An evaluation of intravenous medication preparation times before and after implementation of first-party digital image capture functionality.

PURPOSE: The intravenous (IV) medication compounding workflow has long been associated with preventable medication errors. This has led to the development of technologies designed to enhance the safety of IV compounding workflows. Digital image capture is a component of this technology about which there is relatively limited published literature. This study evaluates image capture implemented within an electronic health record's existing first-party IV workflow solution. METHODS: A retrospective case-control study was conducted to measure IV preparation times before and after digital imaging implementation. Preparations during 3 periods (preimplementation, ≤1 month post implementation, and >1 month post implementation) were matched for 5 variables. A less stringent analysis with matching for 2 variables, as well as an unmatched analysis, were performed post hoc. An employee survey assessed satisfaction with the digital imaging workflow, and revised orders were reviewed to identify new problems introduced by image capture. RESULTS: A total of 134,969 IV dispenses were available for analysis. Median preparation time in the preimplementation and >1 month post implementation cohorts was unchanged in the 5-variable matched analysis (6.87 minutes vs 6.58 minutes, P = 0.14) and increased in the 2-variable matched analysis (6.98 minutes vs 7.35 minutes, P < 0.001) and unmatched analysis (6.55 minutes vs 8.02 minutes, P < 0.001). A large majority of survey respondents (92%) felt that image capture improved patient safety. Of the 105 postimplementation preparations identified as requiring revisions by the checking pharmacist, 24 (22.9%) required revisions directly related to camera functionality. CONCLUSION: Implementation of digital image capture likely increased preparation times. Most IV room staff felt that image capture increased preparation times and were satisfied with how the technology improved patient safety. Image capture introduced camera-specific issues that led to preparations requiring revisions.

Neural matrix factorization++ based recommendation system.

In recent years, Recommender System (RS) research work has covered a wide variety of Artificial Intelligence techniques, ranging from traditional Matrix Factorization (MF) to complex Deep Neural Networks (DNN). Traditional Collaborative Filtering (CF) recommendation methods such as MF, have limited learning capabilities as it only considers the linear combination between user and item vectors. For learning non-linear relationships, methods like Neural Collaborative Filtering (NCF) incorporate DNN into CF methods. Though, CF methods still suffer from cold start and data sparsity. This paper proposes an improved hybrid-based RS, namely Neural Matrix Factorization++ (NeuMF++), for effectively learning user and item features to improve recommendation accuracy and alleviate cold start and data sparsity. NeuMF++ is proposed by incorporating effective latent representation into NeuMF via Stacked Denoising Autoencoders (SDAE). NeuMF++ can also be seen as the fusion of GMF++ and MLP++. NeuMF is an NCF framework which associates with GMF (Generalized Matrix Factorization) and MLP (Multilayer Perceptrons). NeuMF achieves state-of-the-art results due to the integration of GMF linearity and MLP non-linearity. Concurrently, incorporating latent representations has shown tremendous improvement in GMF and MLP, which result in GMF++ and MLP++. Latent representation obtained through the SDAEs' latent space allows NeuMF++ to effectively learn user and item features, significantly enhancing its learning capability. However, sharing feature extractions among GMF++ and MLP++ in NeuMF++ might hinder its performance. Hence, allowing GMF++ and MLP++ to learn separate features provides more flexibility and greatly improves its performance. Experiments performed on a real-world dataset have demonstrated that NeuMF++ achieves an outstanding result of a test root-mean-square error of 0.8681. In future work, we can extend NeuMF++ by introducing other auxiliary information like text or images. Different neural network building blocks can also be integrated into NeuMF++ to form a more robust recommendation model.