Comparative visual analytics for assessing medical records with sequence embedding

Machine learning for data-driven diagnosis has been actively studied in medicine to provide better healthcare.Supporting analysis of a patient cohort similar to a patient under treatment is a key task for clinicians to make decisions with high confidence.However,such analysis is not straightforward due to the characteristics of medical records:high dimensionality,irregularity in time,and sparsity.To address this challenge,we introduce a method for similarity calculation of medical records.Our method employs event and sequence embeddings.While we use an autoencoder for the event embedding,we apply its variant with the self-attention mechanism for the sequence embedding.Moreover,in order to better handle the irregularity of data,we enhance the self-attention mechanism with consideration of different time intervals.We have developed a visual analytics system to support comparative studies of patient records.To make a comparison of sequences with different lengths easier,our system incorporates a sequence alignment method.Through its interactive interface,the user can quickly identify patients of interest and conveniently review both the temporal and multivariate aspects of the patient records.We demonstrate the effectiveness of our design and system with case studies using a real-world dataset from the neonatal intensive care unit of UC Davis.

Sodium variability is associated with increased mortality in severe burn injury

Background:Dysnatremias are associated with increased mortality in critically ill patients. Hypernatremia in burn patients is also associated with poor survival. Based on these findings, we hypothesized that high plasma sodium variability is a marker for increased mortality in severely burn-injured patients. Methods:We performed a retrospective review of adult burn patients with a burn injury of 15%total body surface area (TBSA) or greater from 2010 to 2014. All patients included in the study had at least three serum sodium levels checked during admission. We used multivariate logistic regression analysis to determine if hypernatremia, hyponatremia, or sodium variability independently increased the odds ratio (OR) for death. Results:Two hundred twelve patients met entry criteria. Mean age and%TBSA for the study was 45 ± 18 years and 32 ± 19%. Twenty-nine patients died for a mortality rate of 14%. Serum sodium was measured 10,310 times overall. The median number of serum sodium measurements per patient was 22. Non-survivors were older (59 ± 19 vs. 42 ± 16 years) and suffered from a more severe burn injury (50 ± 25%vs. 29 ± 16%TBSA). While mean sodium was significantly higher for non-survivors (138 ± 3 milliequivalents/liter (meq/l)) than for survivors (135 ± 2 meq/l), mean sodium levels remained within the laboratory reference range (135 to 145 meq/l) for both groups. Non-survivors had a significantly higher median number of hypernatremic (>145 meq/l) measurements (2 vs. 0). Coefficient of variation (CV) was significantly higher in non-survivors (2.85 ± 1.1) than survivors (2.0 ± 0.7). Adjusting for TBSA, age, ventilator days, and intensive care unit (ICU) stay, a higher CV of sodium measurements was associated with mortality (OR 5.8 (95%confidence interval (CI) 1.5 to 22)). Additionally, large variation in sodium ranges in the first 10 days of admission may be associated with increased mortality (OR 1.35 (95%CI 1.06 to1.7)). Conclusions:Increased variability in plasma sodium may be associated with death in severely burned patients.

Pediatric inhalation injury

Smoke inhalation injury can cause severe physiologic perturbations. In pediatric patients, these perturbations cause profound changes in cardiac and pulmonary physiology. In this review, we examine the pathology, early management options, ventilator strategy, and long-term outcomes in pediatric patients who have suffered a smoke inhalation injury.