Inflammation-Responsive Hydrogel Accelerates Diabetic Wound Healing through Immunoregulation and Enhanced Angiogenesis.

Angiogenesis is a prominent component during the highly regulated process of wound healing. The application of exogenous vascular endothelial growth factor (VEGF) has shown considerable potential in facilitating angiogenesis. However, its effectiveness is often curtailed due to chronic inflammation and severe oxidative stress in diabetic wounds. Herein, an inflammation-responsive hydrogel incorporating Prussian blue nanoparticles (PBNPs) is designed to augment the angiogenic efficacy of VEGF. Specifically, the rapid release of PBNPs from the hydrogel under inflammatory conditions effectively alleviates the oxidative stress of the wound, therefore reprogramming the immune microenvironment to preserve the bioactivity of VEGF for enhanced angiogenesis. In vitro and in vivo studies reveal that the PBNPs and VEGF co-loaded hydrogel is biocompatible and possesses effective anti-inflammatory properties, thereby facilitating angiogenesis to accelerate the wound healing process in a type 2 diabetic mouse model.

Models predicting mortality risk of patients with burns to ≥ 50% of the total body surface.

BACKGROUND: Several models predicting mortality risk of burn patients have been proposed. However, models that consider all such patients may not well predict the mortality of patients with extensive burns. METHOD: This retrospective multicentre study recruited patients with extensive burns (≥ 50% of the total body surface area [TBSA]) treated in three hospitals of Eastern China from 1 January 2016 to 30 June 2022. The performances of six predictive models were assessed by drawing receiver operating characteristic (ROC) and calibration curves. Potential predictors were sought via "least absolute shrinkage and selection operator" regression. Multivariate logistic regression was employed to construct a predictive model for patients with burns to ≥ 50% of the TBSA. A nomogram was prepared and the performance thereof assessed by reference to the ROC, calibration, and decision curves. RESULT: A total of 465 eligible patients with burns to ≥ 50% TBSA were included, of whom 139 (29.9%) died. The FLAMES model exhibited the largest area under the ROC curve (AUC) (0.875), followed by the models of Zhou et al. (0.853) and the ABSI model (0.802). The calibration curve of the Zhou et al. model fitted well; those of the other models significantly overestimated the mortality risk. The new nomogram includes four variables: age, the %TBSA burned, the area of full-thickness burns, and blood lactate. The AUCs (training set 0.889; internal validation set 0.934; external validation set 0.890) and calibration curves showed that the nomogram exhibited an excellent discriminative capacity and that the predictions were very accurate. CONCLUSION: For patients with burns to ≥ 50％of the TBSA, the Zhou et al. and FLAMES models demonstrate relatively high predictive ability for mortality. The new nomogram is sensitive, specific, and accurate, and will aid rapid clinical decision-making.

Deep Neural Networks with Multistate Activation Functions.

We propose multistate activation functions (MSAFs) for deep neural networks (DNNs). These MSAFs are new kinds of activation functions which are capable of representing more than two states, including the N-order MSAFs and the symmetrical MSAF. DNNs with these MSAFs can be trained via conventional Stochastic Gradient Descent (SGD) as well as mean-normalised SGD. We also discuss how these MSAFs perform when used to resolve classification problems. Experimental results on the TIMIT corpus reveal that, on speech recognition tasks, DNNs with MSAFs perform better than the conventional DNNs, getting a relative improvement of 5.60% on phoneme error rates. Further experiments also reveal that mean-normalised SGD facilitates the training processes of DNNs with MSAFs, especially when being with large training sets. The models can also be directly trained without pretraining when the training set is sufficiently large, which results in a considerable relative improvement of 5.82% on word error rates.