Development and Validation of an Explainable Deep Learning Model to Predict In-Hospital Mortality for Patients With Acute Myocardial Infarction: Algorithm Development and Validation Study.

BACKGROUND: Acute myocardial infarction (AMI) is one of the most severe cardiovascular diseases and is associated with a high risk of in-hospital mortality. However, the current deep learning models for in-hospital mortality prediction lack interpretability. OBJECTIVE: This study aims to establish an explainable deep learning model to provide individualized in-hospital mortality prediction and risk factor assessment for patients with AMI. METHODS: In this retrospective multicenter study, we used data for consecutive patients hospitalized with AMI from the Chongqing University Central Hospital between July 2016 and December 2022 and the Electronic Intensive Care Unit Collaborative Research Database. These patients were randomly divided into training (7668/10,955, 70%) and internal test (3287/10,955, 30%) data sets. In addition, data of patients with AMI from the Medical Information Mart for Intensive Care database were used for external validation. Deep learning models were used to predict in-hospital mortality in patients with AMI, and they were compared with linear and tree-based models. The Shapley Additive Explanations method was used to explain the model with the highest area under the receiver operating characteristic curve in both the internal test and external validation data sets to quantify and visualize the features that drive predictions. RESULTS: A total of 10,955 patients with AMI who were admitted to Chongqing University Central Hospital or included in the Electronic Intensive Care Unit Collaborative Research Database were randomly divided into a training data set of 7668 (70%) patients and an internal test data set of 3287 (30%) patients. A total of 9355 patients from the Medical Information Mart for Intensive Care database were included for independent external validation. In-hospital mortality occurred in 8.74% (670/7668), 8.73% (287/3287), and 9.12% (853/9355) of the patients in the training, internal test, and external validation cohorts, respectively. The Self-Attention and Intersample Attention Transformer model performed best in both the internal test data set and the external validation data set among the 9 prediction models, with the highest area under the receiver operating characteristic curve of 0.86 (95% CI 0.84-0.88) and 0.85 (95% CI 0.84-0.87), respectively. Older age, high heart rate, and low body temperature were the 3 most important predictors of increased mortality, according to the explanations of the Self-Attention and Intersample Attention Transformer model. CONCLUSIONS: The explainable deep learning model that we developed could provide estimates of mortality and visual contribution of the features to the prediction for a patient with AMI. The explanations suggested that older age, unstable vital signs, and metabolic disorders may increase the risk of mortality in patients with AMI.

LRPL-VIO: A Lightweight and Robust Visual-Inertial Odometry with Point and Line Features.

Visual-inertial odometry (VIO) algorithms, fusing various features such as points and lines, are able to improve their performance in challenging scenes while the running time severely increases. In this paper, we propose a novel lightweight point-line visual-inertial odometry algorithm to solve this problem, called LRPL-VIO. Firstly, a fast line matching method is proposed based on the assumption that the photometric values of endpoints and midpoints are invariant between consecutive frames, which greatly reduces the time consumption of the front end. Then, an efficient filter-based state estimation framework is designed to finish information fusion (point, line, and inertial). Fresh measurements of line features with good tracking quality are selected for state estimation using a unique feature selection scheme, which improves the efficiency of the proposed algorithm. Finally, validation experiments are conducted on public datasets and in real-world tests to evaluate the performance of LRPL-VIO and the results show that we outperform other state-of-the-art algorithms especially in terms of speed and robustness.

Physiological Investigation and Transcriptome Analysis Reveals the Mechanisms of Setaria italica's Yield Formation under Heat Stress.

Setaria italica is an important crop in China that plays a vital role in the Chinese dietary structure. In the last several decades, high temperature has become the most severe climate issue in the world, which causes great harm to the yield and quality formation of millet. In this study, two main cultivated varieties (ZG2 and AI88) were used to explore the photosynthesis and yield index of the whole plant under heat stress. Results implied that photosynthesis was not inhibited during the heat stress, and that the imbalance in sugar transport between different tissues may be the main factor that affects yield formation. In addition, the expression levels of seven SiSUT and twenty-four SiSWEET members were explored. Sugar transporters were heavily affected during the heat stress. The expression of SiSWEET13a was inhibited by heat stress in the stems, which may play a vital role in sugar transport between different tissues. These results provide new insights into the yield formation of crops under heat stress, which will provide guidance to crop breeding and cultivation.

Study on the interaction and gel properties of pork myofibrillar protein with konjac polysaccharides.

BACKGROUND: Natural myofibrillar protein (MP) is sensitive to changes in the microenvironment, such as pH and ionic strength, and therefore can adversely affect the final quality of meat products. The aim of this study was to modify natural MP as well as to improve its functional properties. Therefore, the quality improvement effect of konjac polysaccharides with different concentrations (0, 1.5, 3, 4.5 and 6 g kg-1 protein) on MP gels was investigated. RESULTS: With a concentration of konjac polysaccharides of 6 g kg-1 protein, the composite gel obtained exhibited a significant improvement of water binding (water holding capacity increased by 7.71%) and textural performance (strength increased from 29.12 to 37.55 N mm, an increase of 8.43 N mm). Meanwhile, konjac polysaccharides could help to form more disulfide bonds and non-disulfide covalent bonds, which enhanced the crosslinking of MP and maintained the MP gel network structure. Then, with the preservation of α-helix structure (a significant increase of 8.11%), slower protein aggregation and formation of small aggregates, this supported the formation of a fine and homogeneous network structure and allowed a reduction in water mobility. CONCLUSION: During the heating process, konjac polysaccharides could absorb the surrounding water and fill the gel system, which resulted in an increase in the water content of the gel network and enhanced the gel-forming ability of the gel. Meanwhile, konjac polysaccharides might inhibit irregular aggregation of proteins and promote the formation of small aggregates, which in turn form a homogeneous and continuous gel matrix by orderly arrangement. © 2023 Society of Chemical Industry.

The rice phosphate transporter OsPHT1;7 plays a dual role in phosphorus redistribution and anther development.

Inorganic phosphate (Pi) is the predominant form of phosphorus (P) readily accessible to plants, and Pi Transporter 1 (PHT1) genes are the major contributors to root Pi uptake. However, the mechanisms underlying the transport and recycling of Pi within plants, which are vital for optimizing P use efficiency, remain elusive. Here, we characterized a functionally unknown rice (Oryza sativa) PHT1 member barely expressed in roots, OsPHT1;7. Yeast complementation and Xenopus laevis oocyte assay demonstrated that OsPHT1;7 could mediate Pi transport. Reverse-transcription quantitative polymerase chain reaction and histochemical analyses showed that OsPHT1;7 was preferentially expressed in source leaves and nodes. A further fine-localization analysis by immunostaining showed that OsPHT1;7 expression was restricted in the vascular bundle (VB) sheath and phloem of source leaves as well as in the phloem of regular/diffuse- and enlarged-VBs of nodes. In accordance with this expression pattern, mutation of OsPHT1;7 led to increased and decreased P distribution in source (old leaves) and sink organs (new leaves/panicles), respectively, indicating that OsPHT1;7 is involved in P redistribution. Furthermore, OsPHT1;7 showed an overwhelmingly higher transcript abundance in anthers than other PHT1 members, and ospht1;7 mutants were impaired in P accumulation in anthers but not in pistils or husks. Moreover, the germination of pollen grains was significantly inhibited upon OsPHT1;7 mutation, leading to a >80% decrease in seed-setting rate and grain yield. Taken together, our results provide evidence that OsPHT1;7 is a crucial Pi transporter for Pi transport and recycling within rice plants, stimulating both vegetative and reproductive growth.

Boosting the performance of loss-tolerant measurement-device-independent quantum key distribution.

Measurement-device-independent quantum key distribution can remove all possible detector side channels, and is robust against state preparation flaws when further combined with the loss-tolerant method. However, the secure key rate in this scenario is relatively low, thus hindering its practical application. Here, we first present a four-intensity decoy-state protocol where the signal intensity is modulated only in Z basis for key generation while the decoy intensities are modulated in both Z and X bases for parameter estimation. Moreover, we adopt collective constraint and joint-study strategy in statistical fluctuation analysis. We have also experimentally demonstrated this protocol and the result indicates high performance and good security for practical applications.

Oxidative-antioxidant imbalance in chronic sialadenitis of submandibular gland in human and rat.

OBJECTIVES: This study aimed to evaluate the level of oxidative stress (OS) in human and rat chronic sialadenitis (CS) of the submandibular gland (SMG). METHOD: We collected human SMG tissues and established a rat CS model using Wharton's duct partial ligation (PL). Morphological changes in the SMG were evaluated by HE, Sirius Red, AB/PAS, TUNEL and immunohistochemical (IHC) staining. Oxidative damage and antioxidant capacity were detected by ELISA, commercial assay kits and IHC staining to evaluate their expression levels and locations in the SMG. RESULTS: Histopathological damage were observed in the human and rat CS. In rat PL group, the oxidative damage products (8-OHdG, AOPP, 8-iso-PGF2α and H2 O2 ) were significantly increased. For antioxidants, the PL group had markedly decreased T-AOC and CAT activity, but markedly increased SOD activity. 3-NT, 4-HNE and MDA expression changed during the process of CS, and antioxidant enzymes (CAT, SOD1, SOD2, GPX1 and GPX4) were mainly expressed in ducts. CONCLUSIONS: The oxidative-antioxidant imbalance of CS in human and rats was revealed, the different expression of oxidative damage during the process of CS was detected, and the different antioxidant reaction in acinar and ductal cells was demonstrated.

PCRMLP: A Two-Stage Network for Point Cloud Registration in Urban Scenes.

Point cloud registration plays a crucial role in 3D mapping and localization. Urban scene point clouds pose significant challenges for registration due to their large data volume, similar scenarios, and dynamic objects. Estimating the location by instances (bulidings, traffic lights, etc.) in urban scenes is a more humanized matter. In this paper, we propose PCRMLP (point cloud registration MLP), a novel model for urban scene point cloud registration that achieves comparable registration performance to prior learning-based methods. Compared to previous works that focused on extracting features and estimating correspondence, PCRMLP estimates transformation implicitly from concrete instances. The key innovation lies in the instance-level urban scene representation method, which leverages semantic segmentation and density-based spatial clustering of applications with noise (DBSCAN) to generate instance descriptors, enabling robust feature extraction, dynamic object filtering, and logical transformation estimation. Then, a lightweight network consisting of Multilayer Perceptrons (MLPs) is employed to obtain transformation in an encoder-decoder manner. Experimental validation on the KITTI dataset demonstrates that PCRMLP achieves satisfactory coarse transformation estimates from instance descriptors within a remarkable time of 0.0028 s. With the incorporation of an ICP refinement module, our proposed method outperforms prior learning-based approaches, yielding a rotation error of 2.01° and a translation error of 1.58 m. The experimental results highlight PCRMLP's potential for coarse registration of urban scene point clouds, thereby paving the way for its application in instance-level semantic mapping and localization.

Comparative Transcriptome Analysis Reveals OsBGs and OsGSLs Influence Sugar Transport through Callose Metabolism under Heat Stress in Rice.

Heat or high temperature stress have caused huge damage to many crops and have become the largest threat in terms of the future. Although a huge amount of research has been conducted to explore the mechanisms of heat tolerance and many achievements were accomplished, the mechanism by which how heat stress (HS) influences the yield is still unclear. In this study, RNA-seq analysis indicated that nine 1,3-ß-glucanases (BGs) belonging to the carbohydrate metabolic pathway were expressed differently during heat treatment. Therefore, we identified the BGs and glucan-synthase-likes (GSLs) in three rice ecotypes and processed the analyses of gene gain and loss, phylogenetic relationship, duplication, and syntenic relationship. We found the possibility of an environmental adaption based on BGs and GSLs during evolution. Submicrostructure and dry matter distribution analysis confirmed that HS might block the endoplasmic sugar transport pathway by increasing callose synthesis, which may lead to decreased yield and quality in rice production. This study provides a new clue regarding rice yield and quality under HS and provides guidance to rice cultivation and heat tolerance breeding.

Asymmetric Measurement-Device-Independent Quantum Key Distribution through Advantage Distillation.

Measurement-device-independent quantum key distribution (MDI-QKD) completely closes the security loopholes caused by the imperfection of devices at the detection terminal. Commonly, a symmetric MDI-QKD model is widely used in simulations and experiments. This scenario is far from a real quantum network, where the losses of channels connecting each user are quite different. To adapt such a feature, an asymmetric MDI-QKD model is proposed. How to improve the performance of asymmetric MDI-QKD also becomes an important research direction. In this work, an advantage distillation (AD) method is applied to further improve the performance of asymmetric MDI-QKD without changing the original system structure. Simulation results show that the AD method can improve the secret key rate and transmission distance, especially in the highly asymmetric cases. Therefore, this scheme will greatly promote the development of future MDI-QKD networks.

Recognition of centimeter-level length changes using the intensity of probe light in BOTDA.

We proposed a method based on the Brillouin optical time domain analysis (BOTDA) system to demodulate the length changes of the heating region using the intensity of probe light, under the premise that the temperature in the heating region does not change and the Brillouin frequency shift (BFS) of the heating region is known. This method can realize the recognition of small length under the condition of wide pulse width. The theoretical analysis shows that the different lengths of the heating region will lead to different gains when the spatial resolution corresponding to the pulse width of the pulsed light is larger than the length of the heating region. And through theoretical derivation and simulation, it is concluded that the relationship between the intensity of probe light and the length of the heating region satisfies the exponential growth. Further experiments verify that the exponential growth is approximately linear in the range of small length changes. The length of the heating region can be inversely deduced by using the linear relationship by detecting the intensity of probe light. When the length of sensing fiber is 100 m and the heating region changes from 2 cm to 8 cm, 1 cm heating region changes can be identified.

Centimeter-level spatial resolution Brillouin optical time domain analyzer using mono-pulse self-difference.

To break through the limitation of pulse width on spatial resolution in the Brillouin optical time-domain analysis system (BOTDA), a novel, to the best of our knowledge, scheme based on mono-pulse self-difference (MPSD) is proposed for centimeter-level spatial resolution. It is performed by the self-difference of the temporal trace generated by a single-shot long pulse in the conventional BOTDA. The concept-proof of the proposed approach is demonstrated, and experimental results are presented where distributed temperature sensing with a 5-cm spatial resolution is realized based on a 40-ns pulse along a 2-km sensing fiber.

Extracting Brillouin frequency shift accurately based on particle swarm optimization and a cross-correlation method.

A fitting algorithm based on particle swarm optimization (PSO) and the cross-correlation method (XCM) was proposed to accurately extract features of the Brillouin scattering spectrum. Both the simulation and experiment were conducted to verify the validity of the proposed algorithm. The results show that the algorithm achieves high demodulation accuracy and is suitable for Brillouin scattering spectra with different signal-to-noise ratios, full width at half-maximums, frequency sweep intervals, and spectral symmetries. Compared with the Lorenz curve fitting and XCM algorithm, the extraction error of the PSO-XCM can be optimized up to 99.98% and 99.93%, and the fitting degree can be improved by 98%. Moreover, based on the proposed algorithm, the minimum temperature measurement error can reach 0.06°C and the Brillouin frequency shift measurement error is 0.07 MHz in 10 km sensing fiber for the Brillouin optical time domain analysis (BOTDA) sensing system.

Chemically modified magnetic immobilized phospholipase A1 and its application for soybean oil degumming.

In this paper, the free Phospholipase A1 (PLA1) was immobilized on a magnetic carrier. The average particle diameter of the magnetic carrier was 97 ± 1.3 nm, and the average particle diameter of the magnetically immobilized PLA1 was 105 nm ± 1.3 nm. The enzyme activity was 1940.5 U/g. The magnetic enzyme was chemically modified with formaldehyde, dextran-aldehyde, and dextran-aldehyde-glycine. The proportions of primary amino groups in the modified magnetic immobilized enzyme PLA1 were 0, 53.5% and 47.3%, respectively. The optimum pH of the enzyme after chemical modification was 6.5. When the system temperature was 60 °C, the magnetically immobilized PLA1 modified with dextran-aldehyde-glycine had the optimal activity and stability. This chemically modified magnetic immobilized PLA1 was applied to soybean oil degumming at 60 °C, 6.5 h (reaction time), and 0.10 mg/kg (enzyme dosage). The phosphorus content in the degummed oil was 9.2 mg/kg. The relative enzyme activity was 77.6% after 7 reuses which would be potentially advantageous for industrial applications. SUPPLEMENTARY INFORMATION: The online version contains supplementary material available at (10.1007/s13197-021-05017-4).

Rapid noise removal based dual adversarial network for the Brillouin optical time domain analyzer.

We propose a dual adversarial network (DANet) to improve the signal-to-noise ratio (SNR) of the Brillouin optical time domain analyzer. Rather than inferring the conditional posteriori distribution in the conventional maximum a posteriori (MAP) framework, DANet constructs a joint distribution from two different factorizations corresponding to the noise removal and generation tasks. This method utilizes all the information between the clean-noisy image pairs to preserve data completely without requiring traditional image priors and noise distribution assumptions. Additionally, the clean-noisy image pairs produced by the generator can expand the original dataset to retrain and enhance the denoising effect. The performance of DANet is verified using the simulated and experimental data. Without spatial resolution deterioration, an SNR improvement of 35.51 dB is observed in the simulation, and the Brillouin frequency shift (BFS) uncertainty along the fiber is reduced by 3.56 MHz. Experiments yield a maximum SNR improvement of 19.08 dB, with the BFS uncertainty along the fiber reduced by 0.93 MHz. Significantly, DANet has a processing time of 1.26 s, which is considerably faster than conventional methods, demonstrating its potential for rapid noise removal tasks.

Experimental measurement-device-independent quantum key distribution with the double-scanning method.

The measurement-device-independent quantum key distribution (MDI-QKD) can be immune to all detector side-channel attacks. Moreover, it can be easily implemented combining with the matured decoy-state methods under current technology. It, thus, seems a very promising candidate in practical implementation of quantum communications. However, it suffers from a severe finite-data-size effect in most existing MDI-QKD protocols, resulting in relatively low key rates. Recently, Jiang et al. [Phys. Rev. A103, 012402 (2021).PLRAAN1050-294710.1103/PhysRevA.103.012402] proposed a double-scanning method to drastically increase the key rate of MDI-QKD. Based on Jiang et al.'s theoretical work, here we for the first time, to the best of our knowledge, implement the double-scanning method into MDI-QKD and carry out corresponding experimental demonstration. With a moderate number of pulses of 1010, we can achieve 150 km secure transmission distance, which is impossible with all former methods. Therefore, our present work paves the way toward practical implementation of MDI-QKD.

Machine Learning-Assisted Measurement Device-Independent Quantum Key Distribution on Reference Frame Calibration.

In most of the realistic measurement device-independent quantum key distribution (MDI-QKD) systems, efficient, real-time feedback controls are required to maintain system stability when facing disturbance from either external environment or imperfect internal components. Traditionally, people either use a "scanning-and-transmitting" program or insert an extra device to make a phase reference frame calibration for a stable high-visibility interference, resulting in higher system complexity and lower transmission efficiency. In this work, we build a machine learning-assisted MDI-QKD system, where a machine learning model-the long short-term memory (LSTM) network-is for the first time to apply onto the MDI-QKD system for reference frame calibrations. In this machine learning-assisted MDI-QKD system, one can predict out the phase drift between the two users in advance, and actively perform real-time phase compensations, dramatically increasing the key transmission efficiency. Furthermore, we carry out corresponding experimental demonstration over 100 km and 250 km commercial standard single-mode fibers, verifying the effectiveness of the approach.

Signal-to-noise ratio improvement of Brillouin optical time domain analysis system based on empirical mode decomposition and finite impulse response.

We propose a denoising algorithm based on empirical mode decomposition (EMD) and finite impulse response (FIR) to improve the signal-to-noise ratio (SNR) of Brillouin optical time domain analysis. Denoising results indicate EMD-FIR can effectively reduce noise, and the maximum SNR improvement is 11.69 dB, which is 4.98 dB and 4.26 dB larger than the maximum SNR improvement of wavelet and Butterworth. The temperature uncertainty along the heated section is reduced to 0.62°C by EMD-FIR. The improvement of SNR opens opportunities to apply high measurement accuracy to Brillouin optical time domain analysis and other distributed sensing fields.

The HNF1α-Regulated LncRNA HNF1α-AS1 Is Involved in the Regulation of Cytochrome P450 Expression in Human Liver Tissues and Huh7 Cells.

Expression of cytochrome P450s (P450s) is regulated by epigenetic factors, such as DNA methylation, histone modifications, and noncoding RNAs through different mechanisms. Among these factors, long noncoding RNAs (lncRNAs) have been shown to play important roles in the regulation of gene expression; however, little is known about the effects of lncRNAs on the regulation of P450 expression. The aim of this study was to explore the role of lncRNAs in the regulation of P450 expression by using human liver tissues and hepatoma Huh7 cells. Through lncRNA microarray analysis and quantitative polymerase chain reaction in human liver tissues, we found that the lncRNA hepatocyte nuclear factor 1 alpha antisense 1 (HNF1α-AS1), an antisense RNA of HNF1α, is positively correlated with the mRNA expression of CYP2C8, 2C9, 2C19, 2D6, 2E1, and 3A4 as well as pregnane X receptor (PXR) and constitutive androstane receptor (CAR). Gain- and loss-of-function studies in Huh7 cells transfected with small interfering RNAs or overexpression plasmids showed that HNF1α not only regulated the expression of HNF1α-AS1 and P450s, but also regulated the expression of CAR, PXR, and aryl hydrocarbon receptor (AhR). In turn, HNF1α-AS1 regulated the expression of PXR and most P450s without affecting the expression of HNF1α, AhR, and CAR. Moreover, the rifampicin-induced expression of P450s was also affected by HNF1α and HNF1α-AS1. In summary, the results of this study suggested that HNF1α-AS1 is involved in the HNF1α-mediated regulation of P450s in the liver at both basal and drug-induced levels.

Alterations of Histone Modifications Contribute to Pregnane X Receptor-Mediated Induction of CYP3A4 by Rifampicin.

CYP3A4 is one of the major drug-metabolizing enzymes in human and is responsible for the metabolism of 60% of clinically used drugs. Many drugs are able to induce the expression of CYP3A4, which usually causes drug-drug interactions and adverse drug reactions. This study aims to explore the role of histone modifications in rifampicin-induced expression of CYP3A4 in LS174T cells. We found that the induction of CYP3A4 mRNA (4- to 15-fold) by rifampicin in LS174T cells was associated with increased levels of histone H3 lysine 4 trimethylation (H3K4me3, above 1.8-fold) and H3 acetylation (above 2-fold) and a decreased level of histone H3 lysine 27 trimethylation (H3K27me3, about 50%) in the CYP3A4 promoter. Rifampicin enhanced recruitment to the CYP3A4 promoter of nuclear receptor coactivator 6 (NCOA6, above 3-fold) and histone acetyltransferase p300 (p300, above 1.6-fold). Silencing NCOA6 or p300 by short-hairpin RNAs resulted in inhibition of the CYP3A4 induction as well as altered levels of H3K4me3, H3K27me3, or H3 acetylation in the CYP3A4 promoter. Knockdown of pregnane X receptor (PXR) expression not only suppressed the recruitment of NCOA6 and p300 but also abolished the changes caused by rifampicin in H3K4me3, H3K27me3, and H3 acetylation levels in the CYP3A4 promoter. Moreover, rifampicin treatment enhanced the nuclear accumulation and interactions between PXR and NCOA6/p300. In conclusion, we show that the alterations of histone modifications contribute to the PXR-mediated induction of CYP3A4 by rifampicin.