The comparative ability of commonly used disease severity scores to predict death or a requirement for ICU care in patients hospitalised with possible sepsis in Yangon, Myanmar.

OBJECTIVES: To determine the comparative prognostic utility of commonly used disease prediction scores in adults with presumed community-acquired sepsis in a resource-limited tropical setting. METHODS: This prospective, observational study was performed on the medical ward of a tertiary-referral hospital in Yangon, Myanmar. The ability of the National Early Warning Score 2 (NEWS2), quick NEWS (qNEWS), quick Sequential Organ Failure Assessment (qSOFA) score, Universal Vital Assessment (UVA) and Sequential Organ Failure Assessment (SOFA) scores to predict a complicated inpatient course (death or requirement for intensive care unit (ICU) support) in patients with two or more systemic inflammatory response syndrome criteria was determined. RESULTS: Among the 509 patients, 30 (6%) were HIV-seropositive. The most commonly confirmed diagnoses were tuberculosis (30/509, 5.9%) and measles (26/509, 5.1%). Overall, 75/509 (14.7%) died or required ICU support. All the scores except the qSOFA score, which was inferior, had a similar ability to predict a complicated inpatient course. CONCLUSIONS: In this resource-limited tropical setting, disease severity scores calculated at presentation using only vital signs-such as the NEWS2 score-identified high-risk sepsis patient as well as the SOFA score, which is calculated at 24 h and which also requires laboratory data. Use of these simple clinical scores can be used to facilitate recognition of the high-risk patient and to optimise the use of finite resources.

Upper gastrointestinal anatomy detection with multi-task convolutional neural networks.

Esophagogastroduodenoscopy (EGD) has been widely applied for gastrointestinal (GI) examinations. However, there is a lack of mature technology to evaluate the quality of the EGD inspection process. In this Letter, the authors design a multi-task anatomy detection convolutional neural network (MT-AD-CNN) to evaluate the EGD inspection quality by combining the detection task of the upper digestive tract with ten anatomical structures and the classification task of informative video frames. The authors' model is able to eliminate non-informative frames of the gastroscopic videos and detect the anatomies in real time. Specifically, a sub-branch is added to the detection network to classify NBI images, informative and non-informative images. By doing so, the detected box will be only displayed on the informative frames, which can reduce the false-positive rate. They can determine the video frames on which each anatomical location is effectively examined, so that they can analyse the diagnosis quality. Their method reaches the performance of 93.74% mean average precision for the detection task and 98.77% accuracy for the classification task. Their model can reflect the detailed circumstance of the gastroscopy examination process, which shows application potential in improving the quality of examinations.