Learning efficient representations for concepts has been proven to be an important basis for many applications such as machine translation or document classification. Proper representations of medical concepts such as diagnosis, medication, procedure codes and visits will have broad applications in healthcare analytics. However, in Electronic Health Records (EHR) the visit sequences of patients include multiple concepts (diagnosis, procedure, and medication codes) per visit. This structure provides two types of relational information, namely sequential order of visits and co-occurrence of the codes within each visit. In this work, we propose Med2Vec, which not only learns distributed representations for both medical codes and visits from a large EHR dataset with over 3 million visits, but also allows us to interpret the learned representations confirmed positively by clinical experts. In the experiments, Med2Vec displays significant improvement in key medical applications compared to popular baselines such as Skipgram, GloVe and stacked autoencoder, while providing clinically meaningful interpretation.

OBJECTIVE:To determine the rate of return visits to pediatric emergency departments (EDs) and identify patient-and visit-level factors associated with return visits and hospitalization upon return. DESIGN AND SETTING:Retrospective cohort study of visits to 23 pediatric EDs in 2012 using data from the Pediatric Health Information System. PARTICIPANTS:Patients <18 years old discharged following an ED visit. MEASURES:The primary outcomes were the rate of return visits within 72 hours of discharge from the ED and of return visits within 72 hours resulting in hospitalization. Results: 1,415,721 of the 1,610,201 ED visits to study hospitals resulted in discharge. Of the discharges, 47,294 patients (3.3%) had a return visit. Of these revisits, 9295 (19.7%) resulted in hospitalization. In multivariate analyses, the odds of having a revisit were higher for patients with a chronic condition (odds ratio [OR]

No abstract

Object The rate of readmission after CSF shunt surgery is significant and has caught the attention of purchasers of health care. However, a detailed description of clinical scenarios that lead to readmissions and reoperations after index shunt surgery is lacking in the medical literature. Methods This study included 1755 shunt revision and insertion surgeries that were performed at a single institution between May 1, 2009, and April 30, 2013. Demographic, socioeconomic, and clinical characteristics were prospectively collected in the administrative, business, and operating room databases. Clinical events within the 30 days following discharge were reviewed and analyzed. Two events of interest, Emergency Department (ED) utilization and reoperation, were further analyzed for risk factor associations by using multivariate logistic regression. Results There were 290 readmissions within 30 days of discharge (16.5%). Admission sources included ED (n = 216), hospital transfers (n = 23), and others. Of the 290 readmissions, 184 were associated with an operation, but only 165 of these were performed by the neurosurgical service. These included surgeries for shunt occlusion and externalization (n = 150), wound revision (n = 7), and other neurosurgical procedures that were not shunt related (n = 8). The remaining readmissions (n = 106) were not associated with an operation, and only 59 patients were admitted for issues related to the index shunt surgery. When return to the ED was the dependent variable in a multivariate regression model, patients who returned to the ED were more likely to be from the Atlanta metropolitan area and to be either uninsured or insured with public assistance. When reoperation was the dependent variable, patients whose surgery started after 3 p.m. were more likely to undergo subsequent CSF shunt revision surgery on readmission. Conclusions Of the readmissions within 30 days of shunt surgery, 74.5% were related to the index shunt surgery. Whether and to what extent these readmissions are preventable continues to be controversial. Further study is needed to identify modifiable risk factors that may eventually improve patient care.