the single-center design and lack of external validation. In this study, surgeon and anesthesiologist identifiers contributed meaningfully to predictions, likely reflecting practice variation, documentation practices and case mix. This suggests a limitation in the model\u0026rsquo;s generalizability and the need to recalibrate in external settings. Important contributors such as intraoperative findings, nursing workload, and social determinants were unavailable, and associations derived from feature importance should not be interpreted causally. Future models containing only preoperative features and excluding any intraoperative and postoperative variables could also explore predictions at the time of scheduling.\u003c/p\u003e \u003cp\u003eIn conclusion, gradient-boosted tree models enriched with operative narrative embeddings demonstrated accurate, interpretable prediction of PACU recovery duration at scale. These findings highlight the predictive value of multimodal electronic health record data, particularly unstructured text, for modeling PACU time. External validation and prospective evaluation are required before operational deployment.\u003c/p\u003e"},{"header":"Abbreviations","content":"\u003cp\u003ePACU – Post Anesthesia Care Unit\u003c/p\u003e\n\u003cp\u003eASA – American Society of Anesthesiologists\u003c/p\u003e\n\u003cp\u003eLOS – Length of Stay\u003c/p\u003e\n\u003cp\u003eBMI – Body Mass Index\u003c/p\u003e\n\u003cp\u003eEHR – Electronic Health Record\u003c/p\u003e\n\u003cp\u003eICU – Intensive Care Unit\u003c/p\u003e\n\u003cp\u003eSTROBE – Strengthening the Reporting of Observational Studies in Epidemiology\u003c/p\u003e\n\u003cp\u003eTRIPOD-AI – Transparent Reporting of a Multivariable Prediction Model for Individual Prognosis or Diagnosis – Artificial Intelligence extension\u003c/p\u003e\n\u003cp\u003eMAE – Mean Absolute Error\u003c/p\u003e\n\u003cp\u003eMSE – Mean Squared Error\u003c/p\u003e\n\u003cp\u003eR² / R2 – Coefficient of Determination\u003c/p\u003e\n\u003cp\u003eMAPE – Mean Absolute Percentage Error\u003c/p\u003e\n\u003cp\u003eSMAPE – Symmetric Mean Absolute Percentage Error\u003c/p\u003e\n\u003cp\u003eAI – Artificial Intelligence\u003c/p\u003e\n\u003cp\u003eMICE – Multiple Imputation by Chained Equations\u003c/p\u003e\n\u003cp\u003eXGBoost – Extreme Gradient Boosting\u003c/p\u003e\n\u003cp\u003eLightGBM – Light Gradient Boosting Machine\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eEthics approval and consent to participate\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe study adhered to the Declaration of Helsinki.\u0026nbsp;Ethical approval for this study was obtained from the Western University Research Ethics Board (London, Ontario, Canada). Given the retrospective nature of the study and the use of fully de-identified data, the requirement for informed consent to participate was waived by the Research Ethics Board in accordance with applicable institutional policies and national regulations.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent for publication\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot Applicable.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAvailability of data and materials\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe de-identified dataset used in this study is not publicly available due to hospital privacy and confidentiality regulations. Analytical code and preprocessing scripts are available from the corresponding author upon reasonable request to support reproducibility.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting interests\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors declare that they have no competing financial or personal interests that could have influenced the work reported in this manuscript.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe Academic Medical Organization of Southwestern Ontario (AMOSO)\u0026rsquo;s Innovation Fund (Grant Number INN23-015) and Canada Research Chairs Program [CRC-2022-00078] was used for support of this study.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthor\u0026rsquo;s contributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eStudy conception and design: Noorchenarboo, Kwong, Grolinger, Elnahas\u003c/p\u003e\n\u003cp\u003eAcquisition of data: Noorchenarboo, Elnahas\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eAnalysis and interpretation of data: Noorchenarboo, Kwong, Grolinger, Elnahas\u003c/p\u003e\n\u003cp\u003eDrafting of manuscript: Noorchenarboo, Elnahas\u003c/p\u003e\n\u003cp\u003eCritical revision: Noorchenarboo, Kwong, Grolinger, Elnahas\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAcknowledgements\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eDuring the preparation of this manuscript, the authors used ChatGPT (OpenAI) and Claude (Anthropic) to support text editing (readability and language fluency) and coding tasks (data preprocessing, model development, and visualization). All outputs were reviewed, validated, and edited by the authors, who take full responsibility for the content and integrity of the publication.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eAihua Liu and Yun Shi. Analysis of adverse events in the postanesthesia unit at a tertiary pediatric hospital. \u003cem\u003eJournal of PeriAnesthesia Nursing\u003c/em\u003e, 39(5):750\u0026ndash;756, October 2024.\u003c/li\u003e\n\u003cli\u003eOlubukola O. Nafiu and Vivian Onyewuche. Association of abdominal obesity in children with perioperative respiratory adverse events. \u003cem\u003eJournal of PeriAnesthesia Nursing\u003c/em\u003e, 29(2):84\u0026ndash;93, April 2014.\u003c/li\u003e\n\u003cli\u003eOlubukola O Nafiu, Constance C Burke, Wilson T Chimbira, Ray Ackwerh, Paul I Reynolds, and Shobha Malviya. Prevalence of habitual snoring in children and occurrence of peri-operative adverse events. \u003cem\u003eEuropean Journal of Anaesthesiology\u003c/em\u003e, 28(5):340\u0026ndash;345, May 2011.\u003c/li\u003e\n\u003cli\u003eAnwar Ul Huda and Nasrullah Sheikh. Evaluation of risk factors for prolonged stay in post anaesthesia care unit (pacu) in a tertiary care hospital of saudia arabia. \u003cem\u003eSri Lankan Journal of Anaesthesiology\u003c/em\u003e, 29(1):29\u0026ndash;33, April 2021.\u003c/li\u003e\n\u003cli\u003eT.W. Painter and G.L. Ludbrook. Predicting patients at risk of early postoperative adverse. \u003cem\u003eAnaesthesia and Intensive Care\u003c/em\u003e, 42(5):649 \u0026ndash; 656, 2014. Cited by: 20; All Open Access, Bronze Open Access.\u003c/li\u003e\n\u003cli\u003eAbdulrahman Almansouri, Angela Yang, Anissa Djedid, Ashraf Emara, Javad Nadaf, and Kevin Petrecca. Adverse events within 24 hours after 1070 adult brain tumor surgeries recovered in a neurocritical care unit or a postanesthesia care unit. \u003cem\u003eNeurosurgery Practice\u003c/em\u003e, 6(1), 2025. Cited by: 0.\u003c/li\u003e\n\u003cli\u003eOmer Ozbudak, Hulya Dirol, Ilker Onguc, and Hulya Kahraman. Is asa classification useful in risk stratification for ebus-tbna? \u003cem\u003eTurkish Thoracic Journal\u003c/em\u003e, 22(5):364\u0026ndash;368, October 2021.\u003c/li\u003e\n\u003cli\u003eMargaret R. Flanigan, Sarah G. Bell, Heidi S. Donovan, Jian Zhao, Jennifer M. Holder-Murray, Stephen A. Esper, Gabriella Ficerai-Garland, and Sarah E. Taylor. Variables impacting prolonged post-anesthesia care unit length of stay in gynecologic cancer patients in the era of same day minimally invasive hysterectomy. \u003cem\u003eGynecologic Oncology\u003c/em\u003e, 186:211\u0026ndash;215, July 2024.\u003c/li\u003e\n\u003cli\u003eMaria Teresita B. Aspi and Evangeline Ko-Villa. The use of determinants of length of stay in the post-anesthesia care unit (pacu) at the philippine general hospital among postoperative patients who underwent elective surgeries to create a predictive model for pacu length of stay. \u003cem\u003eActa Medica Philippina\u003c/em\u003e, 54(5), October 2020.\u003c/li\u003e\n\u003cli\u003eAhmad Elsharydah, Daren R Walters, Alwin Somasundaram, Trenton D Bryson, Abu Minhajuddin, Rodney A Gabriel, and Gaganpreet K Grewal. A preoperative predictive model for prolonged post-anaesthesia care unit stay after outpatient surgeries. \u003cem\u003eJournal of Perioperative Practice\u003c/em\u003e, 30(4):91\u0026ndash;96, May 2019.\u003c/li\u003e\n\u003cli\u003eFuquan Fang, Tiantian Liu, Jun Li, Yanchang Yang, Wenxin Hang, Dandan Yan, Sujuan Ye, Pin Wu, Yuhan Hu, and Zhiyong Hu. A novel nomogram for predicting the prolonged length of stay in post-anesthesia care unit after elective operation. \u003cem\u003eBMC Anesthesiology\u003c/em\u003e, 23(1), December 2023.\u003c/li\u003e\n\u003cli\u003eXi Luo, Yanmei Ying, Lu Yin, and Pan Chang. Analysis of risk factors for hypoxemia in pacu for patients undergoing thoracoscopic lung cancer resection based on logistic regression model. \u003cem\u003eBMC Anesthesiology\u003c/em\u003e, 25(1), April 2025.\u003c/li\u003e\n\u003cli\u003eL. Truong, J. L. Moran, and P. Blum. Post anaesthesia care unit discharge: A clinical scoring system versus traditional time-based criteria. \u003cem\u003eAnaesthesia and Intensive Care\u003c/em\u003e, 32(1):33\u0026ndash;42, February 2004.\u003c/li\u003e\n\u003cli\u003eStacy N. Carr, Regina M. Reinsvold, Thomas E. Heering, and Virginia C. Muckler. Integrating the stop-bang questionnaire into the preanesthetic assessment at a military hospital. \u003cem\u003eJournal of PeriAnesthesia Nursing\u003c/em\u003e, 35(4):368\u0026ndash;373, August 2020.\u003c/li\u003e\n\u003cli\u003eMei Liu and Lei Qi. The related factors and countermeasures of hypothermia in patients during the anesthesia recovery period. \u003cem\u003eAmerican journal of translational research\u003c/em\u003e, 13(4):3459, 2021.\u003c/li\u003e\n\u003cli\u003eCyrus Motamed and Jean Louis Bourgain. Comparison of the time to extubation and length of stay in the pacu after sugammadex and neostigmine use in two types of surgery: A monocentric retrospective analysis. \u003cem\u003eJournal of Clinical Medicine\u003c/em\u003e, 10(4):815, February 2021.\u003c/li\u003e\n\u003cli\u003eAndrew Bishara, Catherine Chiu, Elizabeth L. Whitlock, Vanja C. Douglas, Sei Lee, Atul J. Butte, Jacqueline M. Leung, and Anne L. Donovan. Postoperative delirium prediction using machine learning models and preoperative electronic health record data. \u003cem\u003eBMC Anesthesiology\u003c/em\u003e, 22(1), January 2022.\u003c/li\u003e\n\u003cli\u003eTyler J. Loftus, Matthew M. Ruppert, Tezcan Ozrazgat-Baslanti, Jeremy A. Balch, Benjamin Shickel, Die Hu, Philip A. Efron, Patrick J. Tighe, William R. Hogan, Parisa Rashidi, Gilbert R. Upchurch, and Azra Bihorac. Postoperative overtriage to an intensive care unit is associated with low value of care. \u003cem\u003eAnnals of Surgery\u003c/em\u003e, 277(2):179\u0026ndash;185, July 2022.\u003c/li\u003e\n\u003cli\u003eNicholas J. Douville, Lisa Bastarache, Jing He, Kuan-Han H. Wu, Brett Vanderwerff, Emily Bertucci-Richter, Whitney E. Hornsby, Adam Lewis, Elizabeth S. Jewell, Sachin Kheterpal, Nirav Shah, Michael Mathis, Milo C. Engoren, Christopher B. Douville, Ida Surakka, Cristen Willer, and Miklos D. Kertai. Polygenic score for the prediction of postoperative nausea and vomiting: A retrospective derivation and validation cohort study. \u003cem\u003eAnesthesiology\u003c/em\u003e, 142(1):52\u0026ndash;71, September 2024.\u003c/li\u003e\n\u003cli\u003eMa\u0026acute;ıra I. Rudolph, Pauline Y. Ng, Hao Deng, Flora T. Scheffenbichler, Stephanie D. Grabitz, Jonathan P. Wanderer, Timothy T. Houle, and Matthias Eikermann. Comparison of a novel clinical score to estimate the risk of residual neuromuscular block prediction score and the last train-of-four count documented in the electronic anaesthesia record: A retrospective cohort study of electronic data on file. \u003cem\u003eEuropean Journal of Anaesthesiology\u003c/em\u003e, 35(11):883 \u0026ndash; 892, 2018. Cited by: 12; All Open Access, Bronze Open Access.\u003c/li\u003e\n\u003cli\u003ePhilip Chung, Christine T. Fong, Andrew M. Walters, Nima Aghaeepour, Meliha Yetisgen, and Vikas N. O\u0026rsquo;Reilly-Shah. Large language model capabilities in perioperative risk prediction and prognostication. \u003cem\u003eJAMA Surgery\u003c/em\u003e, 159(8):928, August 2024.\u003c/li\u003e\n\u003cli\u003e\u003cem\u003eBMJ\u003c/em\u003e, page q902, April 2024.\u003c/li\u003e\n\u003cli\u003eSamuel Rupp, Elena Ahrens, Maira I. Rudolph, Omid Azimaraghi, Maximilian S. Schaefer, Philipp Fassbender, Carina P. Himes, Preeti Anand, Parsa Mirhaji, Richard Smith, Jeffrey Freda, Matthias Eikermann, and Karuna Wongtangman. Development and validation of an instrument to predict prolonged length of stay in the postanesthesia care unit following ambulatory surgery; [mise au point et validation d\u0026rsquo;un instrument de pr\u0026acute;ediction d\u0026rsquo;une prolongation de la dur\u0026acute;ee de s\u0026acute;ejour en salle de r\u0026acute;eveil apr`es chirurgie ambulatoire]. \u003cem\u003eCanadian Journal of Anesthesia\u003c/em\u003e, 70(12):1939 \u0026ndash; 1949, 2023. Cited by: 7.\u003c/li\u003e\n\u003cli\u003eWon Oak Kim, Hae Keum Kil, Jung Wan Kang, and Hong Ro Park. Prediction on lengths of stay in the postanesthesia care unit following general anesthesia: preliminary study of the neural network and logistic regression delling. \u003cem\u003eJournal of Korean Medical Science\u003c/em\u003e, 15(1):25, 2000.\u003c/li\u003e\n\u003cli\u003eJeffrey L. Tully, William Zhong, Sierra Simpson, Brian P. Curran, Alvaro A. Macias, Ruth S. Waterman, and Rodney A. Gabriel. Machine learning prediction models to reduce length of stay at ambulatory surgery centers through case resequencing. \u003cem\u003eJournal of Medical Systems\u003c/em\u003e, 47(1), July 2023.\u003c/li\u003e\n\u003cli\u003eGuang-Hong Xie, Jun Shen, Fan Li, Huan-Huan Yan, and Ying Qian. Development and validation of a clinical model for predicting delay in postoperative transfer out of the post-anesthesia care unit: A retrospective cohort study. \u003cem\u003eJournal of Multidisciplinary Healthcare\u003c/em\u003e, Volume 17:2535\u0026ndash;2550, May 2024.\u003c/li\u003e\n\u003cli\u003eRodney A. Gabriel, Bhavya Harjai, Sierra Simpson, Nicole Goldhaber, Brian P. Curran, and Ruth S. Waterman. Machine learning-based models predicting outpatient surgery end time and recovery room discharge at an ambulatory surgery center. \u003cem\u003eAnesthesia amp; Analgesia\u003c/em\u003e, 135(1):159\u0026ndash;169, April 2022.\u003c/li\u003e\n\u003cli\u003eShahnam Sedigh Maroufi, Maryam Soleimani Movahed, Azar Ejmalian, Maryam Sarkhosh, and Ali Behmanesh. Postanesthesia care unit (pacu) readiness predictions using machine learning: a comparative study of algorithms. \u003cem\u003eBMC Medical Informatics and Decision Making\u003c/em\u003e, 25(1), March 2025.\u003c/li\u003e\n\u003cli\u003eLeen Alghamdi, Razan Filfilan, Arwa Alghamidi, Roza Alharbi and Haifaa Kayal. Factors associated with prolonged-stay patients within the post-anesthesia care unit: a cohort retrospective study. \u003cem\u003eCureus\u003c/em\u003e., 16(5):e60092, May 2024.\u003c/li\u003e\n\u003cli\u003eMichael Ganter, Stephan Blumenthal, Seraina Dubendorfer, Simone Brunnschweiler, Tim Hofer, Richard Klaghofer, Andreas Zollinger and Christoph Hofer. The length of stay in the post-anesthesia care unit correlates with pain intensity, nausea and vomiting on arrival. \u003cem\u003ePeriop Med\u003c/em\u003e, 3.10, November 2024.\u003c/li\u003e\n\u003cli\u003eQintong Zhang, Feng Xu, Dongsheng Xuan, Li Huang, Min Shi, Zichuan Yue, Dongxue Luo and Manlin Duan. Risk factors for delayed recovery in postanesthesia care unit after surgery: a large and retrospective cohort study. \u003cem\u003eInternational Journal of Surgery\u003c/em\u003e 109(5):p 1281-1290, May 2023.\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"bmc-anesthesiology","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"bane","sideBox":"Learn more about [BMC Anesthesiology](http://bmcanesthesiol.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/bane","title":"BMC Anesthesiology","twitterHandle":"BMC_series","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"PACU recovery time, perioperative prediction, machine learning, gradient boosting, artificial intelligence","lastPublishedDoi":"10.21203/rs.3.rs-8904195/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8904195/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003ch2\u003eBackground\u003c/h2\u003e \u003cp\u003eEfficient recovery management in the Post-Anesthesia Care Unit (PACU) is important for patient safety and care coordination, yet estimating recovery duration remains challenging. Traditional regression models and discharge scoring systems have shown limited accuracy and reliability across patient populations.\u003c/p\u003e\u003ch2\u003eMethods\u003c/h2\u003e \u003cp\u003eWe developed and evaluated machine learning models for PACU recovery prediction using an institution-wide dataset of 203,393 surgical cases collected between January 2012 and July 2025. Structured variables (demographics, surgical metrics, provider identifiers, timing features) were integrated with unstructured operative narratives transformed into biomedical SBERT embeddings, yielding a multi-modal dataset. Model performance was evaluated using robust validation procedures to ensure findings were not dependent on a single patient group or time period.\u003c/p\u003e\u003ch2\u003eResults\u003c/h2\u003e \u003cp\u003eGradient-boosted tree ensembles demonstrated the best performance, with XGBoost achieving a mean absolute error of approximately 42 minutes and explaining over one-third of the variance in recovery time. Nearly 70% of predictions fell within 60 minutes of observed recovery duration. Statistical testing confirmed significant improvements over linear baselines, conventional ensembles, and neural networks. Explainability analyses revealed that operative narratives provided the greatest predictive signal, while provider identifiers and surgical location also contributed, reflecting both patient-level complexity and system-level workflow factors.\u003c/p\u003e\u003ch2\u003eConclusions\u003c/h2\u003e \u003cp\u003eCombining structured perioperative variables with embeddings of operative narratives enables accurate and interpretable PACU recovery prediction at scale. These findings demonstrate the feasibility of multimodal prediction of PACU discharge-ready time and provide a foundation for future evaluation of operational and clinical applications. Future work should include external validation and integration into real-time perioperative systems.\u003c/p\u003e","manuscriptTitle":"Multi-Modal Machine Learning for Prediction of Post-Anesthesia Care Unit Recovery Time","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-03-24 10:28:26","doi":"10.21203/rs.3.rs-8904195/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"reviewerAgreed","content":"190486467538116776770483864987332546643","date":"2026-05-11T14:01:04+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"284434533259252328696032909327333056424","date":"2026-04-22T11:57:16+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2026-03-18T21:40:52+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2026-02-25T10:20:48+00:00","index":"","fulltext":""},{"type":"editorInvited","content":"","date":"2026-02-24T09:58:32+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2026-02-20T21:24:20+00:00","index":"","fulltext":""},{"type":"submitted","content":"BMC Anesthesiology","date":"2026-02-20T21:20:00+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"bmc-anesthesiology","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"bane","sideBox":"Learn more about [BMC Anesthesiology](http://bmcanesthesiol.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/bane","title":"BMC Anesthesiology","twitterHandle":"BMC_series","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"d5f0c158-bbec-4c1a-9cc3-2dbbd76ed13d","owner":[],"postedDate":"March 24th, 2026","published":true,"recentEditorialEvents":[{"type":"reviewerAgreed","content":"190486467538116776770483864987332546643","date":"2026-05-11T14:01:04+00:00","index":51,"fulltext":""}],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[],"tags":[],"updatedAt":"2026-03-24T10:28:26+00:00","versionOfRecord":[],"versionCreatedAt":"2026-03-24 10:28:26","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8904195","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8904195","identity":"rs-8904195","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}