We screened the electronic records of 2,799 patients admitted in Tongji Hospital from January 10th to February 18th, 2020. There were 375 discharged patients including 201 survivors. We built a prognostic prediction model based on XGBoost machine learning algorithm and then tested 29 patients (included 3 patients from other hospital) who were cleared after February 19th. Results:The mean age of the 375 patients was 58.83 years old with 58.7% of males. Fever was the most common initial symptom (49.9%), followed by cough (13.9%), fatigue (3.7%), and dyspnea (2.1%). Our model identified three key clinical features, i.e., lactic dehydrogenase (LDH), lymphocyte and High-sensitivity C-reactive protein (hs-CRP), from a pool of more than 300 features. The clinical route is simple to check and can precisely and quickly assess the risk of death. Therefore, it is of great clinical significance. : medRxiv preprint Conclusion:The three indices-based prognostic prediction model we built is able to predict the mortality risk, and present a clinical route to the recognition of critical cases from severe cases. It can help doctors with early identification and intervention, thus potentially reducing mortality.

The avoidance and suppression of runaway electron (RE) generation during disruptions is of great importance for the safe operation of tokamaks. Massive gas injection is used to suppress the generation of REs, but the poor gas mixing efficiency and extremely high density required to suppress RE generation make the full RE suppression unreliable. The magnetic perturbations provide an alternative RE suppression during disruptions. The use of mode penetration induced by resonant magnetic perturbations (RMPs) to suppress RE generation has been investigated on the J-TEXT tokamak. For a sufficiently long mode penetration duration, robust runaway suppression has been reached during the disruptions. The m/n=2/1 mode RMP with high amplitude excites large magnetic islands inside the plasma and leads to the large-scale destruction of magnetic surfaces during disruptions, which results in RE loss and runaway-free disruptions. The critical island width required for runaway suppression is estimated to be larger than 0.16 as the minor radius. This value might be slightly underestimated because of the misalignment between the electron cyclotron emission diagnostic and the island O-point. NIMROD simulations are used to investigate the effect of magnetic islands on RE generation during disruption, showing that the large magnetic islands have the ability to enhance RE seed loss during disruptions. RMP can excite large magnetic islands in the target plasma without tearing mode and might be a way to prevent RE generation during disruptions.

Disruption prediction is essential for the safe operation of a large scale tokamak. Existing disruption predictors based on machine learning techniques have good prediction performance, but all these methods need large training datasets including many disruptions to develop their successful prediction capability. Future machines are unlikely to provide enough disruption samples since these cause excessive machine damage and the prediction models used are difficult to extrapolate to a machines that the predictor was not trained on. In this paper, a disruption predictor based on a deep learning and anomaly detection technique has been developed. It regards the disruption as an anomaly, and can learn on non-disruptive shots only. The model is trained to extract the hidden features of various nondisruptive shots with a convolutional neural network and a long-shot term memory (LSTM) recurrent neural network. It will predict the future trend of selected diagnostics, then using the predicted future trend and the measured signal to calculate an outlier factor to determine if a disruption is coming. It was tested with J-TEXT discharges in flat top phase and can demonstrate comparable performance to current machine learning disruption prediction techniques, without requiring a disruption data set. This could be applied to future tokamaks and reduce the dependency on disruptive experiments.

Urinary sodium levels are reported to be associated with blood pressure in clinical trials and epidemiology studies. Nevertheless, the public health message of reducing sodium intake in free-living community populations remains under debate. Based on an ongoing prospective study initiated in 2012 with a community-based design in Xinjiang, China, 1668 adults (⩾30 years old) were assessed in the current study for associations between urinary sodium and blood pressure and hypertension in a free-living population of Kazakh people. After excluding 223 people on antihypertensive medication, 1445 participants were analyzed. Second urine samples after waking were used to estimate 24-h urinary sodium excretion, which is a marker for sodium intake. Following analyses, we found that the distribution of systolic and diastolic blood pressures moved upward with increasing quartiles of urinary sodium. After adjusting for age, differences in median systolic blood pressure were 8.5 mm Hg for men and 8.0 mm Hg for women between the top and bottom urinary sodium quartiles, and differences for diastolic blood pressure were 4.7 mm Hg for men and 4.3 mm Hg for women. A significant increased risk for hypertension was observed for the top quartile of urinary sodium after adjusting for age, body mass index, smoking, alcohol consumption, fruit and vegetable consumption, with corresponding odds ratios being 1.61 (95% confidence interval (CI): 1.02-2.54) for men and 1.92 (95% CI: 1.13-3.27) for women. Improving education about reducing salt intake is of particular public importance to reduce blood pressure and the risk for hypertension among the Kazakh people.

Deep-tissue three-dimensional optical imaging of live mammals in vivo with high spatiotemporal resolution in non-invasive manners has been challenging due to light scattering. Here, we developed near-infrared (NIR) light sheet microscopy (LSM) with optical excitation and emission wavelengths up to ~ 1320 nm and ~ 1700 nm respectively, far into the NIR-II (1000-1700 nm) region for 3D optical sectioning through live tissues. Suppressed scattering of both excitation and emission photons allowed one-photon optical sectioning at ~ 2 mm depth in highly scattering brain tissues. NIR-II LSM enabled non-invasive in vivo imaging of live mice, revealing never-before-seen dynamic processes such as highly abnormal tumor microcirculation, and 3D molecular imaging of an important immune checkpoint protein, programmed-death ligand 1 (PD-L1) receptors at the single cell scale in tumors. In vivo two-color near-infrared light sheet sectioning enabled simultaneous volumetric imaging of tumor vasculatures and PD-L1 proteins in live mammals.Optical imaging of biological systems capable of high spatiotemporal resolution in vivo and ex vivo has revolutionized biology and medicine for visualization and understanding of structures, functions and dynamic processes at the cellular and even molecular scale 1,2 . To circumvent light scattering by tissues, in vivo 3D imaging by nonlinear two-photon fluorescence microscopy (670-1070 nm excitation) 3-5 or three-photon microscopy (1300-1700 nm excitation) 6-9 has reached penetration depths ~ 0.7-1.5 mm, benefiting from increased scattering mean free path of the nearinfrared (NIR) excitation employed 6 . Light sheet microscopy (LSM) uses orthogonally arranged planar illumination and wide-field detection, capable of high speed 3D optical sectioning, low photo-damage 10,11 and volumetrically imaging/tracking with subcellular resolution 2 . Currently the excitation and emission of LSM are mostly in the visible except for two photon excitation in the NIR at ~ 940 nm 12 or three photon excitation at 1000 nm 13 . Light scattering has limited LSM to imaging small transparent animals, organisms (zebrafish larvae, drosophila larvae, Medaka embryo, C. elegans, etc.), mammalian tissue samples after chemical clearing 11,14,15 , and mouse brain at a depth of ~ 200 μm after craniotomy 16 .