Using machine learning techniques to develop forecasting algorithms for postoperative complications: protocol for a retrospective study

Development of a machine learning model for predicting pneumothorax risk in coaxial core needle biopsy (≤3 cm)

PURPOSE

The aim is to devise a machine learning algorithm exploiting preoperative clinical data to forecast the hazard of pneumothorax post-coaxial needle lung biopsy (CCNB), thereby informing clinical decision-making and enhancing perioperative care.

METHOD

This retrospective analysis aggregated clinical and imaging data from patients with lung nodules (≤3 cm) biopsies. Variable selection was done using univariate analysis and LASSO regression, with the dataset subsequently divided into training (80 %) and validation (20 %) subsets. Various machine learning (ML) classifiers were employed in a consolidated approach to ascertain the paramount model, which was followed by individualized risk profiling showcased through Shapley Additive eXplanations (SHAP).

RESULTS

Out of the 325 patients included in the study, 19.6% (64/325) experienced postoperative pneumothorax. High-risk factors determined were Cancer, Lesion_type, GOLD, Size, and Depth. The Gaussian Naive Bayes (GNB) classifier demonstrated superior prediction with an Area Under the Curve (AUC) of 0.82 (95% CI 0.71-0.94), complemented by an accuracy rate of 0.8, sensitivity of 0.71, specificity of 0.84, and an F1 score of 0.61 in the test cohort.

CONCLUSION

The formulated prognostic algorithm exhibited commendable efficacy in preoperatively prognosticating CCNB-induced pneumothorax, harboring the potential to refine personalized risk appraisals, steer clinical judgment, and ameliorate perioperative patient stewardship.

Machine-Learned Algorithms to Predict the Risk of Pneumothorax Requiring Chest Tube Placement after Lung Biopsy

PURPOSE

To develop a machine-learned algorithm to predict the risk of postlung biopsy pneumothorax requiring chest tube placement (CTP) to facilitate preprocedural decision making, optimize patient care, and improve resource allocation.

MATERIALS AND METHODS

This retrospective study collected clinical and imaging features of biopsy samples obtained from patients with lung nodule biopsy and included information from 59 procedures resulting in pneumothorax requiring CTP and randomly selected 67 procedures without CTP (convenience sample). The data were divided into 70 and 30 as training and testing sets, respectively. Conventional machine-learned binary classifiers were explored with preprocedural imaging and clinical data as input features and CTP as the output.

RESULTS

There was no single pathognomonic imaging or predictive clinical feature. For the independent test set under the high-specificity mode, a decision tree, logistic regression, and Naïve Bayes classifier achieved accuracies of identifying CTP at 0.79, 0.93, and 0.89 and area under receiver operating curves (AUROCs) of 0.68, 0.76, and 0.82, respectively. Under high-sensitivity mode, a decision tree, logistic regression, and Naïve Bayes achieved accuracies of identifying CTP of 0.60, 0.45, and 0.60 with AUROCs of 0.71, 0.81, and 0.82, respectively. High importance features included lesion character, chronic obstructive pulmonary disease, lesion depth, and age. A coarse decision tree requiring 4 inputs achieved comparable performance as other methods and previous machine learning prediction studies.

CONCLUSIONS

The results support the possibility of predicting pneumothorax requiring CTP after biopsy based on an automated decision support, reliant on readily available preprocedural information.

Social Vulnerability and Surgery Outcomes: A Cross-sectional Analysis

Background

Post-operative complications present a challenge to the healthcare system due to the high unpredictability of their incidence. However, the socioeconomic factors that relate to postoperative complications are still unclear as they can be heterogeneous based on communities, types of surgical services, and sex and gender.

Methods

In this study, we conducted a large population cross-sectional analysis of social vulnerability and the odds of various post-surgical complications. We built statistical logistic regression models of postsurgical complications with social vulnerability index as the independent variable along with sex interaction.

Results

We found that social vulnerability was associated with abnormal heart rhythm with socioeconomic status and housing status being the main association factors. We also found associations of the interaction of social vulnerability and female sex with an increase in odds of heart attack and surgical wound infection.

Conclusions

Our results indicate that social vulnerability measures such as socioeconomic status and housing conditions could be related to health outcomes. This suggests that the domain of preventive medicine should place social vulnerability as a priority to achieve its goals.

Integrating machine learning predictions for perioperative risk management: Towards an empirical design of a flexible-standardized risk assessment tool

BACKGROUND

Surgical patients are complex, vulnerable, and prone to postoperative complications that can potentially be mitigated with quality perioperative risk assessment and management. Several institutions have incorporated machine learning (ML) into their patient care to improve awareness and support clinician decision-making along the perioperative spectrum. Recent research suggests that ML risk prediction can support perioperative patient risk monitoring and management across several situations, including the operating room (OR) to intensive care unit (ICU) handoffs.

OBJECTIVES

Our study objectives were threefold: (1) evaluate whether ML-generated postoperative predictions are concordant with clinician-generated risk rankings for acute kidney injury, delirium, pneumonia, deep vein thrombosis, and pulmonary embolism, and establish their associated risk factors; (2) ascertain clinician end-user suggestions to improve adoption of ML-generated risks and their integration into the perioperative workflow; and (3) develop a user-friendly visualization format for a tool to display ML-generated risks and risk factors to support postoperative care planning, for example, within the context of OR-ICU handoffs.

METHODS

Graphical user interfaces for postoperative risk prediction models were assessed for end-user usability through cognitive walkthroughs and interviews with anesthesiologists, surgeons, certified registered nurse anesthetists, registered nurses, and critical care physicians. Thematic analysis relying on an explanation design framework was used to identify feedback and suggestions for improvement.

RESULTS

17 clinicians participated in the evaluation. ML estimates of complication risks aligned with clinicians' independent rankings, and related displays were perceived as valuable for decision-making and care planning for postoperative care. During OR-ICU handoffs, the tool could speed up report preparation and remind clinicians to address patient-specific complications, thus providing more tailored care information. Suggestions for improvement centered on electronic tool delivery; methods to build trust in ML models; modifiable risks and risk mitigation strategies; and additional patient information based on individual preferences (e.g., surgical procedure).

CONCLUSIONS

ML estimates of postoperative complication risks can provide anticipatory guidance, potentially increasing the efficiency of care planning. We have offered an ML visualization framework for designing future ML-augmented tools and anticipate the development of tools that recommend specific actions to the user based on ML model output.

Protocol for the perioperative outcome risk assessment with computer learning enhancement (Periop ORACLE) randomized study

Background: More than four million people die each year in the month following surgery, and many more experience complications such as acute kidney injury. Some of these outcomes may be prevented through early identification of at-risk patients and through intraoperative risk mitigation. Telemedicine has revolutionized the way at-risk patients are identified in critical care, but intraoperative telemedicine services are not widely used in anesthesiology. Clinicians in telemedicine settings may assist with risk stratification and brainstorm risk mitigation strategies while clinicians in the operating room are busy performing other patient care tasks. Machine learning tools may help clinicians in telemedicine settings leverage the abundant electronic health data available in the perioperative period. The primary hypothesis for this study is that anesthesiology clinicians can predict postoperative complications more accurately with machine learning assistance than without machine learning assistance. Methods: This investigation is a sub-study nested within the TECTONICS randomized clinical trial (NCT03923699). As part of TECTONICS, study team members who are anesthesiology clinicians working in a telemedicine setting are currently reviewing ongoing surgical cases and documenting how likely they feel the patient is to experience 30-day in-hospital death or acute kidney injury. For patients who are included in this sub-study, these case reviews will be randomized to be performed with access to a display showing machine learning predictions for the postoperative complications or without access to the display. The accuracy of the predictions will be compared across these two groups. Conclusion: Successful completion of this study will help define the role of machine learning not only for intraoperative telemedicine, but for other risk assessment tasks before, during, and after surgery. Registration: ORACLE is registered on ClinicalTrials.gov: NCT05042804; registered September 13, 2021.

Real-time risk prediction of colorectal surgery-related post-surgical complications using GRU-D model

BACKGROUND

Post-surgical complications (PSCs) have been an increasing concern for hospitals in light of Medicare penalties for 30-day readmissions. PSCs have become a target for quality improvement and benchmark for the healthcare system. Over half (60 %) of the deep or organ space surgical site infections are discovered after discharge, leading to a readmission. There has thus been a push to develop risk prediction models for targeted preventive interventions for PSCs.

OBJECTIVE

We experiment several Gated Recurrent Unit with Decay (GRU-D) based deep learning architectures with various feature sampling schemes in predicting the risk of colorectal PSCs and compare with atemporal logistic regression models (logit).

METHOD

We used electronic health record (EHR) data of 3,535 colorectal surgical patients involved in the national surgical quality improvement program (NSQIP) between 2006 and 2018. Single layer, stacked layer, and multimodal GRU-D models with sigmoid activation were used to develop risk prediction models. Area Under the Receiver Operating Characteristic curve (AUROC) was calculated by comparing predicted probability of developing complications versus truly observed PSCs (NSQIP adjudicated) within 30 days after surgery. We set up cross-validation and an independent held-out dataset for testing model performance consistency.

RESULTS AND CONCLUSION

The primary contribution of our study is the formulation of a novel real-time PSC risk prediction task using GRU-D with demonstrated clinical utility. GRU-D outperforms the logit model in predicting wound and organ space infection and shows improved performance as additional data points become available. Logit model outperforms GRU-D before surgery for superficial infection and bleeding. For the same sampling scheme, there is no obvious advantage of complex architectures (stacked, multimodal) over single layer GRU-D. Obtaining more data points closer to the occurrence of PSCs is more important than using a more frequent sampling scheme in training GRU-D models. The fourth predicted risk quartile by single layer GRU-D contains 63 %, 59 %, and 66 % organ space infection cases, at 4 h before, 72 h after, and 168 h after surgery, respectively, suggesting its potential application as a bedside risk assessment tool.

Surgery duration: Optimized prediction and causality analysis

Accurate estimation of duration of surgery (DOS) can lead to cost-effective utilization of surgical staff and operating rooms and decrease patients’ waiting time. In this study, we present a supervised DOS nonlinear regression prediction model whose accuracy outperforms earlier results. In addition, unlike previous studies, we identify the features that influence DOS prediction. Further, in difference from others, we study the causal relationship between the feature set and DOS. The feature sets used in prior studies included a subset of the features presented in this study. This study aimed to derive influential effectors of duration of surgery via optimized prediction and causality analysis. We implemented an array of machine learning algorithms and trained them on datasets comprising surgery-related data, to derive DOS prediction models. The datasets we acquired contain patient, surgical staff, and surgery features. The datasets comprised 23,293 surgery records of eight surgery types performed over a 10-year period in a public hospital. We have introduced new, unstudied features and combined them with features adopted from previous studies to generate a comprehensive feature set. We utilized feature importance methods to identify the influential features, and causal inference methods to identify the causal features. Our model demonstrates superior performance in comparison to DOS prediction models in the art. The performance of our DOS model in terms of the mean absolute error (MAE) was 14.9 minutes. The algorithm that derived the model with the best performance was the gradient boosted trees (GBT). We identified the 10 most influential features and the 10 most causal features. In addition, we showed that 40% of the influential features have a significant (p-value = 0.05) causal relationship with DOS. We developed a DOS prediction model whose accuracy is higher than that of prior models. This improvement is achieved via the introduction of a novel feature set on which the model was trained. Utilizing our prediction model, hospitals can improve the efficiency of surgery schedules, and by exploiting the identified causal relationship, can influence the DOS. Further, the feature importance methods we used can help explain the model’s predictions.

Machine learning for real-time aggregated prediction of hospital admission for emergency patients

Machine learning for hospital operations is under-studied. We present a prediction pipeline that uses live electronic health-records for patients in a UK teaching hospital’s emergency department (ED) to generate short-term, probabilistic forecasts of emergency admissions. A set of XGBoost classifiers applied to 109,465 ED visits yielded AUROCs from 0.82 to 0.90 depending on elapsed visit-time at the point of prediction. Patient-level probabilities of admission were aggregated to forecast the number of admissions among current ED patients and, incorporating patients yet to arrive, total emergency admissions within specified time-windows. The pipeline gave a mean absolute error (MAE) of 4.0 admissions (mean percentage error of 17%) versus 6.5 (32%) for a benchmark metric. Models developed with 104,504 later visits during the Covid-19 pandemic gave AUROCs of 0.68–0.90 and MAE of 4.2 (30%) versus a 4.9 (33%) benchmark. We discuss how we surmounted challenges of designing and implementing models for real-time use, including temporal framing, data preparation, and changing operational conditions.

Improving quality and efficiency of care for advanced appendicitis in children

BACKGROUND

Complicated appendicitis encompasses a spectrum of severity with heterogeneity in definition and substantial variation in care. Enhanced recovery after surgery or 'fast-track' protocols aim to reduce practice variation by standardizing care. These initiatives may improve quality and efficiency of care, preserve resources and expedite discharge. This study aims to evaluate the impact of a standardized Enhanced Recovery Pathway (ERP) on the post-operative recovery of children with a subset of complicated appendicitis termed 'advanced' appendicitis.

METHODS

We defined advanced appendicitis as gangrenous or suppurative appendicitis without perforation, contained iatrogenic perforation, or localized purulent fluid. Children with operative findings reflecting these criteria were enrolled in the ERP protocol. Key protocol components include early upgrade of diet, avoidance of intravenous analgesia, abridged intravenous antibiotics, early ambulation and standardized discharge criteria. The study period was May 2018 to June 2019. A historical cohort was used as the comparator group.

RESULTS

Outcomes for 44 children treated under the ERP were compared to 44 historical controls. There was a 20% reduction in median post-operative length of stay (1.80 vs. 2.24 days, p = 0.02). Intravenous analgesia was received by fewer patients (6.8% vs. 36.4%, p = 0.01) with significant reduction in antiemetic requirement (p = 0.03). No significant difference in 30-day complication rates was observed.

CONCLUSION

Reduced post-operative length of stay and reduction in practice variation were achieved after implementation of a 'fast-track' protocol for children with advanced appendicitis. Additional benefits of this protocol include reduced provision of intravenous morphine analgesia, decreased resource use and cost savings.

Use of Machine Learning to Develop and Evaluate Models Using Preoperative and Intraoperative Data to Identify Risks of Postoperative Complications

IMPORTANCE

Postoperative complications can significantly impact perioperative care management and planning.

OBJECTIVES

To assess machine learning (ML) models for predicting postoperative complications using independent and combined preoperative and intraoperative data and their clinically meaningful model-agnostic interpretations.

DESIGN, SETTING, AND PARTICIPANTS

This retrospective cohort study assessed 111 888 operations performed on adults at a single academic medical center from June 1, 2012, to August 31, 2016, with a mean duration of follow-up based on the length of postoperative hospital stay less than 7 days. Data analysis was performed from February 1 to September 31, 2020.

MAIN OUTCOMES AND MEASURES

Outcomes included 5 postoperative complications: acute kidney injury (AKI), delirium, deep vein thrombosis (DVT), pulmonary embolism (PE), and pneumonia. Patient and clinical characteristics available preoperatively, intraoperatively, and a combination of both were used as inputs for 5 candidate ML models: logistic regression, support vector machine, random forest, gradient boosting tree (GBT), and deep neural network (DNN). Model performance was compared using the area under the receiver operating characteristic curve (AUROC). Model interpretations were generated using Shapley Additive Explanations by transforming model features into clinical variables and representing them as patient-specific visualizations.

RESULTS

A total of 111 888 patients (mean [SD] age, 54.4 [16.8] years; 56 915 [50.9%] female; 82 533 [73.8%] White) were included in this study. The best-performing model for each complication combined the preoperative and intraoperative data with the following AUROCs: pneumonia (GBT), 0.905 (95% CI, 0.903-0.907); AKI (GBT), 0.848 (95% CI, 0.846-0.851); DVT (GBT), 0.881 (95% CI, 0.878-0.884); PE (DNN), 0.831 (95% CI, 0.824-0.839); and delirium (GBT), 0.762 (95% CI, 0.759-0.765). Performance of models that used only preoperative data or only intraoperative data was marginally lower than that of models that used combined data. When adding variables with missing data as input, AUROCs increased from 0.588 to 0.905 for pneumonia, 0.579 to 0.848 for AKI, 0.574 to 0.881 for DVT, 0.5 to 0.831 for PE, and 0.6 to 0.762 for delirium. The Shapley Additive Explanations analysis generated model-agnostic interpretation that illustrated significant clinical contributors associated with risks of postoperative complications.

CONCLUSIONS AND RELEVANCE

The ML models for predicting postoperative complications with model-agnostic interpretation offer opportunities for integrating risk predictions for clinical decision support. Such real-time clinical decision support can mitigate patient risks and help in anticipatory management for perioperative contingency planning.

Prediction of Postoperative Complications for Patients of End Stage Renal Disease

End stage renal disease (ESRD) is the last stage of chronic kidney disease that requires dialysis or a kidney transplant to survive. Many studies reported a higher risk of mortality in ESRD patients compared with patients without ESRD. In this paper, we develop a model to predict postoperative complications, major cardiac event, for patients who underwent any type of surgery. We compare several widely-used machine learning models through experiments with our collected data yellow of size 3220, and achieved F1 score of 0.797 with the random forest model. Based on experimental results, we found that features related to operation (e.g., anesthesia time, operation time, crystal, and colloid) have the biggest impact on model performance, and also found the best combination of features. We believe that this study will allow physicians to provide more appropriate therapy to the ESRD patients by providing information on potential postoperative complications.

Potential predictors of type-2 diabetes risk: machine learning, synthetic data and wearable health devices

BACKGROUND

The aim of a recent research project was the investigation of the mechanisms involved in the onset of type 2 diabetes in the absence of familiarity. This has led to the development of a computational model that recapitulates the aetiology of the disease and simulates the immunological and metabolic alterations linked to type-2 diabetes subjected to clinical, physiological, and behavioural features of prototypical human individuals.

RESULTS

We analysed the time course of 46,170 virtual subjects, experiencing different lifestyle conditions. We then set up a statistical model able to recapitulate the simulated outcomes.

CONCLUSIONS

The resulting machine learning model adequately predicts the synthetic dataset and can, therefore, be used as a computationally-cheaper version of the detailed mathematical model, ready to be implemented on mobile devices to allow self-assessment by informed and aware individuals. The computational model used to generate the dataset of this work is available as a web-service at the following address: http://kraken.iac.rm.cnr.it/T2DM .

Machine-Learning-Based Approach to Differential Diagnosis in Tuberculous and Viral Meningitis

Background

Tuberculous meningitis (TBM) is the most severe form of tuberculosis, but differentiating between the diagnosis of TBM and viral meningitis (VM) is difficult. Thus, we have developed machine-learning modules for differentiating TBM from VM.

Material and Methods

For the training data, confirmed or probable TBM and confirmed VM cases were retrospectively collected from five teaching hospitals in Korea between January 2000 - July 2018. Various machine-learning algorithms were used for training. The machine-learning algorithms were tested by the leave-one-out cross-validation. Four residents and two infectious disease specialists were tested using the summarized medical information.

Results

The training study comprised data from 60 patients with confirmed or probable TBM and 143 patients with confirmed VM. Older age, longer symptom duration before the visit, lower serum sodium, lower cerebrospinal fluid (CSF) glucose, higher CSF protein, and CSF adenosine deaminase were found in the TBM patients. Among the various machine-learning algorithms, the area under the curve (AUC) of the receiver operating characteristics of artificial neural network (ANN) with ImperativeImputer for matrix completion (0.85; 95% confidence interval 0.79 - 0.89) was found to be the highest. The AUC of the ANN model was statistically higher than those of all the residents (range 0.67 - 0.72, P <0.001) and an infectious disease specialist (AUC 0.76; P = 0.03).

Conclusion

The machine-learning techniques may play a role in differentiating between TBM and VM. Specifically, the ANN model seems to have better diagnostic performance than the non-expert clinician.

Protocol for a proof-of-concept observational study evaluating the potential utility and acceptability of a telemedicine solution for the post-anesthesia care unit

Introduction: The post-anesthesia care unit (PACU) is a clinical area designated for patients recovering from invasive procedures. There are typically several geographically dispersed PACUs within hospitals. Patients in the PACU can be unstable and at risk for complications. However, clinician coverage and patient monitoring in PACUs is not well regulated and might be sub-optimal. We hypothesize that a telemedicine center for the PACU can improve key PACU functions.

Objectives: The objective of this study is to demonstrate the potential utility and acceptability of a telemedicine center to complement the key functions of the PACU. These include participation in hand-off activities to and from the PACU, detection of physiological derangements, identification of symptoms requiring treatment, recognition of situations requiring emergency medical intervention, and determination of patient readiness for PACU discharge.

Methods and analysis: This will be a single center prospective before-and-after proof-of-concept study. Adults (18 years and older) undergoing elective surgery and recovering in two selected PACU bays will be enrolled. During the initial three-month observation phase, clinicians in the telemedicine center will not communicate with clinicians in the PACU, unless there is a specific patient safety concern. During the subsequent three-month interaction phase, clinicians in the telemedicine center will provide structured decision support to PACU clinicians. The primary outcome will be time to PACU discharge readiness determination in the two study phases. The attitudes of key stakeholders towards the telemedicine center will be assessed. Other outcomes will include detection of physiological derangements, complications, adverse symptoms requiring treatments, and emergencies requiring medical intervention.

Registration: This trial is registered on clinicaltrials.gov, NCT04020887 (16 ^th July 2019).

Machine Learning Model for Risk Prediction of Community-Acquired Acute Kidney Injury Hospitalization From Electronic Health Records: Development and Validation Study

BACKGROUND

Community-acquired acute kidney injury (CA-AKI)-associated hospitalizations impose significant health care needs and contribute to in-hospital mortality. However, most risk prediction models developed to date have focused on AKI in a specific group of patients during hospitalization, and there is limited knowledge on the baseline risk in the general population for preventing CA-AKI-associated hospitalization.

OBJECTIVE

To gain further insight into risk exploration, the aim of this study was to develop, validate, and establish a scoring system to facilitate health professionals in enabling early recognition and intervention of CA-AKI to prevent permanent kidney damage using different machine-learning techniques.

METHODS

A nested case-control study design was employed using electronic health records derived from a group of Chang Gung Memorial Hospitals in Taiwan from 2010 to 2017 to identify 234,867 adults with at least two measures of serum creatinine at hospital admission. Patients were classified into a derivation cohort (2010-2016) and a temporal validation cohort (2017). Patients with the first episode of CA-AKI at hospital admission were classified into the case group and those without CA-AKI were classified in the control group. A total of 47 potential candidate variables, including age, gender, prior use of nephrotoxic medications, Charlson comorbid conditions, commonly measured laboratory results, and recent use of health services, were tested to develop a CA-AKI hospitalization risk model. Permutation-based selection with both the extreme gradient boost (XGBoost) and least absolute shrinkage and selection operator (LASSO) algorithms was performed to determine the top 10 important features for scoring function development.

RESULTS

The discriminative ability of the risk model was assessed by the area under the receiver operating characteristic curve (AUC), and the predictive CA-AKI risk model derived by the logistic regression algorithm achieved an AUC of 0.767 (95% CI 0.764-0.770) on derivation and 0.761 on validation for any stage of AKI, with positive and negative predictive values of 19.2% and 96.1%, respectively. The risk model for prediction of CA-AKI stages 2 and 3 had an AUC value of 0.818 for the validation cohort with positive and negative predictive values of 13.3% and 98.4%, respectively. These metrics were evaluated at a cut-off value of 7.993, which was determined as the threshold to discriminate the risk of AKI.

CONCLUSIONS

A machine learning-generated risk score model can identify patients at risk of developing CA-AKI-related hospitalization through a routine care data-driven approach. The validated multivariate risk assessment tool could help clinicians to stratify patients in primary care, and to provide monitoring and early intervention for preventing AKI while improving the quality of AKI care in the general population.

Machine learning mortality classification in clinical documentation with increased accuracy in visual-based analyses

AIM

The role of machine learning on clinical documentation for predictive outcomes remains undefined. We aimed to compare three neural networks on inpatient providers' notes to predict mortality in neonatal hypoxic-ischaemic encephalopathy (HIE).

METHODS

Using Children's Hospitals Neonatal Database, non-anomalous neonates with HIE treated with therapeutic hypothermia were identified at a single-centre. Data were linked with the initial seven days of documentation. Exposures were derived using the databases and applying convolutional and two recurrent neural networks. The primary outcome was mortality. The predictive accuracy and performance measures for models were determined.

RESULTS

The cohort included 52 eligible infants. Most infants survived (n = 36, 69%) and 23 had severe HIE (44%). Neural networks performed above baseline and differed in their median accuracy for predicting mortality (P = .0001): recurrent models with long short-term memory 69% (25^th , 75^th percentile 65, 73%) and gated-recurrent model units 65% (62, 69%) and convolutional 72% (64, 96%). Convolutional networks' median specificity was 81% (72, 97%).

CONCLUSION

The neural network models demonstrated fundamental validity in predicting mortality using inpatient provider documentation. Convolutional models had high specificity for (excluding) mortality in neonatal HIE. These findings provide a platform for future model training and ultimately tool development to assist clinicians in patient assessments and risk stratifications.

Predicting intra-operative and postoperative consequential events using machine-learning techniques in patients undergoing robot-assisted partial nephrectomy: a Vattikuti Collective Quality Initiative database study

OBJECTIVE

To predict intra-operative (IOEs) and postoperative events (POEs) consequential to the derailment of the ideal clinical course of patient recovery.

MATERIALS AND METHODS

The Vattikuti Collective Quality Initiative is a multi-institutional dataset of patients who underwent robot-assisted partial nephectomy for kidney tumours. Machine-learning (ML) models were constructed to predict IOEs and POEs using logistic regression, random forest and neural networks. The models to predict IOEs used patient demographics and preoperative data. In addition to these, intra-operative data were used to predict POEs. Performance on the test dataset was assessed using area under the receiver-operating characteristic curve (AUC-ROC) and area under the precision-recall curve (PR-AUC).

RESULTS

The rates of IOEs and POEs were 5.62% and 20.98%, respectively. Models for predicting IOEs were constructed using data from 1690 patients and 38 variables; the best model had an AUC-ROC of 0.858 (95% confidence interval [CI] 0.762, 0.936) and a PR-AUC of 0.590 (95% CI 0.400, 0.759). Models for predicting POEs were trained using data from 1406 patients and 59 variables; the best model had an AUC-ROC of 0.875 (95% CI 0.834, 0.913) and a PR-AUC 0.706 (95% CI, 0.610, 0.790).

CONCLUSIONS

The performance of the ML models in the present study was encouraging. Further validation in a multi-institutional clinical setting with larger datasets would be necessary to establish their clinical value. ML models can be used to predict significant events during and after surgery with good accuracy, paving the way for application in clinical practice to predict and intervene at an opportune time to avert complications and improve patient outcomes.

Artificial Intelligence and Machine Learning in Anesthesiology

Commercial applications of artificial intelligence and machine learning have made remarkable progress recently, particularly in areas such as image recognition, natural speech processing, language translation, textual analysis, and self-learning. Progress had historically languished in these areas, such that these skills had come to seem ineffably bound to intelligence. However, these commercial advances have performed best at single-task applications in which imperfect outputs and occasional frank errors can be tolerated.The practice of anesthesiology is different. It embodies a requirement for high reliability, and a pressured cycle of interpretation, physical action, and response rather than any single cognitive act. This review covers the basics of what is meant by artificial intelligence and machine learning for the practicing anesthesiologist, describing how decision-making behaviors can emerge from simple equations. Relevant clinical questions are introduced to illustrate how machine learning might help solve them-perhaps bringing anesthesiology into an era of machine-assisted discovery.

Protocol for the Effectiveness of an Anesthesiology Control Tower System in Improving Perioperative Quality Metrics and Clinical Outcomes: the TECTONICS randomized, pragmatic trial

Introduction: Perioperative morbidity is a public health priority, and surgical volume is increasing rapidly. With advances in technology, there is an opportunity to research the utility of a telemedicine-based control center for anesthesia clinicians that assess risk, diagnoses negative patient trajectories, and implements evidence-based practices. Objectives: The primary objective of this trial is to determine whether an anesthesiology control tower (ACT) prevents clinically relevant adverse postoperative outcomes including 30-day mortality, delirium, respiratory failure, and acute kidney injury. Secondary objectives are to determine whether the ACT improves perioperative quality of care metrics including management of temperature, mean arterial pressure, mean airway pressure with mechanical ventilation, blood glucose, anesthetic concentration, antibiotic redosing, and efficient fresh gas flow. Methods and analysis: We are conducting a single center, randomized, controlled, phase 3 pragmatic clinical trial. A total of 58 operating rooms are randomized daily to receive support from the ACT or not. All adults (eighteen years and older) undergoing surgical procedures in these operating rooms are included and followed until 30 days after their surgery. Clinicians in operating rooms randomized to ACT support receive decision support from clinicians in the ACT. In operating rooms randomized to no intervention, the current standard of anesthesia care is delivered. The intention-to-treat principle will be followed for all analyses. Differences between groups will be presented with 99% confidence intervals; p-values <0.005 will be reported as providing compelling evidence, and p-values between 0.05 and 0.005 will be reported as providing suggestive evidence. Registration: TECTONICS is registered on ClinicalTrials.gov, NCT03923699; registered on 23 April 2019.

Deep-learning model for predicting 30-day postoperative mortality

BACKGROUND

Postoperative mortality occurs in 1-2% of patients undergoing major inpatient surgery. The currently available prediction tools using summaries of intraoperative data are limited by their inability to reflect shifting risk associated with intraoperative physiological perturbations. We sought to compare similar benchmarks to a deep-learning algorithm predicting postoperative 30-day mortality.

METHODS

We constructed a multipath convolutional neural network model using patient characteristics, co-morbid conditions, preoperative laboratory values, and intraoperative numerical data from patients undergoing surgery with tracheal intubation at a single medical centre. Data for 60 min prior to a randomly selected time point were utilised. Model performance was compared with a deep neural network, a random forest, a support vector machine, and a logistic regression using predetermined summary statistics of intraoperative data.

RESULTS

Of 95 907 patients, 941 (1%) died within 30 days. The multipath convolutional neural network predicted postoperative 30-day mortality with an area under the receiver operating characteristic curve of 0.867 (95% confidence interval [CI]: 0.835-0.899). This was higher than that for the deep neural network (0.825; 95% CI: 0.790-0.860), random forest (0.848; 95% CI: 0.815-0.882), support vector machine (0.836; 95% CI: 0.802-870), and logistic regression (0.837; 95% CI: 0.803-0.871).

CONCLUSIONS

A deep-learning time-series model improves prediction compared with models with simple summaries of intraoperative data. We have created a model that can be used in real time to detect dynamic changes in a patient's risk for postoperative mortality.

Who is a high-risk surgical patient?

PURPOSE OF REVIEW

Timely identification of high-risk surgical candidates facilitate surgical decision-making and allows appropriate tailoring of perioperative management strategies. This review aims to summarize the recent advances in perioperative risk stratification.

RECENT FINDINGS

Use of indices which include various combinations of preoperative and postoperative variables remain the most commonly used risk-stratification strategy. Incorporation of biomarkers (troponin and natriuretic peptides), comprehensive objective assessment of functional capacity, and frailty into the current framework enhance perioperative risk estimation. Intraoperative hemodynamic parameters can provide further signals towards identifying patients at risk of adverse postoperative outcomes. Implementation of machine-learning algorithms is showing promising results in real-time forecasting of perioperative outcomes.

SUMMARY

Perioperative risk estimation is multidimensional including validated indices, biomarkers, functional capacity estimation, and intraoperative hemodynamics. Identification and implementation of targeted strategies which mitigate predicted risk remains a greater challenge.

Emergence of digital biomarkers to predict and modify treatment efficacy: machine learning study

OBJECTIVES

Development of digital biomarkers to predict treatment response to a digital behavioural intervention.

DESIGN

Machine learning using random forest classifiers on data generated through the use of a digital therapeutic which delivers behavioural therapy to treat cardiometabolic disease. Data from 13 explanatory variables (biometric and engagement in nature) generated in the first 28 days of a 12-week intervention were used to train models. Two levels of response to treatment were predicted: (1) systolic change ≥10 mm Hg (SC model), and (2) shift down to a blood pressure category of elevated or better (ER model). Models were validated using leave-one-out cross validation and evaluated using area under the curve receiver operating characteristics (AUROC) and specificity- sensitivity. Ability to predict treatment response with a subset of nine variables, including app use and baseline blood pressure, was also tested (models SC-APP and ER-APP).

SETTING

Data generated through ad libitum use of a digital therapeutic in the USA.

PARTICIPANTS

Deidentified data from 135 adults with a starting blood pressure ≥130/80, who tracked blood pressure for at least 7 weeks using the digital therapeutic.

RESULTS

The SC model had an AUROC of 0.82 and a sensitivity of 58% at a specificity of 90%. The ER model had an AUROC of 0.69 and a sensitivity of 32% at a specificity at 91%. Dropping explanatory variables related to blood pressure resulted in an AUROC of 0.72 with a sensitivity of 42% at a specificity of 90% for the SC-APP model and an AUROC of 0.53 for the ER-APP model.

CONCLUSIONS

Machine learning was used to transform data from a digital therapeutic into digital biomarkers that predicted treatment response in individual participants. Digital biomarkers have potential to improve treatment outcomes in a digital behavioural intervention.

Computational Approaches and Challenges to Developing Universal Influenza Vaccines

The traditional design of effective vaccines for rapidly-evolving pathogens, such as influenza A virus, has failed to provide broad spectrum and long-lasting protection. With low cost whole genome sequencing technology and powerful computing capabilities, novel computational approaches have demonstrated the potential to facilitate the design of a universal influenza vaccine. However, few studies have integrated computational optimization in the design and discovery of new vaccines. Understanding the potential of computational vaccine design is necessary before these approaches can be implemented on a broad scale. This review summarizes some promising computational approaches under current development, including computationally optimized broadly reactive antigens with consensus sequences, phylogenetic model-based ancestral sequence reconstruction, and immunomics to compute conserved cross-reactive T-cell epitopes. Interactions between virus-host-environment determine the evolvability of the influenza population. We propose that with the development of novel technologies that allow the integration of data sources such as protein structural modeling, host antibody repertoire analysis and advanced phylodynamic modeling, computational approaches will be crucial for the development of a long-lasting universal influenza vaccine. Taken together, computational approaches are powerful and promising tools for the development of a universal influenza vaccine with durable and broad protection.

Clinical Requirements of Future Patient Monitoring in the Intensive Care Unit: Qualitative Study

Background

In the intensive care unit (ICU), continuous patient monitoring is essential to detect critical changes in patients’ health statuses and to guide therapy. The implementation of digital health technologies for patient monitoring may further improve patient safety. However, most monitoring devices today are still based on technologies from the 1970s.

Objective

The aim of this study was to evaluate statements by ICU staff on the current patient monitoring systems and their expectations for future technological developments in order to investigate clinical requirements and barriers to the implementation of future patient monitoring.

Methods

This prospective study was conducted at three intensive care units of a German university hospital. Guideline-based interviews with ICU staff—5 physicians, 6 nurses, and 4 respiratory therapists—were recorded, transcribed, and analyzed using the grounded theory approach.

Results

Evaluating the current monitoring system, ICU staff put high emphasis on usability factors such as intuitiveness and visualization. Trend analysis was rarely used; inadequate alarm management as well as the entanglement of monitoring cables were rated as potential patient safety issues. For a future system, the importance of high usability was again emphasized; wireless, noninvasive, and interoperable monitoring sensors were desired; mobile phones for remote patient monitoring and alarm management optimization were needed; and clinical decision support systems based on artificial intelligence were considered useful. Among perceived barriers to implementation of novel technology were lack of trust, fear of losing clinical skills, fear of increasing workload, and lack of awareness of available digital technologies.

Conclusions

This qualitative study on patient monitoring involves core statements from ICU staff. To promote a rapid and sustainable implementation of digital health solutions in the ICU, all health care stakeholders must focus more on user-derived findings. Results on alarm management or mobile devices may be used to prepare ICU staff to use novel technology, to reduce alarm fatigue, to improve medical device usability, and to advance interoperability standards in intensive care medicine. For digital transformation in health care, increasing the trust and awareness of ICU staff in digital health technology may be an essential prerequisite.

Trial Registration

ClinicalTrials.gov NCT03514173; https://clinicaltrials.gov/ct2/show/NCT03514173 (Archived by WebCite at http://www.webcitation.org/77T1HwOzk)

What we can learn from Big Data about factors influencing perioperative outcome

PURPOSE OF REVIEW

This narrative review will discuss what value Big Data has to offer anesthesiology and aims to highlight recently published articles of large databases exploring factors influencing perioperative outcome. Additionally, the future perspectives of Big Data and its major pitfalls will be discussed.

RECENT FINDINGS

The potential of Big Data has given an incentive to create nationwide and anesthesia-initiated registries like the MPOG and NACOR. These large databases have contributed in elucidating some of the rare perioperative complications, such as declined cognition after exposure to general anesthesia and epidural hematomas in parturients. Additionally, they are useful in finding patterns such as similar outcome in subtypes of beta-blockers and lower incidence of pneumonia in preoperative influenza vaccinations in the elderly.

SUMMARY

Big Data is becoming increasingly popular with the collaborative collection of registries offering anesthesia a way to explore rare perioperative complications and outcome to encourage further hypotheses testing. Although Big Data has its flaws in security, lack of expertise and methodological concerns, the future potential of analytics combined with genomics, machine learning and real-time decision support looks promising.

An update on postoperative delirium: Clinical features, neuropathogenesis, and perioperative management

PURPOSE OF REVIEW

We present a focused review on postoperative delirium for anesthesiologists, encompassing clinical features, neuropathogenesis, and clinical identification and management strategies based on risk factors and current delirium treatments.

RECENT FINDINGS

The literature on postoperative delirium is dominated by non-experimental studies. We review delirium phenotypes, diagnostic criteria, and present standard nomenclature based on current literature. Disruption of cortical integration of complex information (CICI) may provide a framework to understand the neuropathogenesis of postoperative delirium, as well as risk factors and clinical modifiers in the perioperative period. We further divide risk factors into patient factors, surgical factors, and medical/pharmacological factors, and present specific considerations for each in the preoperative, intraoperative, and postoperative periods.

SUMMARY

Postoperative delirium is prevalent, poorly understood, and often missed with current screening techniques. Proper identification of risk factors is useful for perioperative interventions and can help tailor patient-specific management strategies.

Study protocol for the Anesthesiology Control Tower-Feedback Alerts to Supplement Treatments (ACTFAST-3) trial: a pilot randomized controlled trial in intraoperative telemedicine

Background: Each year, over 300 million people undergo surgical procedures worldwide. Despite efforts to improve outcomes, postoperative morbidity and mortality are common. Many patients experience complications as a result of either medical error or failure to adhere to established clinical practice guidelines. This protocol describes a clinical trial comparing a telemedicine-based decision support system, the Anesthesiology Control Tower (ACT), with enhanced standard intraoperative care. Methods: This study is a pragmatic, comparative effectiveness trial that will randomize approximately 12,000 adult surgical patients on an operating room (OR) level to a control or to an intervention group. All OR clinicians will have access to decision support software within the OR as a part of enhanced standard intraoperative care. The ACT will monitor patients in both groups and will provide additional support to the clinicians assigned to intervention ORs. Primary outcomes include blood glucose management and temperature management. Secondary outcomes will include surrogate, clinical, and economic outcomes, such as incidence of intraoperative hypotension, postoperative respiratory compromise, acute kidney injury, delirium, and volatile anesthetic utilization. Ethics and dissemination: The ACTFAST-3 study has been approved by the Human Resource Protection Office (HRPO) at Washington University in St. Louis and is registered at clinicaltrials.gov ( NCT02830126). Recruitment for this protocol began in April 2017 and will end in December 2018. Dissemination of the findings of this study will occur via presentations at academic conferences, journal publications, and educational materials.

Study protocol for the Anesthesiology Control Tower-Feedback Alerts to Supplement Treatments (ACTFAST-3) trial: a pilot randomized controlled trial in intraoperative telemedicine

Background: Each year, over 300 million people undergo surgical procedures worldwide. Despite efforts to improve outcomes, postoperative morbidity and mortality are common. Many patients experience complications as a result of either medical error or failure to adhere to established clinical practice guidelines. This protocol describes a clinical trial comparing a telemedicine-based decision support system, the Anesthesiology Control Tower (ACT), with enhanced standard intraoperative care. Methods: This study is a pragmatic, comparative effectiveness trial that will randomize approximately 12,000 adult surgical patients on an operating room (OR) level to a control or to an intervention group. All OR clinicians will have access to decision support software within the OR as a part of enhanced standard intraoperative care. The ACT will monitor patients in both groups and will provide additional support to the clinicians assigned to intervention ORs. Primary outcomes include blood glucose management and temperature management. Secondary outcomes will include surrogate, clinical, and economic outcomes, such as incidence of intraoperative hypotension, postoperative respiratory compromise, acute kidney injury, delirium, and volatile anesthetic utilization. Ethics and dissemination: The ACTFAST-3 study has been approved by the Human Resource Protection Office (HRPO) at Washington University in St. Louis and is registered at clinicaltrials.gov ( NCT02830126). Recruitment for this protocol began in April 2017 and will end in December 2018. Dissemination of the findings of this study will occur via presentations at academic conferences, journal publications, and educational materials.