Rodney Allanigue Gabriel

The opioid use disorder (OUD) epidemic in the United States has been a major public health crisis. We developed OUD-REY ( OUD R isk E valuator at 1 Y ear) - a deep learning model that uses longitudinal electronic health record (EHR) data from the All of Us research program to predict risk of OUD at one year for patients with an opioid prescription. On internal validation, OUD-REY achieved an AUROC of 91.2% and AUPRC of 77.0%. We leveraged a pretrained language model to accommodate thousands of unseen, out-of-vocabulary clinical features in a patient’s longitudinal EHR history. We externally validated our model on a new dataset released by All of Us that was temporally separated from our development set and on a dataset containing ~6 million patients from medical centers of the Veterans Affairs (VA) nationwide over a span of 20 years. On external validation, our model had an AUROC and AUPRC of 86.4% and 23.1% (OUD prevalence of 2.9%), respectively, on our temporally separated All of Us validation set. It had an AUROC and AUPRC of 85.5% and 19.5% (OUD prevalence of 3.5%), respectively, on the VA dataset. Our fairness auditing showed that the model was generally fair for patients with different genders, race, and ethnic backgrounds. This model may potentially improve prevention and care of OUD.

Was not provided for this article.

Opioid use disorder (OUD) is a chronic condition characterized by an uncontrollable, problematic pattern of opioid use that continues despite severe negative consequences.1 Opioids are often given within a medical setting in response to moderate-to-severe pain following a surgical procedure and can lead to postoperative persistent opioid use (POU), often defined as continuing to use opioids 3 months postoperatively.2 3 Recent evidence suggests that POU is a significant risk factor for developing downstream OUD.4 5 The objective of this study was to report the timing and frequency of a subsequent documented diagnosis code for moderate-to-severe OUD among surgical patients who were prescribed opioids 90–180 days after major surgery.

Postoperative pain remains a common and clinically important problem, with inadequate control contributing to delayed recovery, prolonged opioid exposure, and risk of chronic postsurgical pain. Machine learning has been suggested as a strategy to improve postoperative pain management by predicting pain trajectories, estimating opioid requirements, identifying patients at risk for poor outcomes and improving the clinical management. This critical review examines current applications of machine learning in postoperative pain, including preoperative risk stratification, intraoperative and postoperative prediction, objective assessment, wearable monitoring, discharge opioid prescribing, and use in specialized populations.We found that although many models demonstrate moderate to strong predictive performance, their clinical utility remains uncertain. Most studies are retrospective, single-center, or procedure-specific, with limited external validation and heterogeneous outcome definitions. While many models incorporate postoperative variables, this influences the stage of care at which they can be applied, making them more useful for guiding postoperative management than for informing preoperative or intraoperative interventions. As a result, future work should move beyond prediction accuracy toward prospective validation, actionable hypothesis driven multi-modal models, and large-scale trials that determine whether machine learning–guided care improves pain control while reducing unnecessary opioid exposure.

Objectives: Our study aimed to identify social determinants of health (SDoH), demographic, and clinical features associated with opioid use disorder (OUD) and then to develop a risk calculator, termed Social and Clinical Opioid Use Disorder Predisposition Evaluation (SCOPE), that clinicians may use at the bedside to risk-stratify for OUD. Methods: Participant data from the All of Us Research Program were used. Multivariable logistic regression was used to identify various SDoH, demographic, and clinical predictors for OUD. The dependent variable of OUD was defined as occurring ≥90 days after participants completed a survey that queried questions related to SDoH. Calculated odds ratios from the model were used to construct a score-based calculator, SCOPE, and model performance was measured by area under the receiver operating characteristic curve (AUROC), sensitivity, and specificity. Results: There were 624,431 included participants, of whom 2475 (0.39%) had OUD that was diagnosed >90 days after survey completion. SCOPE’s performance was assessed using 10-fold cross-validation, resulting in an average AUROC [SD] of 0.85 [0.011], specificity of 78% [0.16%], and sensitivity of 78% [2.42%]. Defining a cutoff score of 8 as high-risk yielded a sensitivity of 0.79 [95% CI: 0.76–0.83], specificity of 0.76 [95% CI: 0.75–0.76], positive likelihood ratio of 3.35, and negative likelihood ratio of 0.26. Conclusions: We identified individual- and SDoH-related risk factors for OUD and developed SCOPE, which may be used as a clinical decision-making aid for personalized risk identification of OUD.

Background: Early prediction of in-hospital mortality in critically ill patients can aid clinicians in optimizing treatment. The objective was to develop a multimodal deep-learning model, using structured and unstructured clinical data, to predict in-hospital mortality risk among critically ill patients after their initial 24 h intensive care unit (ICU) admission. Methods: This study used data from Medical Information Mart for Intensive Care III (MIMIC-III), MIMIC-IV, the Electronic Intensive Care Unit Collaborative Research Database (eICU), and the high time-resolution ICU data set (HiRID). A multimodal model was developed on the MIMIC data sets, featuring time series components occurring within the first 24 h of ICU admission and predicting risk of subsequent inpatient mortality. Inputs included time-invariant variables, time-variant variables, clinical notes, and chest x-ray images. External validation occurred in a temporally separated MIMIC population, the high time-resolution ICU data set, and the eICU data set. Area under the receiver operating characteristics (AUROC), area under the precision–recall curves (AUPRC), and Brier scores were reported. Results: A total of 203,434 ICU admissions from more than 200 hospitals between 2001 to 2022 were included, in which mortality rate ranged from 5.2 to 7.9% across the four data sets. The model integrating structured data points had AUROC, AUPRC, and Brier scores of 0.92 (95% CI, 0.90 to 0.93), 0.53 (95% CI, 0.49 to 0.57), and 0.19 (95% CI, 0.18 to 0.20), respectively. The model was externally validated on eight different institutions within the eICU data set, demonstrating AUROCs ranging from 0.84 to 0.92. When including only patients with available clinical notes and imaging data, the AUROC, AUPRC, and Brier score improved from 0.87 (95% CI, 0.85 to 0.89) to 0.89 (95% CI, 0.87 to 0.91), 0.43 [(95% CI, 0.36 to 0.51) to 0.48 (95% CI, 0.40 to 0.56), and 0.37 (95% CI, 0.36 to 0.39) to 0.17 (95% CI, 0.16 to 0.19), respectively. The difference in AUROC was statistically significant based on a DeLong test ( P = 0.0167). Conclusions: These findings highlight the importance of incorporating multiple sources of patient information for mortality prediction and the importance of external validation.

We developed machine learning (ML) models to perform continuous hourly prediction of arterial blood gas (ABG) and basic metabolic panel (BMP) laboratory values in critically ill patients. We evaluated its impact on prediction of the need for renal replacement therapy (RRT). We compared the performance of the deep learning models using laboratory variables imputed hourly by our ML estimators versus models using laboratory variables imputed by a previous-value baseline. Our ML model incorporated various predictors, including previous laboratory values, administered medications, clinical events, fluid input/output, hemodynamics, and ventilator settings. Accuracy of laboratory imputations were compared using mean absolute error (MAE) and root mean squared error (RMSE). We trained bi-directional long short-term memory models (Bi-LSTM) to predict need for renal replacement therapy (RRT) that differed only in how laboratory variables were imputed over time. We compared performance using area under the receiver operating characteristics curve (AUROC) and area under the precision-recall curve (AUPRC). XGBoost achieved an average error reduction of 33% in MAE and 32% in RMSE across ABG variables, and 19% and 21% across BMP variables, respectively. Bi-LSTM models using our ML-based hourly laboratory estimates (AUROC 0.951, 95% CI 0.942–0.960; AUPRC 0.419, 95% CI 0.375–0.465) outperformed models using previous-value laboratory imputation (AUROC 0.923, 95% CI 0.910–0.935; AUPRC 0.343, 95% CI 0.301–0.386). Our real-time imputation models led to significant error reduction of most ABG and BMP targets, improving deep learning model performance for the need of RRT.

Complex regional pain syndrome (CRPS) is a multifactorial neuropathic disorder characterized by pain disproportionate to injury. We report a 58-year-old man with CRPS type II of the right lower extremity after traumatic nerve injury, diagnosed via Budapest criteria. After limited response to medical and rehabilitative therapy, ultrasound-guided percutaneous cryoneurolysis was performed on saphenous, deep peroneal, sural, and superficial peroneal nerves. The procedure yielded >80% pain reduction within 1 week and sustained analgesia with improved function at 9 months. This case suggests cryoneurolysis may represent a safe, minimally invasive option for refractory CRPS, warranting further controlled investigation.

Objectives: We evaluated bidirectional long short-term memory models for predicting inpatient mortality using different approaches to processing vital signs data collected during the initial 24 h of intensive care unit (ICU) admissions. Materials and Methods: We compared 3 vital-sign representations: (1) raw data recorded every 5 min, (2) preprocessed data averaged hourly, and (3) preprocessed data using biomarker representations that extends a digital oximetry biomarker toolbox of PhysioZoo software, applied to blood pressure, heart rate, temperature, respiratory rate, and SpO2. Results: Across 2 large ICU datasets, HiRID and eICU, models trained on the frequency-normalized representation achieved higher discrimination and lower Brier scores than those trained on raw 5-min and hourly averaged data. Discussion: The use of biomarker representations of vital signs yielded the largest improvements in discrimination and overall probabilistic performance reflected by lower Brier scores for predicting inpatient mortality by deep learning. Conclusion: Thus, we recommend using a similar approach to vital signs preprocessing for time-series predictive models.