Application of CT and MRI images based on artificial intelligence to predict lymph node metastases in patients with oral squamous cell carcinoma: a subgroup meta-analysis.

Background: The performance of artificial intelligence (AI) in the prediction of lymph node (LN) metastasis in patients with oral squamous cell carcinoma (OSCC) has not been quantitatively evaluated. The purpose of this study was to conduct a systematic review and meta-analysis of published data on the diagnostic performance of CT and MRI based on AI algorithms for predicting LN metastases in patients with OSCC. Methods: We searched the Embase, PubMed (Medline), Web of Science, and Cochrane databases for studies on the use of AI in predicting LN metastasis in OSCC. Binary diagnostic accuracy data were extracted to obtain the outcomes of interest, namely, the area under the curve (AUC), sensitivity, and specificity, and compared the diagnostic performance of AI with that of radiologists. Subgroup analyses were performed with regard to different types of AI algorithms and imaging modalities. Results: Fourteen eligible studies were included in the meta-analysis. The AUC, sensitivity, and specificity of the AI models for the diagnosis of LN metastases were 0.92 (95% CI 0.89-0.94), 0.79 (95% CI 0.72-0.85), and 0.90 (95% CI 0.86-0.93), respectively. Promising diagnostic performance was observed in the subgroup analyses based on algorithm types [machine learning (ML) or deep learning (DL)] and imaging modalities (CT vs. MRI). The pooled diagnostic performance of AI was significantly better than that of experienced radiologists. Discussion: In conclusion, AI based on CT and MRI imaging has good diagnostic accuracy in predicting LN metastasis in patients with OSCC and thus has the potential for clinical application. Systematic Review Registration: https://www.crd.york.ac.uk/PROSPERO/#recordDetails, PROSPERO (No. CRD42024506159).

Structural basis of neuropeptide Y signaling through Y1 and Y2 receptors.

Neuropeptide Y (NPY), a 36-amino-acid peptide, functions as a neurotransmitter in both the central and peripheral nervous systems by activating the NPY receptor subfamily. Notably, NPY analogs display varying selectivity and exert diverse physiological effects through their interactions with this receptor family. [Pro34]-NPY and [Leu31, Pro34]-NPY, mainly acting on Y1R, reportedly increases blood pressure and postsynaptically potentiates the effect of other vasoactive substances above all, while N-terminal cleaved NPY variants in human body primary mediates angiogenesis and neurotransmitter release inhibition through Y2R. However, the recognition mechanisms of Y1R and Y2R with specific agonists remain elusive, thereby hindering subtype receptor-selective drug development. In this study, we report three cryo-electron microscopy (cryo-EM) structures of Gi2-coupled Y1R and Y2R in complexes with NPY, as well as Y1R bound to a selective agonist [Leu31, Pro34]-NPY. Combined with cell-based assays, our study not only reveals the conserved peptide-binding mode of NPY receptors but also identifies an additional sub-pocket that confers ligand selectivity. Moreover, our analysis of Y1R evolutionary dynamics suggests that this sub-pocket has undergone functional adaptive evolution across different species. Collectively, our findings shed light on the molecular underpinnings of neuropeptide recognition and receptor activation, and they present a promising avenue for the design of selective drugs targeting the NPY receptor family.

WNT16 as a promising biomarker for systemic lupus erythematosus and its role in regulating cell proliferation and apoptosis.

OBJECTIVES: To investigate the expression and function of WNT16, a member of the WNT family protein, in the context of systemic lupus erythematosus (SLE). METHODS: WNT16 expression was assessed in peripheral blood mononuclear cells (PBMCs) from 35 SLE patients and 25 healthy individuals using quantitative polymerase chain reaction. Additionally, serum WNT16 protein levels were quantified via enzyme-linked immunosorbent assay in 162 SLE patients, 96 healthy controls (HC), and disease controls comprised 154 individuals with rheumatoid arthritis (RA) and Sjögren's syndrome (SS). We investigated the associations between WNT16 protein levels and clinical manifestations, laboratory indices, and disease activity in SLE patients. Receiver operating characteristic (ROC) curve analysis was employed to evaluate the diagnostic potential of serum WNT16 for SLE. Furthermore, we performed a knockdown assay on Jeko-1 cells and assessed cell proliferation and apoptosis using Cell Counting Kit-8 and flow cytometry. RESULTS: WNT16 mRNA in SLE patients' PBMCs were significantly lower than those in HC. Furthermore, serum WNT16 in SLE patients were markedly reduced compared to HC, RA, and SS cohorts. ROC curve analysis indicated that plasma WNT16 levels could serve as a potential biomarker for SLE identification (AUC=0.809, SLE vs. HC; AUC=0.760, SLE vs. RA; AUC=0.710, SLE vs. SS). Notably, a weak positive correlation was observed between WNT16 protein and both alkaline phosphatase and lymphocyte percentages. Conversely, a weak negative correlation existed between WNT16 and low-density lipoprotein, neutrophil percentage, and the incidence of pleurisy and disease activity. Additionally, our study confirmed that WNT16 knockdown impairs cell proliferation and enhances apoptosis. CONCLUSIONS: Serum WNT16 levels effectively differentiate SLE patients from healthy controls and individuals with other autoimmune disorders. WNT16 serves as a potential biomarker with high sensitivity. The diminished expression of WNT16 in SLE may have a significant role in its pathogenesis through the regulation of cell proliferation and apoptosis.

Early patient-reported outcomes after robotic-assisted versus video-assisted thoracoscopic lobectomy.

BACKGROUND: Robotic-assisted thoracoscopic surgery (RATS) can achieve traditional clinical outcomes comparable to those of video-assisted thoracoscopic surgery (VATS). However, patient-reported outcomes (PROs) during the early period after RATS and VATS remain unclear. This study aimed to utilize longitudinal electronic PRO (ePRO) assessments to evaluate symptom burden and functional status between these approaches from patients' perspective. METHODS: This study comprised patients who underwent lobectomy via RATS or VATS for non-small cell lung cancer. We collected multiple-time-point PROs data from the prospective longitudinal study via an ePRO system. Symptom severity and function status were assessed using the perioperative symptom assessment for patients undergoing lung surgery and were analyzed between groups using linear mixed-effects models. RESULTS: Of the 164 patients, 42 underwent RATS and 122 underwent VATS. After propensity score matching (PSM), 42 RATS and 84 VATS exhibited similar baseline characteristics. During the 7-day postoperative period, participants underwent RATS reported milder pain (p = 0.014), coughing (p < 0.001), drowsiness (p = 0.001), and distress (p = 0.045) compared with those underwent VATS. Moreover, participants in RATS group showed less functional interference with walking (p < 0.001) and general activity (p < 0.001). RATS exhibited a shorter postoperative hospitalization (p = 0.021) but higher hospital cost (p < 0.001). Meanwhile, short-term clinical outcomes of operative time, dissected lymph node stations, chest tube drainage, and postoperative complication rates were comparable. CONCLUSION: PROs are important metrics for assessing patients' recovery after lobectomy. Compared with VATS, RATS may induce less symptom burden and better functional status for patients in the early postoperative period.

Cushioning mechanism of the metatarsals during landing for the skateboarding ollie maneuver.

Skateboarding is an Olympic event with frequent jumping and landing, where the cushioning effect by the foot structure (from the arch, metatarsals, etc.) and damping performance by sports equipment (shoes, insoles, etc.) can greatly affect an athlete's sports performance and lower the risk of limb injury. Skateboarding is characterized by the formation of a "man-shoe-skateboard system," which makes its foot cushioning mechanism different from those of other sports maneuvers, such as basketball vertical jump and gymnastics broad jump. Therefore, it is necessary to clarify the cushioning mechanism of the foot structure upon landing on a skateboard. To achieve this, a multibody finite element model of the right foot, shoe, and skateboard was created using Mimics, Geomagic, and ANSYS. Kinetic data from the ollie maneuver were used to determine the plantar pressure and Achilles tendon force at three characteristics (T1, T2, and T3). The stress and strain on the foot and metatarsals (MT1-5) were then simulated. The simulation results had an error of 6.98% compared to actual measurements. During landing, the force exerted on the internal soft tissues tends to increase. The stress and strain variations were highest on MT2, MT3, and MT4. Moreover, the torsion angle of MT1 was greater than those of the other metatarsals. Additionally, the displacements of MT2, MT3, and MT4 were higher than those of the other parts. This research shows that skateboarders need to absorb the ground reaction force through the movements of the MTs for ollie landing. The soft tissues, bones, and ligaments in the front foot may have high risks of injury. The developed model serves as a valuable tool for analyzing the foot mechanisms in skateboarding; furthermore, it is crucial to enhance cushioning for the front foot during the design of skateboard shoes to reduce potential injuries.

Cross-Linking CdSO4-Type Nets with Hexafluorosilicate Anions to Form an Ultramicroporous Material for Efficient C2H2/CO2 and C2H2/C2H4 Separation.

A 44.610.8 topology hybrid ultramicroporous material (HUM), {[Cu1.5F(SiF6)(L)2.5]·G}n, (L = 4,4'-bisimidazolylbiphenyl, G = guest molecules), 1, formed by cross-linking interpenetrated 3D four-connected CdSO4-type nets with hexafluorosilicate anions is synthesized and evaluated in the context of gas sorption and separation herein. 1 is the first HUM functionalized with two different types of fluorinated sites (SiF6 2- and F- anions) lining along the pore surface. The optimal pore size (≈5 Å) combining mixed and high-density electronegative fluorinated sites enable 1 to preferentially adsorb C2H2 over CO2 and C2H4 by hydrogen bonding interactions with a high C2H2 isosteric heat of adsorption (Qst) of ≈42.3 kJ mol-1 at zero loading. The pronounced discriminatory sorption behaviors lead to excellent separation performance for C2H2/CO2 and C2H2/C2H4 that surpasses many well-known sorbents. Dynamic breakthrough experiments are conducted to confirm the practical separation capability of 1, which reveal an impressive separation factor of 6.1 for equimolar C2H2/CO2 mixture. Furthermore, molecular simulation and density functional theory (DFT) calculations validate the strong binding of C2H2 stems from the chelating fix of C2H2 between SiF6 2- anion and coordinated F- anion.

The role of TGF-ß signaling in muscle atrophy, sarcopenia and cancer cachexia.

Skeletal muscle, comprising a significant proportion (40 to 50 percent) of total body weight in humans, plays a critical role in maintaining normal physiological conditions. Muscle atrophy occurs when the rate of protein degradation exceeds protein synthesis. Sarcopenia refers to age-related muscle atrophy, while cachexia represents a more complex form of muscle wasting associated with various diseases such as cancer, heart failure, and AIDS. Recent research has highlighted the involvement of signaling pathways, including IGF1-Akt-mTOR, MuRF1-MAFbx, and FOXO, in regulating the delicate balance between muscle protein synthesis and breakdown. Myostatin, a member of the TGF-ß superfamily, negatively regulates muscle growth and promotes muscle atrophy by activating Smad2 and Smad3. It also interacts with other signaling pathways in cachexia and sarcopenia. Inhibition of myostatin has emerged as a promising therapeutic approach for sarcopenia and cachexia. Additionally, other TGF-ß family members, such as TGF-ß1, activin A, and GDF11, have been implicated in the regulation of skeletal muscle mass. Furthermore, myostatin cooperates with these family members to impair muscle differentiation and contribute to muscle loss. This review provides an overview of the significance of myostatin and other TGF-ß signaling pathway members in muscular dystrophy, sarcopenia, and cachexia. It also discusses potential novel therapeutic strategies targeting myostatin and TGF-ß signaling for the treatment of muscle atrophy.

Glutamate rescues heat stress-induced apoptosis of Sertoli cells by enhancing the activity of antioxidant enzymes and activating the Trx1-Akt pathway in vitro.

Heat stress reduces the number of Sertoli cells, which is closely related to an imbalanced redox status. Glutamate functions to maintain the equilibrium of redox homeostasis. However, the role of glutamate in heat treated Sertoli cells remains unclear. Herein, Sertoli cells from 3-week-old piglets were treated at 44 °C for 30 min (heat stress). Glutamate levels increased significantly following heat stress treatment, followed by a gradual decrease during recovery, while glutathione (GSH) showed a gradual increase. The addition of exogenous glutamate (700 µM) to Sertoli cells before heat stress significantly reduced the heat stress-induced apoptosis rate, mediated by enhanced levels of antioxidant substances (superoxide dismutase (SOD), total antioxidant capacity (TAC), and GSH) and reduced levels of oxidative substances (reactive oxygen species (ROS) and malondialdehyde (MDA)). Glutamate addition to Sertoli cells before heat stress upregulated the levels of glutamate-cysteine ligase, modifier subunit (Gclm), glutathione synthetase (Gss), thioredoxin (Trx1) and B-cell leukemia/lymphoma 2 (Bcl-2), and the ratio of phosphorylated Akt (protein kinase B)/total Akt. However, it decreased the levels of Bcl2-associated X protein (Bax) and cleaved-caspase 3. Addition of the inhibitor of glutaminase (Gls1), Bptes (Bis-2-(5-phenylacetamido-1,3,4-thiadiazol-2-yl)ethyl sulfide, 30 µM)to Sertoli cells before heat stress reversed these effects. These results inferred that glutamate rescued heat stress-induced apoptosis in Sertoli cells by enhancing activity of antioxidant enzymes and activating the Trx1-Akt pathway. Thus, glutamate supplementation might represent a novel strategy to alleviate the negative effect of heat stress.

Multi-Relational Deep Hashing for Cross-Modal Search.

Deep cross-modal hashing retrieval has recently made significant progress. However, existing methods generally learn hash functions with pairwise or triplet supervisions, which involves learning the relevant information by splicing partial similarity between data pairs; notably, this approach only captures the data similarity locally and incompletely, resulting in sub-optimal retrieval performance. In this paper, we propose a novel Multi-Relational Deep Hashing (MRDH) approach, which can fully bridge the modality gap by comprehensively modeling the similarity relationship between data in different modalities. In more detail, to investigate the inter-modal relationships, we constrain the consistency of cross-modal pairwise similarities to maintain the semantic similarity across modalities. Moreover, to further capture complete similarity information, we design a new similarity metric, which we term cross-modal global similarity, by encouraging hash codes of similar data pairs from different modalities to approach a common center and hash codes for dissimilar pairs to converge to different centers. Adopting this approach enables our model to generate more discriminative hash codes. Extensive experiments on three benchmark datasets demonstrate the superiority of our method on cross-modal hashing retrieval.

Asynchronous Parallel Large-Scale Gaussian Process Regression.

Gaussian process regression (GPR) is an important nonparametric learning method in machine learning research with many real-world applications. It is well known that training large-scale GPR is a challenging task due to the required heavy computational cost and large volume memory. To address this challenging problem, in this article, we propose an asynchronous doubly stochastic gradient algorithm to handle the large-scale training of GPR. We formulate the GPR to a convex optimization problem, i.e., kernel ridge regression. After that, in order to efficiently solve this convex kernel problem, we first use the random feature mapping method to approximate the kernel model and then utilize two unbiased stochastic approximations, i.e., stochastic variance reduced gradient and stochastic coordinate descent, to update the solution asynchronously and in parallel. In this way, our algorithm scales well in both sample size and dimensionality, and speeds up the training computation. More importantly, we prove that our algorithm has a global linear convergence rate. Our experimental results on eight large-scale benchmark datasets with both regression and classification tasks show that the proposed algorithm outperforms the existing state-of-the-art GPR methods.

[Multi-index evaluation of different parts of Amomum villosum].

To investigate the quality differences between the seeds and husks of Amomum villosum and explore the rationality of using the seeds without husks, this study determined the content of protocatechuic acid, vanillic acid, epicatechin, quercitrin, volatile oil, water extract, and ethanol extract. The 2,2-diphenyl-1-picrylhydrazyl(DPPH), 2,2-azino-bis(3-ethylbenzothiazoline-6-sulfonic acid)(ABTS), and hydroxyl radical scavenging activities were determined to evaluate the antioxidant activities of seeds and husks. The quality differences between the seeds and husks were assessed through orthogonal partial least squares-discriminant analysis(OPLS-DA) and analytic hierarchy process(AHP) combined with the entropy weight method(EWM). Significant differences(P<0.05) were observed in all 10 indicators between the seeds and husks. The levels of epicatechin, quercetin, and volatile oil were higher in the seeds, whereas those of protocatechuic acid, vanillic acid, water extract, and ethanol extract were higher in the husks. The seeds showed stronger scavenging ability against DPPH and ABTS radicals, while the husks showed a stronger scavenging effect on hydroxyl radicals. OPLS-DA significantly discriminated between the seeds and husks. Furthermore, volatile oil, water extract, DPPH radical scavenging rate, quercitrin, ABTS radical scavenging rate, hydroxyl radical scavenging rate, and vanillic acid were selected as the main differential indicators by variable importance in projection(VIP). Comprehensive scores calculated by AHP combined with EWM indicated that the seeds were superior to husks in terms of overall quality. However, there are still some dominant components and a certain antioxidant effect in the husks. Therefore, it is suggested to using Amomi Fructus with a certain amount of husks or utilizing the husks for other purposes.

A novel FK506-loading mesoporous silica nanoparticle homing to lymph nodes for transplant rejection treatment.

Tacrolimus (FK506) is an effective therapeutic for transplant rejection in clinical practice, primarily inhibiting rejection by suppressing the activation and proliferation of allogeneic T cells in the lymph nodes (LNs). However, conventional administration methods face challenges in directly delivering free FK506 to the LNs. In this study, we introduce a novel LN-targeted delivery system based on mesoporous silica nanoparticles (MSNs-FK506-MECA79). These particles were designed to selectively target high endothelial venules in LNs; this was achieved through surface modification with MECA79 antibodies. Their mean size and zeta potential were 201.18 ± 5.98 nm and - 16.12 ± 0.36 mV, respectively. Our findings showed that MSNs-FK506-MECA79 could accumulate in LNs and increase the local concentration of FK506 from 28.02 ± 7.71 ng/g to 123.81 ± 76.76 ng/g compared with the free FK506 treatment group. Subsequently, the therapeutic efficacy of MSNs-FK506-MECA79 was evaluated in a skin transplantation model. The treatment with MSNs-FK506-MECA79 could lead to a decrease in the infiltration of T cells in the grafts, a reduction in the grade of rejection, and a significant prolongation of survival. Consequently, this study presents a promising strategy for the active LN-targeted delivery of FK506 and improving the immunotherapeutic effects on transplant rejection.

Unsupervised Out-of-Distribution Object Detection via PCA-Driven Dynamic Prototype Enhancement.

To promote the application of object detectors in real scenes, out-of-distribution object detection (OOD-OD) is proposed to distinguish whether detected objects belong to the ones that are unseen during training or not. One of the key challenges is that detectors lack unknown data for supervision, and as a result, can produce overconfident detection results on OOD data. Thus, this task requires to synthesize OOD data for training, which achieves the goal of enhancing the ability of localizing and discriminating OOD objects. In this paper, we propose a novel method, i.e., PCA-Driven dynamic prototype enhancement, to explore exploiting Principal Component Analysis (PCA) to extract simulative OOD data for training and obtain dynamic prototypes that are related to the current input and are helpful for boosting the discrimination ability. Concretely, the last few principal components of the backbone features are utilized to calculate an OOD map that involves plentiful information that deviates from the correlation distribution of the input. The OOD map is further used to extract simulative OOD data for training, which alleviates the impact of lacking unknown data. Besides, for in-distribution (ID) data, the category-level semantic information of objects between the backbone features and the high-level features should be kept consistent. To this end, we utilize the residual principal components to extract dynamic prototypes that reflect the semantic information of the current backbone features. Next, we define a contrastive loss to leverage these prototypes to enlarge the semantic gap between the simulative OOD data and the features from the residual principal components, which improves the ability of discriminating OOD objects. In the experiments, we separately verify our method on OOD-OD and incremental object detection. The significant performance gains demonstrate the superiorities of our method.

Causality-Invariant Interactive Mining for Cross-Modal Similarity Learning.

In the real world, how to effectively learn consistent similarity measurement across different modalities is essential. Most of the existing similarity learning methods cannot deal well with cross-modal data due to the modality gap and have obvious performance degeneration when applied to cross-modal data. To tackle this problem, we propose a novel cross-modal similarity learning method, called Causality-Invariant Interactive Mining (CIIM), that can effectively capture informative relationships among different samples and modalities to derive the modality-consistent feature embeddings in the unified metric space. Our CIIM tackles the modality gap from two aspects, i.e., sample-wise and feature-wise. Specifically, we start from the sample-wise view and learn the single-modality and hybrid-modality proxies for exploring the cross-modal similarity with the elaborate metric losses. In this way, sample-to-sample and sample-to-proxy correlations are both taken into consideration. Furthermore, we conduct the causal intervention to eliminate the modality bias and reconstruct the invariant causal embedding in the feature-wise aspect. To this end, we force the learned embeddings to satisfy the specific properties of our causal mechanism and derive the causality-invariant feature embeddings in the unified metric space. Extensive experiments on two cross-modality tasks demonstrate the superiority of our proposed method over the state-of-the-art methods.

Earliest Prepared core technology in Eurasia from Nihewan (China): Implications for early human abilities and dispersals in East Asia.

Organized flaking techniques to obtain predetermined stone tools have been traced back to the early Acheulean (also known as mode 2) in Africa and are seen as indicative of the emergence of advanced technical abilities and in-depth planning skills among early humans. Here, we report one of the earliest known examples of prepared core technology in the archaeological record, at the Cenjiawan (CJW) site in the Nihewan basin of China, dated 1.1 Mya. The operational schemes reconstructed from the CJW refit sets, together with shaping patterns observed in the retouched tools, suggest that Nihewan basin toolmakers had the technical abilities of mode 2 hominins, and developed different survival strategies to adapt to local raw materials and environments. This finding predates the previously earliest known prepared core technology from Eurasia by 0.3 My, and the earliest known mode 2 sites in East Asia by a similar amount of time, thus suggesting that hominins with advanced technologies may have migrated into high latitude East Asia as early as 1.1 Mya.

Identification of fangchinoline as a broad-spectrum enterovirus inhibitor through reporter virus based high-content screening.

Hand, foot, and mouth disease (HFMD) is a common pediatric illness mainly caused by enteroviruses, which are important human pathogens. Currently, there are no available antiviral agents for the therapy of enterovirus infection. In this study, an excellent high-content antiviral screening system utilizing the EV-A71-eGFP reporter virus was developed. Using this screening system, we screened a drug library containing 1042 natural compounds to identify potential EV-A71 inhibitors. Fangchinoline (FAN), a bis-benzylisoquinoline alkaloid, exhibits potential inhibitory effects against various enteroviruses that cause HFMD, such as EV-A71, CV-A10, CV-B3 and CV-A16. Further investigations revealed that FAN targets the early stage of the enterovirus life cycle. Through the selection of FAN-resistant EV-A71 viruses, we demonstrated that the VP1 protein could be a potential target of FAN, as two mutations in VP1 (E145G and V258I) resulted in viral resistance to FAN. Our research suggests that FAN is an efficient inhibitor of EV-A71 and has the potential to be a broad-spectrum antiviral drug against human enteroviruses.

Nanoparticle-based T cell immunoimaging and immunomodulatory for diagnosing and treating transplant rejection.

T cells serve a pivotal role in the rejection of transplants, both by directly attacking the graft and by recruiting other immune cells, which intensifies the rejection process. Therefore, monitoring T cells becomes crucial for early detection of transplant rejection, while targeted drug delivery specifically to T cells can significantly enhance the effectiveness of rejection therapy. However, regulating the activity of T cells within transplanted organs is challenging, and the prolonged use of immunosuppressive drugs is associated with notable side effects and complications. Functionalized nanoparticles offer a potential solution by targeting T cells within transplants or lymph nodes, thereby reducing the off-target effects and improving the long-term survival of the graft. In this review, we will provide an overview of recent advancements in T cell-targeted imaging molecular probes for diagnosing transplant rejection and the progress of T cell-regulating nanomedicines for treating transplant rejection. Additionally, we will discuss future directions and the challenges in clinical translation.

Class-Incremental Unsupervised Domain Adaptation via Pseudo-Label Distillation.

Class-Incremental Unsupervised Domain Adaptation (CI-UDA) requires the model can continually learn several steps containing unlabeled target domain samples, while the source-labeled dataset is available all the time. The key to tackling CI-UDA problem is to transfer domain-invariant knowledge from the source domain to the target domain, and preserve the knowledge of the previous steps in the continual adaptation process. However, existing methods introduce much biased source knowledge for the current step, causing negative transfer and unsatisfying performance. To tackle these problems, we propose a novel CI-UDA method named Pseudo-Label Distillation Continual Adaptation (PLDCA). We design Pseudo-Label Distillation module to leverage the discriminative information of the target domain to filter the biased knowledge at the class- and instance-level. In addition, Contrastive Alignment is proposed to reduce domain discrepancy by aligning the class-level feature representation of the confident target samples and the source domain, and exploit the robust feature representation of the unconfident target samples at the instance-level. Extensive experiments demonstrate the effectiveness and superiority of PLDCA. Code is available at code.

Initial Upper Palaeolithic material culture by 45,000 years ago at Shiyu in northern China.

The geographic expansion of Homo sapiens populations into southeastern Europe occurred by ∼47,000 years ago (∼47 ka), marked by Initial Upper Palaeolithic (IUP) technology. H. sapiens was present in western Siberia by ∼45 ka, and IUP industries indicate early entries by ∼50 ka in the Russian Altai and 46-45 ka in northern Mongolia. H. sapiens was in northeastern Asia by ∼40 ka, with a single IUP site in China dating to 43-41 ka. Here we describe an IUP assemblage from Shiyu in northern China, dating to ∼45 ka. Shiyu contains a stone tool assemblage produced by Levallois and Volumetric Blade Reduction methods, the long-distance transfer of obsidian from sources in China and the Russian Far East (800-1,000 km away), increased hunting skills denoted by the selective culling of adult equids and the recovery of tanged and hafted projectile points with evidence of impact fractures, and the presence of a worked bone tool and a shaped graphite disc. Shiyu exhibits a set of advanced cultural behaviours, and together with the recovery of a now-lost human cranial bone, the record supports an expansion of H. sapiens into eastern Asia by about 45 ka.