Search

Abstract Associations between delirium and postoperative adverse events in cardiovascular surgery have been reported and the preoperative identification of high-risk patients of delirium is needed to implement focused interventions. We aimed to develop and validate machine learning models to predict post-cardiovascular surgery delirium. Patients aged ≥ 40 years who underwent cardiovascular surgery at a single hospital were prospectively enrolled. Preoperative and intraoperative factors were assessed. Each patient was evaluated for postoperative delirium 7 days after surgery. We developed machine learning models using the Bernoulli naive Bayes, Support vector machine, Random forest, Extra-trees, and XGBoost algorithms. Stratified fivefold cross-validation was performed for each developed model. Of the 87 patients, 24 (27.6%) developed postoperative delirium. Age, use of psychotropic drugs, cognitive function (Mini-Cog

IntroductionPostoperative delirium (POD) is common and life-threatening, however, with intensive interventions, a potentially preventable clinical syndrome. Although electroencephalography (EEG) is a promising biomarker of delirium, standard 20-leads EEG holds difficulties for screening usage in clinical practice.ObjectiveWe aimed to develop an accurate algorithm to predict POD using EEG data obtained from portable device.MethodsWe recruited 128 patients who underwent scheduled cardiovascular surgery. Cognitive function assessments were conducted, and portable EEG recordings were obtained prior to surgery.ResultsAmong the patients, 47 (36.7%) patients with POD were identified and they did not significantly differ from patients without POD in sex ratio, age, cognitive function, or treatment duration of intensive care unit. However, significant differences were observed in the preoperative EEG power spectrum densities at various frequencies, especially gamma activity, between patients with and without POD. POD was successfully predicted using preoperative EEG data with a machine learning algorithm, yielding accuracy of 86% and area under the receiver operating characteristic curve of 0.93.DiscussionThis study provides new insights into the objective and biological vulnerability to delirium. The developed algorithm can be applied in general hospitals without advanced equipment and expertise, thereby enabling the reduction of POD occurrences with intensive interventions for high-risk patients.

About 40 percent of chronic heart failure patients require re-hospitalization within a year because their cardiac function never fully recover even if appropriately treated. In Japan, heart failure notebooks have been used to prevent re-hospitalization for outpatients. However, the notebook has no significant efficacy for prevention. We developed a watch-type wearable device and software to evaluate third heart sound (S3), which is a specific cardiac murmur related to heart failure exacerbation. Feasibility study with the device was done in Osaka University Hospital from January to April 2022 to verify the sensitivity and the specificity for detecting S3 referring to cardiologist's labels. We collected 100 heart sound data from 11 patients. S3 was identified by the cardiologist in onefourth of the sounds. Our algorithm detected S3 with a sensitivity and a specificity of 80%. Social implementation of S3 detection technology of us may reduce heart failure recurrence in the future. [ABSTRACT FROM AUTHOR]