Machine learning-assisted MALDI-TOF MS toward rapid classification of milk products.

This study established a method for rapid classification of milk products by combining matrix assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS) analysis with machine learning techniques. The analysis of 2 different types of milk products was used as an example. To select key variables as potential markers, integrated machine learning strategies based on 6 feature selection techniques combined with support vector machine (SVM) classifier were implemented to screen the informative features and classify the milk samples. The models were evaluated and compared by accuracy, Akaike information criterion (AIC), and Bayesian information criterion (BIC). The results showed the least absolute shrinkage and selection operator (LASSO) combined with SVM performs best, with prediction accuracy of 100 ± 0%, AIC of -360 ± 22, and BIC of -345 ± 22. Six features were selected by LASSO and identified based on the available protein molecular mass data. These results indicate that MALDI-TOF MS coupled with machine learning technique could be used to search for potential key targets for authentication and quality control of food products.

Theoretical comparison of fructose with methylglucoside for the production of formate and levulinate catalyzed by Brønsted acids in a methanol solution.

For the conversion of fructose/methylglucoside (MG) into both methyl formate (MF) and methyl levulinate (MLev), the C-source of formate [HCOO]- remains unclear at the molecular level. Herein, reaction mechanisms catalyzed by [CH3OH2]+ in a methanol solution were theoretically investigated at the PBE0/6-311++G(d,p) level. For the conversion of fructose into MF and MLev, the formate [HCOO]- comes from the C1-atom of fructose, in which the rate-determining step lies in the reaction of 5-hydroxymethylfurfural (HMF) with CH3OH to yield MF and MLev. The reaction of fructose with CH3OH kinetically tends to generate HMF intermediates rather than yield (MF + MLev). When MG is dissolved in a methanol solution, its O2, O3, and O4 atoms are closer to the first layer of the solvent than O1, O5, and O6 atoms. For the dehydration of MG with methanol into MF and MLev, the formate [HCOO]- stems from the dominant C1- and secondary C3-atoms of MG. Kinetically, MG is ready to yield (MF + MLev), whereas fructose can induce the reaction to remain at the HMF intermediate, inhibiting the further conversion of HMF with CH3OH into MF and MLev. If MG isomerizes into fructose, the reaction will be more preferable for yielding HMF rather than (MF + MLev).

Fruit freshness detection based on multi-task convolutional neural network.

Background: Fruit freshness detection by computer vision is essential for many agricultural applications, e.g., automatic harvesting and supply chain monitoring. This paper proposes to use the multi-task learning (MTL) paradigm to build a deep convolutional neural work for fruit freshness detection. Results: We design an MTL model that optimizes the freshness detection (T1) and fruit type classification (T2) tasks in parallel. The model uses a shared CNN (convolutional neural network) subnet and two FC (fully connected) task heads. The shared CNN acts as a feature extraction module and feeds the two task heads with common semantic features. Based on an open fruit image dataset, we conducted a comparative study of MTL and single-task learning (STL) paradigms. The STL models use the same CNN subnet with only one specific task head. In the MTL scenario, the T1 and T2 mean accuracies on the test set are 93.24% and 88.66%, respectively. Meanwhile, for STL, the two accuracies are 92.50% and 87.22%. Statistical tests report significant differences between MTL and STL on T1 and T2 test accuracies. We further investigated the extracted feature vectors (semantic embeddings) from the two STL models. The vectors have an averaged 0.7 cosine similarity on the entire dataset, with most values lying in the 0.6-0.8 range. This indicates a between-task correlation and justifies the effectiveness of the proposed MTL approach. Conclusion: This study proves that MTL exploits the mutual correlation between two or more relevant tasks and can maximally share their underlying feature extraction process. we envision this approach to be extended to other domains that involve multiple interconnected tasks.

A Georeferenced Dataset for Mapping and Assessing Subgrade Defects in China's High-Speed Railways.

China has the world's longest high-speed rail (HSR) network, marked by dense transportation and complex operations. However, frequent train use coupled with extreme weather conditions has led to rising subgrade issues. Existing railway defect records suffer from inconsistency, hindering direct applicability. Currently, there is a lack of a relevant dataset dedicated to HSR subgrade defects. To bridge this gap, we developed a comprehensive georeferenced dataset that encompasses defect records extracted from peer-reviewed literature published between 1999 and 2023 in China. Rigorous quality control procedures were implemented to eliminate duplicate data and ensure the accuracy of the dataset. The dataset consists of georeferenced records for eight different defects, spanning across 661 locations and categorized at various scales, ranging from provinces to townships. The most commonly reported defect types include subgrade settlement, frost damage, uplift deformation, and mud pumping. This dataset provides a comprehensive map of historical subgrade defects affecting high-speed railways in China. It could facilitate operational risk assessments and the prediction of subgrade performance.

Multi-defect risk assessment in high-speed rail subgrade infrastructure in China.

This study addresses the escalating risk of high-speed railway (HSR) infrastructure in China, amplified by climate warming, increased rainfall, frequent extreme weather, and geohazard events. Leveraging a georeferenced dataset of recent HSR defects obtained through an extensive literature review, we employ machine learning techniques for a quantitative multi-defect risk assessment. Climatic, geomorphological, geohydrological, and anthropogenic variables influencing HSR subgrade safety are identified and ranked. Climatic factors significantly impact frost damage and mud pumping, while geomorphological variables exhibit greater influence on settlement and uplift deformation defects. Notably, frost damage is prevalent in the northeast and northwest, mud pumping along the southeast coast, and settlement and uplift deformation in the northwest and central areas. The generated comprehensive risk map underscores high-risk zones, particularly the Menyuan Hui Autonomous and Minle County sections of the Lanzhou-Urumqi HSR, emphasizing the need for focused attention and preventive actions to mitigate potential losses and ensure operational continuity.

Discovery, optimization and biological evaluation of novel HBsAg production inhibitors.

Hepatitis B virus (HBV) infection is a major global health problem. HBsAg inhibitors are expected to reduce the production of HBsAg via inhibiting host proteins PAPD5 & PAPD7 and finally achieve the ideal goal of "functional cure". In this work, a series of tetrahydropyridine (THP) derivatives with a bridged ring were synthesized and evaluated for their inhibiting HBsAg production and HBV DNA activity. Among them, compound 17i was identified as potent HBsAg production inhibitor with excellent in vitro anti-HBV potency (HBV DNA EC50 = 0.018 µM, HBsAg EC50 = 0.044 µM) and low toxicity (CC50 > 100 µM). Moreover, 17i exhibited favorable in vitro/in vivo DMPK properties in mice. 17i could also significantly reduce serum HBsAg and HBV DNA levels (1.08 and 1.04 log units, respectively) in HBV transgenic mice.

Morphological Rule-Constrained Object Detection of Key Structures in Infant Fundus Image.

The detection of optic disc and macula is an essential step for ROP (Retinopathy of prematurity) zone segmentation and disease diagnosis. This paper aims to enhance deep learning-based object detection with domain-specific morphological rules. Based on the fundus morphology, we define five morphological rules, i.e., number restriction (maximum number of optic disc and macula is one), size restriction (e.g., optic disc width: 1.05 +/- 0.13 mm), distance restriction (distance between the optic disc and macula/fovea: 4.4 +/- 0.4 mm), angle/slope restriction (optic disc and macula should roughly be positioned in the same horizontal line), position restriction (In OD, the macula is on the left side of the optic disc; vice versa for OS). A case study on 2953 infant fundus images (with 2935 optic disc instances and 2892 macula instances) proves the effectiveness of the proposed method. Without the morphological rules, naïve object detection accuracies of optic disc and macula are 0.955 and 0.719, respectively. With the proposed method, false-positive ROIs (region of interest) are further ruled out, and the accuracy of the macula is raised to 0.811. The IoU (intersection over union) and RCE (relative center error) metrics are also improved.

The Treatment Satisfaction in Patients and Their Partners Treated with Low-Intensity Extracorporeal Shock Wave Therapy and Sildenafil: A Prospective Non-Randomized Controlled Study.

Background: Phosphodiesterase 5 inhibitors (PDE5Is) and other more invasive options merely provide symptomatic relief rather than a permanent improvement in erectile dysfunction (ED), whereas the long-term improvement in ED via low-intensity extracorporeal shockwave therapy (Li-ESWT) has been confirmed. So far, no comparative study of sildenafil versus Li-ESWT has been conducted with respect to treatment satisfaction. Objective: In this study, we aim to compare erectile function status and satisfaction rates in patients who received sildenafil or Li-ESWT for ED. Methods: Patients complaining of ED were considered candidates. Participants chose to enter one of two active treatment groups according to their treatment intention-either a 9-week Li-ESWT regimen or 100 mg on-demand sildenafil. The erectile function was evaluated using the erectile function domain of the International Index of Erectile Function questionnaires (IIEF-EF), while the treatment satisfaction was evaluated using the Erectile Dysfunction Inventory of Treatment Satisfaction questionnaires (EDITS). Results: We enrolled 72 participants in the study (42 in the Li-ESWT group and 30 in the sildenafil group). Patients in both groups were young men. Four weeks after the last session, the IIEF-EF score for Li-ESWT and sildenafil was 16.3± 5.5 and 18.3± 6.5 (P > 0.05), respectively. The total EDITS index of the patient version and the partner version were similar in the two groups. Among EDITS questions measuring overall satisfaction and efficacy duration, the score was higher in the Li-ESWT group. Conclusion: We found that Li-ESWT may have better satisfaction than on-demand sildenafil for young ED patients. However, further studies are needed to determine the factors influencing satisfaction.

Design and synthesis of boron-containing ALK inhibitor with favorable in vivo efficacy.

ALK is an attractive therapeutic target for the treatment of non-small cell lung cancer. As an emerging element in medicinal chemistry, boron has achieved great success in the discovery of antitumor drugs and antibacterial agents. Through construction of a BCC (boron-containing compound) compound library and broad kinase screening, we found the ALK inhibitor hit compound 10a. Structural optimization by CADD and isosterism revealed that lead compound 10k has improved activity (ALKL1196M IC50 = 8.4 nM, NCI-H2228 cells IC50 = 520 nM) and better in vitro metabolic stability (human liver microsomes, T1/2 = 238 min). Compound 10k showed good in vivo efficacy in a nude mouse NCI-H2228 lung cancer xenograft model with a TGI of 52 %. Molecular simulation analysis results show that the hydroxyl group on the oxaborole forms a key hydrogen bond with Asn1254 or Asp1270, and this binding site provides a new idea for drug design. This is the first publicly reported lead compound for a boron-containing ALK inhibitor.

Improved YOLOX detection algorithm for contraband in X-ray images.

It is important to perform contraband inspections on items before they are taken into public places in order to ensure the safety of people and property. At present, the mainstream method of judging contraband is that security inspectors observe the X-ray image of objects and judge whether they belong to contraband. Unfortunately, contraband is often hidden under other normal objects. In a high-intensity working environment, security inspectors are very prone to missed detection and wrong detection. To this end, a detection framework based on computer vision technology is proposed, which is trained and improved on the basis of the current state-of-the-art YOLOX object detection network, and adopts strategies such as feature fusion, adding a double attention mechanism and classifying regression loss. Compared with the benchmark YOLOX-S model, the proposed method achieves a higher average accuracy, with an improvement of 5.0% on the public safety SIXray dataset, opening the way to large-scale automatic detection of contraband in public places.

Matrix Factorization-Based Dimensionality Reduction Algorithms─A Comparative Study on Spectroscopic Profiling Data.

Spectroscopic profiling data used in analytical chemistry can be very high-dimensional. Dimensionality reduction (DR) is an effective way to handle the potential "curse of dimensionality" problem. Among the existing DR algorithms, many can be categorized as a matrix factorization (MF) problem, which decomposes the original data matrix X into the product of a low-dimensional matrix W and a dictionary matrix H. First, this paper provides a theoretical reformulation of relevant DR algorithms under a unified MF perspective, including PCA (principal component analysis), NMF (non-negative matrix factorization), LAE (linear autoencoder), RP (random projection), SRP (sparse random projection), VQ (vector quantization), AA (archetypical analysis), and ICA (independent component analysis). From this perspective, an open-sourced toolkit has been developed to integrate all of the above algorithms with a unified API. Second, we made a comparative study on MF-based DR algorithms. In a case study of TOF (time-of-flight) mass spectra, the eight algorithms extracted three components from the original 27,619 features. The results are compared by a set of DR quality metrics, e.g., reconstruction error, pairwise distance/ranking property, computational cost, local and global structure preservations, etc. Finally, based on the case study result, we summarized guidelines for DR algorithm selection. (1) For reconstruction quality, choose ICA. In the case study, ICA, PCA, and NMF have high reconstruction qualities (reconstruction error < 2%), ICA being the best. (2) To keep the pairwise topological structure, choose PCA. PCA best preserves the pairwise distance/ranking property. (3) For edge computing and IoT scenarios, choose RP or SRP if reconstruction is not required and the JL-lemma condition is met. The RP family has the best computational performance in the experiment, almost 10-100 times faster than its peers.

Changes in temporal pattern and spatial distribution of environmental pollutants in 8 Asian countries owing to COVID-19 pandemic.

This study investigated the changes in air pollutant's concentration, spatio-temporal distribution and sensitivity of changes in air pollutant's concentration during pre and post COVID-19 outbreak. We employed Google Earth Engine Platform to access remote sensing datasets of air pollutants across Asian continent. Air pollution and cumulative confirmed-COVID cases data of Asian countries (Afghanistan, Bangladesh, China, India, Iran, Iraq, Pakistan, and Saudi Arabia) have been collected and analyzed for 2019 and 2020. The results indicate that aerosol index (AI) and nitrogen dioxide (NO2) is significantly reduced during COVID outbreak i.e. in year 2020. In addition, we found significantly positive (P < 0.05, 95% confidence interval, two-tailed) correlation between changes in AI and NO2 concentration for net active-COVID case increment in almost each country. For other atmospheric gases i.e. carbon monoxide (CO), formaldehyde (HCHO), ozone (O3), and Sulfur dioxide (SO2), insignificant and/or significant negative correlation is also observed. These results suggest that the atmospheric concentration of AI and NO2 are good indicators of human activities. Furthermore, the changes in O3 shows significantly negative correlation for net active-COVID case increment. In conclusion, we observed significant positive environmental impact of COVID-19 restrictions in Asia. This study would help and assist environmentalist and policy makers in restraining air pollution by implementing efficient restrictions on human activities with minimal economic loss.

Significant CO2 sink over the Tibet's largest lake: Implication for carbon neutrality across the Tibetan Plateau.

In this study, the CO2 flux over Siling Co Lake, the largest lake in Tibet, was analyzed employing the eddy covariance observation from 26 April 2014 to 22 March 2015. The results showed that Siling Co Lake acts as a net sink of CO2 with annual absorption of 315.5 g C m-2 a-1. The inorganic processes might be the main reason of CO2 absorption in Siling Co Lake owing to its high pH as an endorheic lake. Therefore, it is essential to consider the carbon absorption of lakes when assessing the carbon sink function of the Tibetan Plateau (TP). There would be more carbon surplus on the TP when the carbon absorption by lakes is considered, which will greatly contribute to the realization of carbon neutralization across TP. This study highlights the need to strengthen the systematic observation of long-term series of CO2 flux in lakes on the TP, and further analyse the mechanism of CO2 exchange between lakes and the atmosphere.

Immune suppressive function of IL-1α release in the tumor microenvironment regulated by calpain 1.

Interleukin-1α (IL-1α) plays an important role in inflammation and hematopoiesis. Many tumors have increased IL-1α expression. However, the immune regulatory role of secreted IL-1α in tumor development and whether it can be targeted for cancer therapy are still unclear. Here, we found that tumoral-secreted IL-1α significantly promoted hepatocellular carcinoma (HCC) development in vivo. Tumoral-released IL-1α were found to inhibit T and NK cell activation, and the killing capacity of CD8+ T cells. Moreover, MDSCs were dramatically increased by tumoral-released IL-1α in both spleens and tumors. Indeed, higher tumoral IL-1α expression is associated with increased tumoral infiltration of MDSCs in HCC patients. Further studies showed that tumoral-released IL-1α promoted MDSC recruitment to the tumor microenvironment through a CXCR2-dependent mechanism. Depletion of MDSCs could diminish the tumor-promoting effect of tumoral-released IL-1α. On the contrary, systemic administration of recombinant IL-1α protein significantly inhibited tumor development by activating T cells. In fact, IL-1α protein could promote T cell activation and enhance the cytotoxicity of CD8+ T cells in vitro. Thus, our study demonstrated that tumoral-released IL-1α promoted tumor development through recruiting MDSCs to inhibit T cell activation, while systemic IL-1α directly promoted anti-tumor T cell responses. We further identified calpain 1 as the major intracellular protease mediating tumoral IL-1α secretion. Calpain 1 KO tumors had diminished IL-1α release and reduced tumor development. Thus, our findings provide new insights into the functions of secreted IL-1α in tumor immunity and its implications for immunotherapy.

Spectroscopic profiling-based geographic herb identification by neural network with random weights.

Daodi medicinal material plays an important role in traditional Chinese medicine (TCM). This study researches and validates the NNRW (neural network with random weights) model on spectroscopic profiling data for geographical origin identification. NNRW is a special neural network model that does not require an iterative training process. It has been proved effective in various resource-limited data-driven applications. However, whether NNRW works for spectroscopic profiling data remains to be explored. In this study, the Raman and UV (ultraviolet) profiling data of 160 radix astragali samples from four geographic regions are trained and evaluated by four classification models, i.e., NNRW, MLP (multi-layer perceptron), SVM (support vector machine), and DTC (decision tree classifier). Their validation accuracies are 96.3%, 98.0%, 98.4%, and 92.8% respectively. The training/fitting times are 0.372 ms (milli-seconds), 57.9 ms, 2.033 ms, and 3.351 ms, respectively. This study shows that NNRW has a significant training time cut while keeping a high prediction accuracy, and it is a promising solution to resource-limited edge computing applications.

Comparative Analysis Reveals Novel Changes in Plasma Metabolites and Metabolomic Networks of Infants With Retinopathy of Prematurity.

Purpose: Advances in mass spectrometry have provided new insights into the role of metabolomics in the etiology of several diseases. Studies on retinopathy of prematurity (ROP), for example, overlooked the role of metabolic alterations in disease development. We employed comprehensive metabolic profiling and gold-standard metabolic analysis to explore major metabolites and metabolic pathways, which were significantly affected in early stages of pathogenesis toward ROP. Methods: This was a multicenter, retrospective, matched-pair, case-control study. We collected plasma from 57 ROP cases and 57 strictly matched non-ROP controls. Non-targeted ultra-high-performance liquid chromatography-tandem mass spectroscopy (UPLC-MS/MS) was used to detect the metabolites. Machine learning was employed to reveal the most affected metabolites and pathways in ROP development. Results: Compared with non-ROP controls, we found a significant metabolic perturbation in the plasma of ROP cases, which featured an increase in the levels of lipids, nucleotides, and carbohydrate metabolites and lower levels of peptides. Machine leaning enabled us to distinguish a cluster of metabolic pathways (glycometabolism, redox homeostasis, lipid metabolism, and arginine pathway) were strongly correlated with the development of ROP. Moreover, the severity of ROP was associated with the levels of creatinine and ribitol; also, overactivity of aerobic glycolysis and lipid metabolism was noted in the metabolic profile of ROP. Conclusions: The results suggest a strong correlation between metabolic profiling and retinal neovascularization in ROP pathogenesis. These findings provide an insight into the identification of novel metabolic biomarkers for the diagnosis and prevention of ROP, but the clinical significance requires further validation.

Task-adaptive eigenvector-based projection (EBP) transform for compressed sensing: A case study of spectroscopic profiling sensor.

The compressed sensing (CS) theory requires the signal to be sparse under some transform. For most signals (e.g., speech and photos), the non-adaptive transform bases, such as discrete cosine transform (DCT), discrete Fourier transform (DFT), and Walsh-Hadamard transform (WHT), can meet this requirement and perform quite well. However, one limitation of these non-adaptive transforms is that we cannot leverage domain-specific knowledge to improve CS efficiency. This study presents a task-adaptive eigenvector-based projection (EBP) transform. The EBP basis has an equivalent effect of the principal component loading matrix and can generate a sparse representation in the latent space. In a Raman spectroscopic profiling case study, EBP demonstrates better performance than its non-adaptive counterparts. At the 1% CS sampling ratio (k), the reconstruction relative mean square errors of DCT, DFT, WHT and EBP are 0.33, 0.68, 0.32, and 0.00, respectively. At a fixed k, EBP achieves much better reconstruction quality than the non-adaptive counterparts. For specific domain tasks, EBP can significantly lower the CS sampling ratio and reduce the overall measurement cost.

pyDRMetrics - A Python toolkit for dimensionality reduction quality assessment.

High-dimensional data are pervasive in this bigdata era. To avoid the curse of the dimensionality problem, various dimensionality reduction (DR) algorithms have been proposed. To facilitate systematic DR quality comparison and assessment, this paper reviews related metrics and develops an open-source Python package pyDRMetrics. Supported metrics include reconstruction error, distance matrix, residual variance, ranking matrix, co-ranking matrix, trustworthiness, continuity, co-k-nearest neighbor size, LCMC (local continuity meta criterion), and rank-based local/global properties. pyDRMetrics provides a native Python class and a web-oriented API. A case study of mass spectra is conducted to demonstrate the package functions. A web GUI wrapper is also published to support user-friendly B/S applications.

Discovery of a Silicon-Containing Pan-Genotype Hepatitis C Virus NS5A Inhibitor.

We describe a study leading to the discovery of compound 11, a pan-genotypic hepatitis C virus (HCV) nonstructural protein 5A (NS5A) inhibitor with excellent potency, metabolic stability, and pharmacokinetics (PK). Compound 11 incorporating a 4-silapiperidine group was discovered by further optimizing our previous lead with a triethylsilyl moiety. It displayed great potency against genotype 1 subtype a (GT1a), -1b, -2a, -3a, -4a, -5a, and -6a with an EC50 range of 0.33-17 pM and improved potency against the resistance-associated variant GT1a_M28T. Pharmacokinetics (PK) study indicated that compound 11 has reasonable PK exposures with a high liver distribution in preclinical animal species (mouse, rat, and dog). The results of a 14 day repeat-dose toxicity study identified the safety of compound 11.

Adaptive compressed sensing of Raman spectroscopic profiling data for discriminative tasks.

Raman spectroscopy is widely used in discriminative tasks. It provides a wide-range physio-chemical fingerprint in a rapid and non-invasive way. The Raman spectrometry uses a sensor array to convert photon signals into digital spectroscopic data. This analog-to-digital process can benefit from the compressed sensing (CS) technique. The major benefits include less memory usage, shorter acquisition time, and more cost-efficient sensor. Traditional compressed sensing and reconstruction is a series of mathematical operations performed on the signal. Meanwhile, for discriminative tasks, both the signal and the categorical information are involved. For such scenarios, this paper proposes a method that uses both domain signal and categorical information to optimize CS hyper-parameters, including 1) the sampling ratio or the sensing matrix, 2) the basis matrix for the sparse transform, and 3) the regularization rate or shrinkage factor for L1-norm minimization. A case study of formula milk brand identification proves the proposed method can generate effective compressed sensing while preserving enough discriminative power in the reconstructed signal. Under the optimized hyper-parameters, a 100% classification accuracy is retained by only sampling 20% of the original signal.