Deep transfer learning with multimodal embedding to tackle cold-start and sparsity issues in recommendation system.

Recommender systems (RSs) have become increasingly vital in the modern information era and connected economy. They play a key role in business operations by generating personalized suggestions and minimizing information overload. However, the performance of traditional RSs is limited by data sparseness and cold-start issues. Though deep learning-based recommender systems (DLRSs) are very popular, they underperform when considering rating matrices with sparse entries. Despite their performance improvements, DLRSs also suffer from data sparsity, cold start, serendipity, and generalizability issues. We propose a multistage model that uses multimodal data embedding and deep transfer learning for effective and personalized product recommendations, and is designed to overcome data sparsity and cold-start issues. The proposed model includes two phases. In the first-offline-phase, a deep learning technique is implemented to learn hidden features from a large image dataset (targeting new item cold start), and a multimodal data embedding is used to produce dense user feature and item feature vectors (targeting user cold start). This phase produces three different similarity matrices that are used as inputs for the second-online-phase to generate a list of top-n relevant items for a target user. We analyzed the accuracy and effectiveness of the proposed model against the existing baseline RSs using a Brazilian E-commerce dataset. The results show that our model scored 0.5882 for MAE and 0.4011 for RMSE which is lower than baseline RSs which indicates that the model achieved an improved accuracy and was able to minimize the typical cold start and data sparseness issues during the recommendation process.

Is sarcopenia and sarcopenic obesity associated with clinical and pathological outcomes in patients undergoing radical nephroureterectomy?

PURPOSE: To investigate the association between sarcopenia and sarcopenic obesity on clinical, perioperative, and oncologic outcomes in patients with upper-tract urothelial carcinoma (UTUC) undergoing radical nephroureterectomy (RNU). METHODS: Retrospective review of our institutional UTUC database was performed to identify all patients who underwent radical nephroureterectomy from 2002-2016. Skeletal Muscle Index (SMI) was measured at the L3 vertebral level and standardized according to patient height (cm2/m2). Sarcopenia was defined as<55cm2/m2 for men and<39cm2/m2 for women. Sarcopenic obesity was also assessed in patients with BMI>30kg/m2. Unadjusted logistic regression and Wilcoxon rank sum tests examined the relationship between sarcopenia and variables. RESULTS: A total of 100 patients (66 men and 34 women) with a mean age of 68 years, BMI of 30, Charlson comorbidity index of 4.0, tumor size of 3.5, and SMI of 50.8cm2/m2 were included. Furthermore, 42 patients (42%) were sarcopenic, and 18 patients (18%) had sarcopenic obesity. Median EBL was 150ml, OR duration was 322 minutes, and length of stay was 5.0 days. Sarcopenia was associated with several clinical factors including decreasing BMI, male sex, and coronary artery disease, albeit without association with any perioperative or oncologic outcomes. Sarcopenic obesity was similarly associated with several clinical variables including male sex, diabetes mellitus, hyperlipidemia, as well as increased EBL (P = 0.047) and non-bladder cancer disease relapse (P = 0.049). CONCLUSIONS: This contemporary cohort of patients undergoing RNU highlights the association of nonmodifiable risk factors with sarcopenia and disease relapse with sarcopenic obesity. Larger studies are necessary to further validate these observations.

In vitro antimicrobial susceptibility of bacterial enteropathogens isolated from international travelers to Mexico, Guatemala, and India from 2006 to 2008.

The incidence rates of travelers' diarrhea (TD) have remained high for the last 50 years. More recently, there have been increasing recommendations for self-initiated therapy and use of prophylactic drugs for TD. We last examined the in vitro susceptibilities of commonly used antibiotics against TD pathogens in 1997. We now examine 456 enteropathogens isolated from adult travelers to Mexico, India, and Guatemala with diarrhea acquired between 2006 and 2008 to determine changes in susceptibility against 10 different antimicrobials by the agar dilution method. Traditional antibiotics, such as ampicillin, trimethoprim-sulfamethoxazole, and doxycycline, continue to show high levels of resistance. Current first-line antibiotic agents, including fluoroquinolones and azithromycin, showed significantly higher MICs than in our earlier study, and MIC(90) levels were above the Clinical and Laboratory Standards Institute cutoffs for resistance. There were significant geographical differences in resistance patterns when Central America was compared with India. Entertoxigenic Escherichia coli (ETEC) isolates showed increased resistance to ciprofloxacin (P = 0.023) and levofloxacin (P = 0.0078) in India compared with Central America. Enteroaggregative E. coli (EAEC) isolates from Central America showed increased resistance to nearly all of the antibiotics tested. Compared to MICs of isolates 10 years prior, there were 4- to 10-fold increases in MIC(90) values for ceftriaxone, ciprofloxacin, levofloxacin, and azithromycin for both ETEC and EAEC. There were no significant changes in rifaximin MICs. Rising MICs over time imply the need for continuous surveillance of susceptibility patterns worldwide and geographically specific recommendations in TD therapy.