Leveraging explainable artificial intelligence to optimize clinical decision support.

OBJECTIVE: To develop and evaluate a data-driven process to generate suggestions for improving alert criteria using explainable artificial intelligence (XAI) approaches. METHODS: We extracted data on alerts generated from January 1, 2019 to December 31, 2020, at Vanderbilt University Medical Center. We developed machine learning models to predict user responses to alerts. We applied XAI techniques to generate global explanations and local explanations. We evaluated the generated suggestions by comparing with alert's historical change logs and stakeholder interviews. Suggestions that either matched (or partially matched) changes already made to the alert or were considered clinically correct were classified as helpful. RESULTS: The final dataset included 2 991 823 firings with 2689 features. Among the 5 machine learning models, the LightGBM model achieved the highest Area under the ROC Curve: 0.919 [0.918, 0.920]. We identified 96 helpful suggestions. A total of 278 807 firings (9.3%) could have been eliminated. Some of the suggestions also revealed workflow and education issues. CONCLUSION: We developed a data-driven process to generate suggestions for improving alert criteria using XAI techniques. Our approach could identify improvements regarding clinical decision support (CDS) that might be overlooked or delayed in manual reviews. It also unveils a secondary purpose for the XAI: to improve quality by discovering scenarios where CDS alerts are not accepted due to workflow, education, or staffing issues.

Electroconvulsive therapy in a cochlear implant patient.

OBJECTIVE: To report the first use of electroconvulsive therapy (ECT) in a patient with a cochlear implant (CI). STUDY DESIGN: Clinical Capsule Report. SETTING: University hospital. PATIENT: : A 17-year-old boy who underwent Nucleus 22 cochlear implantation in 1995 presented with delirious mania in 2009. Aggressive pharmacologic management was ineffective, and ECT was recommended due to the potentially lethal nature of his psychiatric illness. INTERVENTIONS: After careful consideration by a multidisciplinary team, unilateral ECT on the side opposite the CI without removal of the device was recommended. Electroconvulsive therapy was performed on hospital Days 20 and 21. Integrity testing of the CI was performed on hospital Day 38. MAIN OUTCOME MEASURES: Subjective and objective assessment of cochlear implant functioning and response to ECT. RESULTS: Electroconvulsive therapy was well tolerated and contributed to alleviation of presenting symptoms. The patient used his CI without subjective degradation of performance. Integrity testing of the CI after ECT confirmed proper functioning of the device. CONCLUSION: This is the first report of ECT in a patient with CI. Unilateral ECT was performed contralateral to the CI without subjective or objective decline in performance. This Clinical Report motivates further study regarding the use of ECT in CI patients.