A noise audit of human-labeled benchmarks for machine commonsense reasoning.

With the advent of large language models, evaluating and benchmarking these systems on important AI problems has taken on newfound importance. Such benchmarking typically involves comparing the predictions of a system against human labels (or a single 'ground-truth'). However, much recent work in psychology has suggested that most tasks involving significant human judgment can have non-trivial degrees of noise. In his book, Kahneman suggests that noise may be a much more significant component of inaccuracy compared to bias, which has been studied more extensively in the AI community. This article proposes a detailed noise audit of human-labeled benchmarks in machine commonsense reasoning, an important current area of AI research. We conduct noise audits under two important experimental conditions: one in a smaller-scale but higher-quality labeling setting, and another in a larger-scale, more realistic online crowdsourced setting. Using Kahneman's framework of noise, our results consistently show non-trivial amounts of level, pattern, and system noise, even in the higher-quality setting, with comparable results in the crowdsourced setting. We find that noise can significantly influence the performance estimates that we obtain of commonsense reasoning systems, even if the 'system' is a human; in some cases, by almost 10 percent. Labeling noise also affects performance estimates of systems like ChatGPT by more than 4 percent. Our results suggest that the default practice in the AI community of assuming and using a 'single' ground-truth, even on problems requiring seemingly straightforward human judgment, may warrant empirical and methodological re-visiting.

Association between baseline hemoglobin level and early neurological deterioration after intravenous thrombolysis in patients with acute ischemic stroke. / åŸºçº¿è¡€çº¢è›‹ç™½æ°´å¹³å¯¹æ€¥æ€§ç¼ºè¡€æ€§å’ä¸­æ‚£è€ é™è„‰æº¶æ “åŽæ—©æœŸç¥žç»åŠŸèƒ½æ¶åŒ–çš„å½±å“.

OBJECTIVES: To investigate the association between baseline hemoglobin level and early neurologic deterioration (END) after intravenous thrombolysis in patients with acute ischemic stroke (AIS). METHODS: Data of AIS patients who received intravenous thrombolytic therapy at multiple hospitals across the country between January 2017 and July 2020 were collected from the online database Acute Stroke Patients for stroke Management Quality Evaluation (CASE-Ⅱ, NCT04487340). Binary logistic regression analysis was used to study the factors affecting the occurrence of END after intravenous thrombolytic therapy, and the correlation between baseline hemoglobin level and END was investigated by limiting cubic spline curve. RESULTS: A total of 8162 patients were included. Patients with END had lower baseline hemoglobin levels (136 and 140 g/L, P<0.01) and higher rates of anemia (24.2% and 16.9%, P<0.01) compared with non-END patients. Binary logistic regression analysis showed that baseline hemoglobin level (OR=0.995, 95%CI: 0.991-0.999, P<0.05) and anemia (OR=1.238, 95%CI: 1.055-1.454, P<0.01) were independently correlated with the occurrence of END after intravenous thrombolysis in AIS patients. Restricted cubic spline regression showed that there was a U-shaped relationship between hemoglobin level and the risk of END after intravenous thrombolysis in AIS patients (P<0.01); while the relationship only existed in male patients (P<0.05), not in female patients (P>0.05). CONCLUSIONS: There is a U-shaped correlation between baseline hemoglobin level and the risk of END in AIS patients after intravenous thrombolysis, that is, both lower and higher hemoglobin level may increase the risk of END.

Uptake of ox-LDL by binding to LRP6 mediates oxidative stress-induced BMSCs senescence promoting obesity-related bone loss.

Obesity has long been thought to be a main cause of hyperlipidemia. As a systemic disease, the impact of obesity on organs, tissues and cells is almost entirely negative. However, the relationship between obesity and bone loss is highly controversial. On the one hand, obesity has long been thought to have a positive effect on bone due to increased mechanical loading on the skeleton, conducive to increasing bone mass to accommodate the extra weight. On the other hand, obesity-related metabolic oxidative modification of low-density lipoprotein (LDL) in vivo causes a gradual increase of oxidized LDL (ox-LDL) in the bone marrow microenvironment. We have reported that low-density lipoprotein receptor-related protein 6 (LRP6) acts as a receptor of ox-LDL and mediates the bone marrow stromal cells (BMSCs) uptake of ox-LDL. We detected elevated serum ox-LDL in obese mice. We found that ox-LDL uptake by LRP6 led to an increase of intracellular reactive oxygen species (ROS) in BMSCs, and N-acetyl-L-cysteine (NAC) alleviated the cellular senescence and impairment of osteogenesis induced by ox-LDL. Moreover, LRP6 is a co-receptor of Wnt signaling. We found that LRP6 preferentially binds to ox-LDL rather than dickkopf-related protein 1 (DKK1), both inhibiting Wnt signaling and promoting BMSCs senescence. Mesoderm development LRP chaperone (MESD) overexpression inhibits ox-LDL binding to LRP6, attenuating oxidative stress and BMSCs senescence, eventually rescuing bone phenotype.

Unlatching the Additional Zinc Storage Ability of Vanadium Nitride Nanocrystallites.

Vanadium-based materials, due to their diverse valence states and open-framework lattice, are promising cathodes for aqueous zinc ion batteries (AZIBs), but encounters the major challenges of in situ electrochemical activation process, potent polarity of the aqueous electrolyte and periodic expansion/contraction for efficient Zn2+ storage. Herein, architecting vanadium nitride (VN) nanosheets over titanium-based hollow nanoarrays skeletal host (denoted VNTONC) can simultaneously modulate address those challenges by creating multiple interfaces and maintaining the (1 1 1) phase of VN, which optimizes the Zn2+ storage and the stability of VN. Benefiting from the modulated crystalline thermodynamics during the electrochemical activation of VN, two outcomes are achieved; I) the cathode transforms into a nanocrystalline structure with increased active sites and higher conductivity and; II) a significant portion of the (1 1 1) crystal facets is retained in the process leading to the additional Zn2+ storage capacity. As a result, the as-prepared VNTONC electrode demonstrates remarkable discharge capacities of 802.5 and 331.8 mAh g-1 @ 0.5 and 6.0 A g-1 , respectively, due to the enhanced kinetics as validated by theoretical calculations. The assembled VNTONC||Zn flexible ZIB demonstrates excellent Zn storage properties up to 405.6 mAh g-1 , and remarkable robustness against extreme operating conditions.

Ultra-high-performance liquid chromatography-quadrupole-time of flight-mass spectrometry combined with network pharmacology for analysis of potential quality markers of three processed products of Qingpi.

Qingpi, a well-known traditional Chinese medicine for qi-regulating and commonly processed into three types of pieces, has been widely used in the clinical application of liver disease for thousands of years. In this study, an ultra-high-performance liquid chromatography-quadrupole-time of flight-mass spectrometry approach along with multivariate statistical analysis was developed to assess and characterize the differentiations of three processed products and confirm the potential quality markers of Qingpi. In addition, a systematic analysis combined with network pharmacology and molecular docking was performed to clarify the potential mechanism of Qingpi for the treatment of liver disease. As a result, 18 components were identified and an integrated network of Qingpi-Components-Target-Pathway-Liver Disease was constructed. Eight compounds were finally screened out as the potential quality markers acting on ten main targets and pathways of liver disease. Molecular docking analysis results indicated that the quality markers had a good binding activity with the targets. Overall, this work preliminarily identified the potential quality markers of three processed products of Qingpi, and predicted its targets in the prevention and treatment of liver disease, which can provide supporting information for further study of the pharmacodynamic substances and mechanisms of Qingpi.

Melatonin, a potentially effective drug for the treatment of infected bone nonunion.

Osteomyelitis (OM), characterized by heterogeneity and complexity in treatment, has a high risk of infection recurrence which may cause limb disability. Management of chronic inactive osteomyelitis (CIOM) without typical inflammatory symptoms is a great challenge for orthopedic surgeons. On the basis of data analysis of 1091 OM cases, we reported that latent osteogenic decline in CIOM patients was the main cause of secondary surgery. Our research shows that impairment of osteoblasts capacity in CIOM patients is associated with ferroptosis of osteoblasts caused by internalization of Staphylococcus aureus. Further studies show that melatonin could alleviate ferroptosis of osteoblasts in infected states through Nox4/ROS/P38 axis and protect the osteogenic ability of CIOM patients. Knockout of NADPH oxidase 4 (Nox4) in vivo could effectively relieve ferroptosis of osteoblasts in the state of infection and promote osteogenesis. Through a large number of clinical data analyses combined with molecular experiments, this study clarified that occult osteogenic disorders in CIOM patients were related to ferroptosis of osteoblasts. We revealed that melatonin might be a potential therapeutic drug for CIOM patients and provided a new insight for the treatment of OM.

In vitro characterization of alternative l-threonate and d-erythronate catabolic pathways.

l-threonate is the metabolite of vitamin C, while d-erythronate is the metabolite of N-acetyl-d-glucosamine, the nutritional supplement for joint health. They are widely distributed in the environment and human biofluids. Nevertheless, the catabolisms of l-threonate and d-erythronate are sparsely reported. Here we explored the functional diversity of an acid sugar kinase family (Pfam families PF07005-PF17042), and discovered a novel 2-oxo-tetronate kinase. The conserved genome neighborhood of the 2-oxo-tetronate kinase encodes members of class-II fructose-bisphosphate aldolase family (F_bP_aldolase, PF01116) and a dehydrogenase family (PF03446-PF14833). Instructed by this analysis, we experimentally verified that these enzymes are capable of degrading l-threonate into dihydroxyacetone phosphate (DHAP) in Arthrobacter sp. ZBG10, Clostridium scindens ATCC 35704, and Pseudonocardia dioxanivorans ATCC 55486. Meanwhile, a convergent catabolic pathway for d-erythronate was characterized in P. dioxanivorans ATCC 55486. Moreover, the phylogenetic distribution analysis indicates that the biological range of the identified l-threonate and d-erythronate catabolic pathways appears to extend mostly to members of the Actinomycetota, Cyanobacteriota, Bacillota, Pseudomonadota, and Bacteroidota phyla.

Optimal weak measurement scheme for chiral molecular detection based on photonic spin Hall effect.

In this paper, we propose a precision method to measure the chiroptical signal of Artemisinin solutions using the photonic spin Hall effect (PSHE) on the Ce:YIG-YIG-SiO2 structure as a probe. The effects of transmission distance, incident angles, applied magnetic fields of different directions, and beam waist of light on the weak measurement system are analytically investigated through simulations. It is found that decreasing the beam waist of the incident spot, increasing the incident angle, increasing the transmission distance, and adding a longitudinal magnetic field is conducive to enhancing the amplification transverse shift of PSHE, thus the measurement sensitivity is greatly improved. Based on the optimal weak measurement scheme, the detection limit for concentration measurement of artemisinin based on optical rotatory (OR) was reduced to 0.05 mg/ml. The measurement precision of the OR angle has been improved to 10-7rad.

Quantifying confidence shifts in a BERT-based question answering system evaluated on perturbed instances.

Recent work on transformer-based neural networks has led to impressive advances on multiple-choice natural language processing (NLP) problems, such as Question Answering (QA) and abductive reasoning. Despite these advances, there is limited work still on systematically evaluating such models in ambiguous situations where (for example) no correct answer exists for a given prompt among the provided set of choices. Such ambiguous situations are not infrequent in real world applications. We design and conduct an experimental study of this phenomenon using three probes that aim to 'confuse' the model by perturbing QA instances in a consistent and well-defined manner. Using a detailed set of results based on an established transformer-based multiple-choice QA system on two established benchmark datasets, we show that the model's confidence in its results is very different from that of an expected model that is 'agnostic' to all choices that are incorrect. Our results suggest that high performance on idealized QA instances should not be used to infer or extrapolate similarly high performance on more ambiguous instances. Auxiliary results suggest that the model may not be able to distinguish between these two situations with sufficient certainty. Stronger testing protocols and benchmarking may hence be necessary before such models are deployed in front-facing systems or ambiguous decision making with significant human impact.

Case Report: A novel de novo variant of COL1A1 in fetal genetic osteogenesis imperfecta.

Objective: Osteogenesis imperfecta (OI) is a rare genetic disorder. Clinical severity is heterogeneous. The purpose of this study was to investigate the genetic characteristics of a fetus with OI by whole exome sequencing (WES) and identify the cause of the disease. Methods: In this study, a fetus with osteogenic dysplasia was referred to our hospital. DNA was extracted from the aborted fetal tissue and peripheral blood of the parents. To identify the pathogenic genes, we conducted the trio-WES using DNA. A de novo variant in the COL1A1 gene is suspected to be the cause of the OI phenotype. We used Sanger sequencing for validation and various bioinformatics methods (such as SIFT, PolyPhen2, Mutation Taster, conservative analysis, SWISS Model, glycosylation site prediction, and I-Mutant 2.0) for analysis. Results: Both WES and Sanger sequencing identified a novel de novo variant of COL1A1 (c. 1309G>A, p. Gly437Ser) in a fetus with OI. Bioinformatic analysis showed that the affected residue, p. Gly437, was highly conserved in multiple species and predicted that the variant was deleterious and may have an impact on protein function. This variant is present in highly conserved glycine residues of Gly-X-Y sequence repeats of the triple helical region of the collagen type I α chain, which may be the cause of OI. Conclusion: This study revealed that the c.1309G>A (p. Gly437Ser) variant in the COL1A1 gene may be the genetic cause of fetal OI in this case. The discovery of this variant enriched the variation spectrum of OI. WES improves the accurate diagnosis of fetal OI, and doctors can provide patients with appropriate genetic counseling.

TG-CSR: A human-labeled dataset grounded in nine formal commonsense categories.

Machine Common Sense Reasoning is the subfield of Artificial Intelligence that aims to enable machines to behave or make decisions similarly to humans in everyday and ordinary situations. To measure progress, benchmarks in the form of question-answering datasets have been developed and published in the community to evaluate machine commonsense models, including large language models. We describe the individual label data produced by six human annotators originally used in computing ground truth for the Theoretically-Grounded Commonsense Reasoning (TG-CSR) benchmark's composing datasets. According to a set of instructions, annotators were provided with spreadsheets containing the original TG-CSR prompts and asked to insert labels in specific spreadsheet cells during annotation sessions. TG-CSR data is organized in JSON files, individual raw label data in a spreadsheet file, and individual normalized label data in JSONL files. The release of individual labels can enable the analysis of the labeling process itself, including studies of noise and consistency across annotators.

High-fat diet promotes tumor growth in the patient-derived orthotopic xenograft (PDOX) mouse model of ER positive endometrial cancer.

Endometrial cancer, one of the common gynecological malignancies, is affected by several influencing factors. This study established a unique patient-derived orthotopic xenograft (PDOX) nude mouse model for the study of influencing factors in ER positive endometrial cancer. The aim of this study was to demonstrate that a high-fat diet can affect the growth of ER positive endometrial cancer PDOX model tumors. The tumor tissues were expanded by subcutaneous transplantation in nude mice, and then the subcutaneous tumor tissues were orthotopically implanted into the nude mouse uterus to establish the PDOX model. After modeling, they were divided into high-fat diet group and normal diet group for 8 weeks of feeding, which showed that high-fat diet significantly promoted tumor growth (P < 0.001) and increased the protein expression level of ERα in tumor tissues. This study demonstrates that PDOX models of endometrial cancer can embody the role of dietary influences on tumor growth and that this model has the potential for preclinical studies of cancer promoting factors.

Bone-derived PDGF-BB drives brain vascular calcification in male mice.

Brain vascular calcification is a prevalent age-related condition often accompanying neurodegenerative and neuroinflammatory diseases. The pathogenesis of large-vessel calcifications in peripheral tissue is well studied, but microvascular calcification in the brain remains poorly understood. Here, we report that elevated platelet-derived growth factor BB (PDGF-BB) from bone preosteoclasts contributed to cerebrovascular calcification in male mice. Aged male mice had higher serum PDGF-BB levels and a higher incidence of brain calcification compared with young mice, mainly in the thalamus. Transgenic mice with preosteoclast-specific Pdgfb overexpression exhibited elevated serum PDGF-BB levels and recapitulated age-associated thalamic calcification. Conversely, mice with preosteoclast-specific Pdgfb deletion displayed diminished age-associated thalamic calcification. In an ex vivo cerebral microvascular culture system, PDGF-BB dose-dependently promoted vascular calcification. Analysis of osteogenic gene array and single-cell RNA-Seq (scRNA-Seq) revealed that PDGF-BB upregulated multiple osteogenic differentiation genes and the phosphate transporter Slc20a1 in cerebral microvessels. Mechanistically, PDGF-BB stimulated the phosphorylation of its receptor PDGFRß (p-PDGFRß) and ERK (p-ERK), leading to the activation of RUNX2. This activation, in turn, induced the transcription of osteoblast differentiation genes in PCs and upregulated Slc20a1 in astrocytes. Thus, bone-derived PDGF-BB induced brain vascular calcification by activating the p-PDGFRß/p-ERK/RUNX2 signaling cascade in cerebrovascular cells.

Quantifying COVID-19 policy impacts on subjective well-being during the early phase of the pandemic: A cross-sectional analysis of United States survey data from March to August 2020.

To stop the spread of COVID-19, a number of public health policies and restrictions were implemented during the pre-vaccination phase of the pandemic. This study provides a quantitative assessment of how these policies impacted subjective well-being (SWB) in the United States over a 6-month period spanning March to August 2020. We study two specific research objectives. First, we aim to quantify the impacts of COVID-19 public health policies at different levels of stringency on SWB. Second, we train and implement a conditional inference tree model for predicting individual SWB based both on socio-demographic characteristics and policies then in place. Our results indicate that policies such as enforcing strict stay-at-home requirements and closing workplaces were negatively associated with SWB, and that an individual's socio-demographic characteristics, including income status, job, and gender, conditionally interact with policies such as workplace closure in a predictive model of SWB. Therefore, although such policies may have positive health implications, they also have secondary environmental and social implications that need to be taken into account in any cost-benefit analysis of such policies for future pandemic preparedness. Our proposed methodology suggests a way to quantify such impacts through the lens of SWB, and to further advance the science of pandemic preparedness from a public health perspective.

Fabrication of triple-crosslinked gelatin/alginate hydrogels for controlled release applications.

Hydrogels have been demonstrated as smart drug carriers to recognize the tumor microenvironment for cancer treatment, where the dynamic crosslinks in the hydrogel network contribute to the stimuli-responsive features but also result in poor stability and weak mechanical property of the hydrogels. Here, phenylboronic acid-grafted polyethyleneimine (PBA-PEI)-modified gelatin (PPG) was synthesized to crosslink alginate dialdehyde (ADA) through imine bonds and boronate ester bonds, and then calcium ions (Ca2+) were added to introduce the third calcium-carboxylate crosslinking in the network to form the triple-crosslinked PPG/ADA-Ca2+ hydrogels. Given the three types of dynamic bonds in the network, PPG/ADA-Ca2+ hydrogels possessed a self-healing manner, stimuli-responsiveness, and better mechanical properties compared to single- or double-crosslinked hydrogels. The controlled release capability of PPG/ADA-Ca2+ hydrogels was also demonstrated, showing the encapsulated molecules can be rapidly released from the hydrogel network in the presence of hydrogen peroxide while the release rate can be slowed down at acidic pH. Furthermore, PPG/ADA-Ca2+ hydrogels presented selected cytotoxicity and drug delivery to cancer cells due to the regulated degradation by the cellular microenvironment. Taken together, PPG/ADA-Ca2+ hydrogels have been demonstrated as promising biomaterials with multiple desirable properties and dynamic features to perform controlled molecule release for biomedical applications.

Coffee-Associated Endophytes: Plant Growth Promotion and Crop Protection.

Endophytic microbes are a ubiquitous group of plant-associated communities that colonize the intercellular or intracellular host tissues while providing numerous beneficial effects to the plants. All the plant species are thought to be associated with endophytes, majorly constituted with bacteria and fungi. During the last two decades, there has been a considerable movement toward the study of endophytes associated with coffee plants. In this review, the main consideration is given to address the coffee-associated endophytic bacteria and fungi, particularly their action on plant growth promotion and the biocontrol of pests. In addition, we sought to identify and analyze the gaps in the available research. Additionally, the potential of endophytes to improve the quality of coffee seeds is briefly discussed. Even though there are limited studies on the subject, the potentiality of coffee endophytes in plant growth promotion through enhancing nitrogen fixation, availability of minerals, nutrient absorption, secretion of phytohormones, and other bioactive metabolites has been well recognized. Further, the antagonistic effect against various coffee pathogenic bacteria, fungi, nematodes, and also insect pests leads to the protection of the crop. Furthermore, it is recognized that endophytes enhance the sensory characteristics of coffee as a new field of study.

Observation of robust edge mode and in-gap corner mode in Kagome surface-wave photonic crystals.

Recent theory has demonstrated that Kagome photonic crystals (PCs) support first-order and second-order topological phenomena. Here, we extend the topological physics of the Kagome lattice to surface electromagnetic waves and experimentally show a Kagome surface-wave PC. Under the protection of first-order and second-order topologies, both robust edge modes and in-gap corner modes are observed. The robust transport of edge modes is demonstrated by high transmission through the waveguide with a sharp bend. The localized corner mode is found at the corner with one isolated rod when a triangle-shaped sample is constructed. Our work not only shows a platform to mimic the topological physics in classical wave systems, but also offers a potential application in designing high-performance photonic devices.

Second Chern crystals with inherently non-trivial topology.

Chern insulators have been generalized to many classical wave systems and thereby lead to many potential applications such as robust waveguides, quantum computation and high-performance lasers. However, the band structure of a material can be either topologically trivial or non-trivial, depending on how the crystal structure is designed. Here, we propose a second Chern crystal in a four-dimensional parameter space by introducing two extra synthetic translation dimensions. Since the topology of the bulk bands in the synthetic translation space is intrinsically non-trivial, our proposed four-dimensional crystal is guaranteed to be topologically non-trivial regardless of the crystal's detailed configuration. We derive the topologically protected modes on the lower dimensional boundaries of such a crystal via dimension reduction. Remarkably, we observe the one-dimensional gapless dislocation modes and confirm their robustness in experiments. Our findings provide novel perspectives on topologically non-trivial crystals and may inspire designs of classical wave devices.

Modulation of cellular metabolism by protein crotonylation regulates pancreatic cancer progression.

Protein lysine crotonylation has been recently identified as a vital posttranslational modification in cellular processes, particularly through the modification of histones. We show that lysine crotonylation is an important modification of the cytoplastic and mitochondria proteins. Enzymes in glycolysis, the tricarboxylic acid (TCA) cycle, fatty acid metabolism, glutamine metabolism, glutathione metabolism, the urea cycle, one-carbon metabolism, and mitochondrial fusion/fission dynamics are found to be extensively crotonylated in pancreatic cancer cells. This modulation is mainly controlled by a pair of crotonylation writers and erasers including CBP/p300, HDAC1, and HDAC3. The dynamic crotonylation of metabolic enzymes is involved in metabolism regulation, which is linked with tumor progression. Interestingly, the activation of MTHFD1 by decrotonylation at Lys354 and Lys553 promotes the development of pancreatic cancer by increasing resistance to ferroptosis. Our study suggests that crotonylation represents a metabolic regulatory mechanism in pancreatic cancer progression.