In-Hospital Cancer Mortality Prediction by Multimodal Learning of Non-English Clinical Texts.

Predicting important outcomes in patients with complex medical conditions using multimodal electronic medical records remains challenge. We trained a machine learning model to predict the inpatient prognosis of cancer patients using EMR data with Japanese clinical text records, which has been considered difficult due to its high context. We confirmed high accuracy of the mortality prediction model using clinical text in addition to other clinical data, suggesting applicability of this method to cancer.

Live-cell single-molecule labeling and analysis of myosin motors with quantum dots.

Quantum dots (QDs) are a powerful tool for quantitatively analyzing dynamic cellular processes by single-particle tracking. However, tracking of intracellular molecules with QDs is limited by their inability to penetrate the plasma membrane and bind to specific molecules of interest. Although several techniques for overcoming these problems have been proposed, they are either complicated or inconvenient. To address this issue, in this study, we developed a simple, convenient, and nontoxic method for labeling intracellular molecules in cells using HaloTag technology and electroporation. We labeled intracellular myosin motors with this approach and tracked their movement within cells. By simultaneously imaging myosin movement and F-actin architecture, we observed that F-actin serves not only as a rail but also as a barrier for myosin movement. We analyzed the effect of insulin on the movement of several myosin motors, which have been suggested to regulate intracellular trafficking of the insulin-responsive glucose transporter GLUT4, but found no significant enhancement in myosin motor motility as a result of insulin treatment. Our approach expands the repertoire of proteins for which intracellular dynamics can be analyzed at the single-molecule level.