Machine Learning Model for Chest Radiographs: Using Local Data to Enhance Performance.

PURPOSE: To develop and assess the performance of a machine learning model which screens chest radiographs for 14 labels, and to determine whether fine-tuning the model on local data improves its performance. Generalizability at different institutions has been an obstacle to machine learning model implementation. We hypothesized that the performance of a model trained on an open-source dataset will improve at our local institution after being fine-tuned on local data. METHODS: In this retrospective, institutional review board approved study, an ensemble of neural networks was trained on open-source datasets of chest radiographs for the detection of 14 labels. This model was then fine-tuned using 4510 local radiograph studies, using radiologists' reports as the gold standard to evaluate model performance. Both the open-source and fine-tuned models' accuracy were tested on 802 local radiographs. Receiver-operator characteristic curves were calculated, and statistical analysis was completed using DeLong's method and Wilcoxon signed-rank test. RESULTS: The fine-tuned model identified 12 of 14 pathology labels with area under the curves greater than .75. After fine-tuning with local data, the model performed statistically significantly better overall, and specifically in detecting six pathology labels (P < .01). CONCLUSIONS: A machine learning model able to accurately detect 14 labels simultaneously on chest radiographs was developed using open-source data, and its performance was improved after fine-tuning on local site data. This simple method of fine-tuning existing models on local data could improve the generalizability of existing models across different institutions to further improve their local performance.

Empiric antibiotics for peripartum bacteremia: A chart review from a quaternary Canadian centre.

OBJECTIVE: To evaluate the effectiveness of empiric antibiotic protocols for peripartum bacteremia at a quaternary institution by describing incidence, microbial epidemiology, clinical source of infection, susceptibility patterns, and maternal and neonatal outcomes. METHODS: Retrospective chart review of peripartum patients with positive blood cultures between 2010 and 2018. RESULTS: The incidence of peripartum bacteremia was 0.3%. The most cultured organisms were Escherichia coli (51, 26.7%), Streptococcus spp. (52, 27.2%), and anaerobic spp. (35, 18.3%). Of the E. coli cases, 54.9% (28), 19.6% (10), and 19.6% (10) were resistant to ampicillin, first- and third-generation cephalosporins, respectively. Clinical sources of infection included intra-amniotic infection/endometritis (115, 67.6%), upper and/or lower urinary tract infection (23, 13.5%), and soft tissue infection (8, 4.7%). Appropriate empiric antibiotics were prescribed in 137 (83.0%) cases. There were 7 ICU admissions (4.2%), 18 pregnancy losses (9.9%), 9 neonatal deaths (5.5%), and 6 cases of neonatal bacteremia (3.7%). CONCLUSION: Peripartum bacteremia remains uncommon but associated with maternal morbidity and neonatal morbidity and mortality. Current empiric antimicrobial protocols at our site remain appropriate, but continuous monitoring of antimicrobial resistance patterns is critical given the presence of pathogens resistant to first-line antibiotics.