E-learning and practical performance in musculoskeletal ultrasound: a multicentre randomized study.

OBJECTIVES: To examine the effect of pre-course e-learning on residents' practical performance in musculoskeletal ultrasound (MSUS). METHODS: This was a multicentre, randomized controlled study following the Consolidated Standards of Reporting Trials (CONSORT) statement. Residents with no or little MSUS experience were randomized to either an e-learning group or a traditional group. One week before a 2-day face-to-face MSUS course, the e-learning group received access to an interactive platform consisting of online lectures, assignments, and practical instruction videos aligned with the content of the course. The traditional group only received standard pre-course information (program, venue, and time). All participants performed a pre- and post-course practical MSUS examination and were assessed by two individual raters, blinded to the group allocation, using the validated Objective Structured Assessment of Ultrasound Skills (OSAUS) tool. RESULTS: Twenty-eight participants completed the study. There were no statistically significant differences in the pre- or post-course practical MSUS performance between the e-learning group and the traditional group; the mean pre-course OSAUS score (s.d.) in the -learning group was 5.4 (3.7) compared with 5.2 (2.4) in the traditional group (P = 0.8), whereas the post-course OSAUS score in the e-learning group was 11.1 (2.8) compared with 10.9 (2.4) in the traditional group (P = 0.8). There was a significant difference between the mean pre- and post-course scores (5.74 points, P < 0.001). The OSAUS assessment tool demonstrated good inter-rater reliability (intra-class correlation = 0.84). CONCLUSION: We found no significant impact of pre-course e-learning on novices' acquisition of practical MSUS skills. Hands-on training is of the utmost importance and improves MSUS performance significantly. The OSAUS assessment tool is an applicable tool with high interrater reliability. TRIAL REGISTRATION: https://clinicaltrials.gov/ NCT04959162.

Deep learning detects and visualizes bleeding events in electronic health records.

BACKGROUND: Bleeding is associated with a significantly increased morbidity and mortality. Bleeding events are often described in the unstructured text of electronic health records, which makes them difficult to identify by manual inspection. OBJECTIVES: To develop a deep learning model that detects and visualizes bleeding events in electronic health records. PATIENTS/METHODS: Three hundred electronic health records with International Classification of Diseases, Tenth Revision diagnosis codes for bleeding or leukemia were extracted. Each sentence in the electronic health record was annotated as positive or negative for bleeding. The annotated sentences were used to develop a deep learning model that detects bleeding at sentence and note level. RESULTS: On a balanced test set of 1178   sentences, the best-performing deep learning model achieved a sensitivity of 0.90, specificity of 0.90, and negative predictive value of 0.90. On a test set consisting of 700 notes, of which 49 were positive for bleeding, the model achieved a note-level sensitivity of 1.00, specificity of 0.52, and negative predictive value of 1.00. By using a sentence-level model on a note level, the model can explain its predictions by visualizing the exact sentence in a note that contains information regarding bleeding. Moreover, we found that the model performed consistently well across different types of bleedings. CONCLUSIONS: A deep learning model can be used to detect and visualize bleeding events in the free text of electronic health records. The deep learning model can thus facilitate systematic assessment of bleeding risk, and thereby optimize patient care and safety.