Clinical decision-making substantially impacts patients’ lives and their quality of life. However, the black-box nature of AI-powered clinical decision support systems (CDSSs) complicates the interpretation of how decisions are derived. Explainable AI (XAI) improves acceptance and trust with explanations, but the effectiveness of different methods remains uncertain. We compared the acceptance, trust, satisfaction and usability of various explanatory methods among clinicians. We also explored the factors associated with acceptance levels for each item using trust, satisfaction and usability score questionnaires. Surgeons and physicians (N = 63), who had prescribed blood products before surgery, made decisions before and after receiving one of three CDSS explanation methods, each comprising six vignettes, in a counterbalanced design. We found empirical evidence, which indicates that providing a clinical explanation enhances clinicians’ acceptance than presenting ‘results only’ or ‘results with SHapley Additive exPlanations (SHAP)’. Additionally, trust, satisfaction and usability were correlated with acceptance. This study suggests best practices for the strategic application of the XAI–CDSS in the medical field.

Mobile and tablet apps, ubiquitous and pervasive computing, wearable computing and domotics for health.

Background: Delirium frequently occurs among patients admitted to the intensive care unit (ICU). There is limited evidence to support interventions to treat or resolve delirium in patients who have already developed delirium. Therefore, the early recognition and prevention of delirium are important in the management of critically ill patients. Objective: This study aims to develop and validate a delirium prediction model within 24 hours of admission to the ICU using electronic health record data. The algorithm was named the Prediction of ICU Delirium (PRIDE). Methods: This is a retrospective cohort study performed at a tertiary referral hospital with 120 ICU beds. We only included patients who were 18 years or older at the time of admission and who stayed in the medical or surgical ICU. Patients were excluded if they lacked a Confusion Assessment Method for the ICU record from the day of ICU admission or if they had a positive Confusion Assessment Method for the ICU record at the time of ICU admission. The algorithm to predict delirium was developed using patient data from the first 2 years of the study period and validated using patient data from the last 6 months. Random forest (RF), Extreme Gradient Boosting (XGBoost), deep neural network (DNN), and logistic regression (LR) were used. The algorithms were externally validated using MIMIC-III data, and the algorithm with the largest area under the receiver operating characteristics (AUROC) curve in the external data set was named the PRIDE algorithm. Results: A total of 37,543 cases were collected. After patient exclusion, 12,409 remained as our study population, of which 3816 (30.8%) patients experienced delirium incidents during the study period. Based on the exclusion criteria, out of the 96,016 ICU admission cases in the MIMIC-III data set, 2061 cases were included, and 272 (13.2%) delirium incidents occurred. The average AUROCs and 95% CIs for internal validation were 0.916 (95% CI 0.916-0.916) for RF, 0.919 (95% CI 0.919-0.919) for XGBoost, 0.881 (95% CI 0.878-0.884) for DNN, and 0.875 (95% CI 0.875-0.875) for LR. Regarding the external validation, the best AUROC were 0.721 (95% CI 0.72-0.721) for RF, 0.697 (95% CI 0.695-0.699) for XGBoost, 0.655 (95% CI 0.654-0.657) for DNN, and 0.631 (95% CI 0.631-0.631) for LR. The Brier score of the RF model is 0.168, indicating that it is well-calibrated. Conclusions: A machine learning approach based on electronic health record data can be used to predict delirium within 24 hours of ICU admission. RF, XGBoost, DNN, and LR models were used, and they effectively predicted delirium. However, with the potential to advise ICU physicians and prevent ICU delirium, prospective studies are required to verify the algorithm’s performance.

Objectives: Although medical artificial intelligence (AI) systems that assist healthcare professionals in critical care settings are expected to improve healthcare, skepticism exists regarding whether their potential has been fully actualized. Therefore, we aimed to conduct a qualitative study with physicians and nurses to understand their needs, expectations, and concerns regarding medical AI; explore their expected responses to recommendations by medical AI that contradicted their judgments; and derive strategies to implement medical AI in practice successfully.Methods: Semi-structured interviews were conducted with 15 healthcare professionals working in the emergency room and intensive care unit in a tertiary teaching hospital in Seoul. The data were interpreted using summative content analysis. In total, 26 medical AI topics were extracted from the interviews. Eight were related to treatment recommendation, seven were related to diagnosis prediction, and seven were related to process improvement.Results: While the participants expressed expectations that medical AI could enhance their patients’ outcomes, increase work efficiency, and reduce hospital operating costs, they also mentioned concerns regarding distortions in the workflow, deskilling, alert fatigue, and unsophisticated algorithms. If medical AI decisions contradicted their judgment, most participants would consult other medical staff and thereafter reconsider their initial judgment.Conclusions: Healthcare professionals wanted to use medical AI in practice and emphasized that artificial intelligence systems should be trustworthy from the standpoint of healthcare professionals. They also highlighted the importance of alert fatigue management and the integration of AI systems into the workflow.

Background: Patient safety in the intensive care unit (ICU) is one of the most critical issues, and unplanned extubation (UE) is considered the most adverse event for patient safety. Prevention and early detection of such an event is an essential but difficult component of quality care. Objective: This study aimed to develop and validate prediction models for UE in ICU patients using machine learning. Methods: This study was conducted in an academic tertiary hospital in Seoul, Republic of Korea. The hospital had approximately 2000 inpatient beds and 120 ICU beds. As of January 2019, the hospital had approximately 9000 outpatients on a daily basis. The number of annual ICU admissions was approximately 10,000. We conducted a retrospective study between January 1, 2010, and December 31, 2018. A total of 6914 extubation cases were included. We developed a UE prediction model using machine learning algorithms, which included random forest (RF), logistic regression (LR), artificial neural network (ANN), and support vector machine (SVM). For evaluating the model’s performance, we used the area under the receiver operating characteristic curve (AUROC). The sensitivity, specificity, positive predictive value, negative predictive value, and F1 score were also determined for each model. For performance evaluation, we also used a calibration curve, the Brier score, and the integrated calibration index (ICI) to compare different models. The potential clinical usefulness of the best model at the best threshold was assessed through a net benefit approach using a decision curve. Results: Among the 6914 extubation cases, 248 underwent UE. In the UE group, there were more males than females, higher use of physical restraints, and fewer surgeries. The incidence of UE was higher during the night shift as compared to the planned extubation group. The rate of reintubation within 24 hours and hospital mortality were higher in the UE group. The UE prediction algorithm was developed, and the AUROC for RF was 0.787, for LR was 0.762, for ANN was 0.763, and for SVM was 0.740. Conclusions: We successfully developed and validated machine learning–based prediction models to predict UE in ICU patients using electronic health record data. The best AUROC was 0.787 and the sensitivity was 0.949, which was obtained using the RF algorithm. The RF model was well-calibrated, and the Brier score and ICI were 0.129 and 0.048, respectively. The proposed prediction model uses widely available variables to limit the additional workload on the clinician. Further, this evaluation suggests that the model holds potential for clinical usefulness.

Background: Physicians’ alert overriding behavior is considered to be the most important factor leading to failure of computerized provider order entry (CPOE) combined with a clinical decision support system (CDSS) in achieving its potential adverse drug events prevention effect. Previous studies on this subject have focused on specific diseases or alert types for well-defined targets and particular settings. The emergency department is an optimal environment to examine physicians’ alert overriding behaviors from a broad perspective because patients have a wider range of severity, and many receive interdisciplinary care in this environment. However, less than one-tenth of related studies have targeted this physician behavior in an emergency department setting. Objective: The aim of this study was to describe alert override patterns with a commercial medication CDSS in an academic emergency department. Methods: This study was conducted at a tertiary urban academic hospital in the emergency department with an annual census of 80,000 visits. We analyzed data on the patients who visited the emergency department for 18 months and the medical staff who treated them, including the prescription and CPOE alert log. We also performed descriptive analysis and logistic regression for assessing the risk factors for alert overrides. Results: During the study period, 611 physicians cared for 71,546 patients with 101,186 visits. The emergency department physicians encountered 13.75 alerts during every 100 orders entered. Of the total 102,887 alerts, almost two-thirds (65,616, 63.77%) were overridden. Univariate and multivariate logistic regression analyses identified 21 statistically significant risk factors for emergency department physicians’ alert override behavior. Conclusions: In this retrospective study, we described the alert override patterns with a medication CDSS in an academic emergency department. We found relatively low overrides and assessed their contributing factors, including physicians’ designation and specialty, patients’ severity and chief complaints, and alert and medication type.