Identifying allergic-rhinitis-associated genes with random-walk-based method in PPI network.

Allergic rhinitis is a common allergic disease with a complex pathogenesis and many unresolved issues. Studies have shown that the incidence of allergic rhinitis is closely related to genetic factors, and research on the related genes could help further understand its pathogenesis and develop new treatment methods. In this study, 446 allergic rhinitis-related genes were obtained on the basis of the DisGeNET database. The protein-protein interaction network was searched using the random-walk-with-restart algorithm with these 446 genes as seed nodes to assess the linkages between other genes and allergic rhinitis. Then, this result was further examined by three screening tests, including permutation, interaction, and enrichment tests, which aimed to pick up genes that have strong and special associations with allergic rhinitis. 52 novel genes were finally obtained. The functional enrichment test confirmed their relationships to the biological processes and pathways related to allergic rhinitis. Furthermore, some genes were extensively analyzed to uncover their special or latent associations to allergic rhinitis, including IRAK2 and MAPK, which are involved in the pathogenesis of allergic rhinitis and the inhibition of allergic inflammation via the p38-MAPK pathway, respectively. The new found genes may help the following investigations for understanding the underlying molecular mechanisms of allergic rhinitis and developing effective treatments.

Future directions of noninvasive prediction of esophageal variceal bleeding: No worry about the present computed tomography inefficiency.

In this editorial, we comment on the minireview by Martino A, published in the recent issue of World Journal of Gastrointestinal Endoscopy 2023; 15 (12): 681-689. We focused mainly on the possibility of replacing the hepatic venous pressure gradient (HVPG) and endoscopy with noninvasive methods for predicting esophageal variceal bleeding. The risk factors for bleeding were the size of the varices, the red sign and the Child-Pugh score. The intrinsic core factor that drove these changes was the HVPG. Therefore, the present studies investigating noninvasive methods, including computed tomography, magnetic resonance imaging, elastography, and laboratory tests, are working on correlating imaging or serum marker data with intravenous pressure and clinical outcomes, such as bleeding. A single parameter is usually not enough to construct an efficient model. Therefore, multiple factors were used in most of the studies to construct predictive models. Encouraging results have been obtained, in which bleeding prediction was partly reached. However, these methods are not satisfactory enough to replace invasive methods, due to the many drawbacks of different studies. There is still plenty of room for future improvement. Prediction of the precise timing of bleeding using various models, and extracting the texture of variceal walls using high-definition imaging modalities to predict the red sign are interesting directions to lay investment on.

Exploring non-curative endoscopic submucosal dissection: Current treatment optimization and future indication expansion.

The increasing popularity of endoscopic submucosal dissection (ESD) as a treatment for early gastric cancer has highlighted the importance of quality assessment in achieving curative resections. This article emphasizes the significance of evaluating ESD quality, not only for curative cases but also for non-curative ones. Postoperative assessment relies on the endoscopic curability (eCura) classification, but management strategies for eCuraC-1 tumour with a positive horizontal margin are unclear. Current research primarily focuses on comparing additional surgical procedures in high-risk patients, while studies specifically targeting eCuraC-1 patients are limited. Exploring management strategies and follow-up outcomes for such cases could provide valuable insights. Furthermore, the application of molecular imaging using near-infrared fluorescent tracers holds promise for precise tumour diagnosis and navigation, potentially impacting the management of early-stage gastric cancer patients. Advancing research in these areas is essential for improving the overall efficacy of endoscopic techniques and refining treatment indications.

Identification of Protein-Protein Interaction Associated Functions Based on Gene Ontology.

Protein-protein interactions (PPIs) involve the physical or functional contact between two or more proteins. Generally, proteins that can interact with each other always have special relationships. Some previous studies have reported that gene ontology (GO) terms are related to the determination of PPIs, suggesting the special patterns on the GO terms of proteins in PPIs. In this study, we explored the special GO term patterns on human PPIs, trying to uncover the underlying functional mechanism of PPIs. The experimental validated human PPIs were retrieved from STRING database, which were termed as positive samples. Additionally, we randomly paired proteins occurring in positive samples, yielding lots of negative samples. A simple calculation was conducted to count the number of positive samples for each GO term pair, where proteins in samples were annotated by GO terms in the pair individually. The similar number for negative samples was also counted and further adjusted due to the great gap between the numbers of positive and negative samples. The difference of the above two numbers and the relative ratio compared with the number on positive samples were calculated. This ratio provided a precise evaluation of the occurrence of GO term pairs for positive samples and negative samples, indicating the latent GO term patterns for PPIs. Our analysis unveiled several nuclear biological processes, including gene transcription, cell proliferation, and nutrient metabolism, as key biological functions. Interactions between major proliferative or metabolic GO terms consistently correspond with significantly reported PPIs in recent literature.

[Health Assessment of the Stream Ecosystem in Beijing, China].

With economic development, the health of river ecosystems is becoming severely threatened because of the increasing effects of human activities on river ecosystems. Here, 101 sites along regional river systems in Beijing rivers were investigated from autumn 2020 to summer 2021. A total of 34 metrics, including aquatic organisms, hydrology, water quality, and habitat, were calculated to be the candidate indicators. Principal component and correlation analyses were used to select the core metrics from the candidate indicators, and the weight of each core metric was estimated using the entropy method. The integrated index of stream ecological health was constructed to assess the health condition of the Beijing rivers. The results of the PCA and correlation analyses revealed that eleven metrics were selected as the core metrics to construct the integrated index of stream ecological health, including water temperature, flow velocity, BOD5, NH4+-N, Cu, the density of phytoplankton and zooplankton, the Shannon-Wiener diversity index of macroinvertebrates and fish, the BMWP index, and the qualitative habitat evaluation index. According to the health assessment results, 4.95% of the sampling sites were healthy, 23.76% were subhealthy, and 71.29% were in a fair or below healthy state. The river health status showed strong spatial heterogeneity, and the river health statuses in the northern and western regions were relatively good, whereas the river health status in the central and southeastern regions were relatively poor. The results of four aspects stream ecosystem assessment showed that the overall water quality of the rivers was "subhealthy" and the aquatic organisms and habitat were "general poor," but the hydrology was "poor." The evaluation results of five water systems demonstrated that the Chaobai River had the best health status, followed by that of the Yongding River, Daqing River, and Jiyun River, and the Beiyun River had the worst health status. Maintaining river ecological baseflow, ensuring river system connectivity, and improving and restoring the river habitat environment are the key aspects of river ecological restoration and protection in Beijing in the future.

Application of noninvasive test (acoustic attenuation imaging and ultrasonic shear wave elastography) to grade nonalcoholic fatty liver disease: An observational study.

The aim of this study was to validate the diagnostic efficacy of acoustic attenuation imaging (ATI) and ultrasonic shear wave elastography (SWE) in classifying nonalcoholic fatty liver disease (NAFLD). A total of 100 patients with NAFLD were recruited from our hospital between January 2021 and December 2022. Patient demographics and clinical data were collected, and 2-dimensional ultrasound was used to screen patients based on liver echo characteristics. Patients without liver space-occupying lesions underwent routine ultrasound examinations. Imaging or serology was used to confirm the presence of fatty liver in patients or healthy individuals. Patients with alcoholic liver disease (alcohol equivalent content < 20 g/day for women, <30 g/day for men), as well as those with lenticular degeneration, total parenteral nutrition, autoimmune liver disease, drug-induced hepatitis, and viral hepatitis, were excluded from the study. Out of the 100 included patients, 24 had normal liver, 21 had mild fatty liver, 30 had moderate fatty liver, and 25 had severe fatty liver. There were age differences between the normal group and patients with mild fatty liver, and the average body mass index (BMI) varied across the 4 groups. As the severity of the disease increased, the average BMI also increased (P < .05). The ATI scores and SWE scores differed significantly among the different groups (P < .05), with both scores showing an upward trend as the fatty liver condition worsened. Correlation analysis revealed positive correlations between ATI and SWE scores and the degree of fatty liver (P < .05), positive correlations with BMI (P < .05), and negative correlations with high-density lipoprotein cholesterol expression (P < .05). The area under the curve (AUC) for the ATI score in diagnosing different degrees of fatty liver was > 0.750, and the AUC for the SWE score was also > 0.750. The AUC for SWE score in diagnosing different degrees of fatty liver ranged from 1.01 to 4.57, while the combined AUC for ATI and SWE scores was > 0.850, with respective cutoff values of 3.62, 5.72, and 7.57 based on the maximum approximate entry index. The combination of ATI and SWE has a significant impact on the grading diagnosis of NAFLD, and its application can be extended to clinical practice.

Printing 3D mesh-like grooves on zinc surface to enhance the stability of aqueous zinc ion batteries.

Aqueous zinc ion batteries (AZIBs) are receiving broad attention owing to their high safety and low cost. However, the high mechanical strength and irreversible growth of zinc dendrites limit the practical application of AZIBs. Herein, regular mesh-like gullies are built on the surface of zinc foil (M150 Zn) by using simple model pressing method and stainless steel mesh as a mold. Due to the charge-enrichment effect, zinc ion deposition and stripping will be preferentially carried out in the grooves to keep the outer surface flat. In addition, zinc is exposed to 002 crystal surface in the gully after being pressed, and the deposited zinc is more inclined to grow at a small angle, so that it has a sedimentary morphology parallel to the basement. Consequently, at a current density of 0.5 mA cm-2, the M150 zinc anode has a voltage hysteresis of only 35 mV and a cycle life of up to 400 h (relative to a zinc foil of 96 mV and 160 h). Even more imposing is that the full cell has a capacity retention of approximately 100% after 1000 cycles at 2 A g-1 and a specific capacity of almost 60 mAh g-1 when activated carbon is used as the cathode. It is a promising method to improve the stable cycle performance of AZIBs by using a simple method to realize the non-prominent dendrites on the surface of zinc electrode.

Nitrogen-rich Graphite Flake from Hemp as Anode Material for High Performance Lithium-ion Batteries.

Biomass-derived carbon (BC) has attracted extensive attention as anode material for lithium ion batteries (LiBs) due to its natural hierarchical porous structure and rich heteroatoms that can adsorb Li+ . However, the specific surface area of pure biomass carbon is generally small, so we can help NH3 and inorganic acid produced by urea decomposition to strip biomass, improve its specific surface area and enrich nitrogen elements. The nitrogen-rich graphite flake obtained by the above treatment of hemp is named NGF. The product that has a high nitrogen content of 10.12% has a high specific surface area of 1151.1 m2 g-1 . In the lithium ion battery test, the capacity of NGF is 806.6 mAh g-1 at 30 mA g-1 , which is twice than that of BC. NGF also showed excellent performance that is 429.2 mAh g-1 under high current testing at 2000 mA g-1 . The reaction process kinetics is analyzed and we found that the outstanding rate performance is attributed to the large-scale capacitance control. In addition, the results of the constant current intermittent titration test indicate that the diffusion coefficient of NGF is greater than that of BC. This work proposes a simple method of nitrogen-rich activated carbon, which has a significantly commercial prospect.

Daphnia magna uptake and excretion of luminescence-labelled polystyrene nanoparticle as visualized by high sensitivity real-time optical imaging.

The environmental and ecological consequences of nanoplastics (NPs) draw increasing research interests and social concerns. However, the in situ and real-time detection of NPs from living organisms and transferring media remains as a major technical obstacle for scientific investigation. Herein we report a novel time-gated imaging (TGI) strategy capable of real-time visualizing the intake of NPs by an individual living organism, which is based on the polystyrene NPs labelled with lanthanide up-conversion luminescence. The limit of detection (LOD) of the TGI apparatus was 600 pg (SNR = 3) in a field of view of 2.4 × 3.8 mm. Taking Daphnia magna as the aquatic model, we investigated the dynamics of uptake and accumulation of NPs (500 µg/L) for 24 h, and the subsequent excretion process (in clean medium) for 48 h, and quantitively analyzed the distribution and the overall mass of NPs deposited in D. magna. The uptake of NPs via filter-feeding occurred in a few minutes, whereas a longer accumulation was found, in a timescale of several hours. And similar behaviors (bi-phase elimination) were also seen in the excretion, indicating the migration of NPs into the circulatory system. The average mass of NPs accumulated in an individual D. magna was ∼12 ng after 24 h exposure, indicating that D. magna as a filter feeder tends to retain NPs. The observed NPs accumulation in D. magna exemplifies the potential risk of aquatic ecosystem on exposure to NP contamination.

Integrating Genetics, Transcriptomics, and Proteomics in Lung Tissue to Investigate Chronic Obstructive Pulmonary Disease.

The integration of transcriptomic and proteomic data from lung tissue with chronic obstructive pulmonary disease (COPD)-associated genetic variants could provide insight into the biological mechanisms of COPD. Here, we assessed associations between lung transcriptomics and proteomics with COPD in 98 subjects from the Lung Tissue Research Consortium. Low correlations between transcriptomics and proteomics were generally observed, but higher correlations were found for COPD-associated proteins. We integrated COPD risk SNPs or SNPs near COPD-associated proteins with lung transcripts and proteins to identify regulatory cis-quantitative trait loci (QTLs). Significant expression QTLs (eQTLs) and protein QTLs (pQTLs) were found regulating multiple COPD-associated biomarkers. We investigated mediated associations from significant pQTLs through transcripts to protein levels of COPD-associated proteins. We also attempted to identify colocalized effects between COPD genome-wide association studies and eQTL and pQTL signals. Evidence was found for colocalization between COPD genome-wide association study signals and a pQTL for RHOB and an eQTL for DSP. We applied weighted gene co-expression network analysis to find consensus COPD-associated network modules. Two network modules generated by consensus weighted gene co-expression network analysis were associated with COPD with a false discovery rate lower than 0.05. One network module is related to the catenin complex, and the other module is related to plasma membrane components. In summary, multiple cis-acting determinants of transcripts and proteins associated with COPD were identified. Colocalization analysis, mediation analysis, and correlation-based network analysis of multiple omics data may identify key genes and proteins that work together to influence COPD pathogenesis.

UV-derived double crosslinked PEO-based solid polymer electrolyte for room temperature.

The low ionic conductivity at room temperature and poor dimensional stability at high temperature of polyethylene oxide (PEO)-based solid electrolytes greatly limit the development and utilization of solid polymer electrolytes (SPEs). To reconcile the contradiction between electrochemical performance and mechanical strength of PEO-based SPEs, a cross-linking structure with active -CH2CH2O- soft chains that doped with rigid segments is designed and prepared through a method of green ultraviolet irradiation without solvent. The obtained solid film shows a high ionic conductivity of 0.2 mS·cm-1 and an ionic transference number of 0.51 at room temperature. The activation energy value of 1.92 kJ·mol-1 gives evidence for a favorable migration mechanism of PTP-SPE. A combination of flexibility and strength can be realized by molecular structure design with a tensile elongation of 40%. The reversible overpotential in galvanostatic cycling over 500 h of a Li||Li symmetrical cell indicates that the compact PTP-SPE can inhibit the formation of lithium dendrites. This work provides a new strategy for designing high-performance composite solid electrolytes at room temperature.

Machine Learning for Predicting Distant Metastasis of Medullary Thyroid Carcinoma Using the SEER Database.

Objectives: We aimed to establish an effective machine learning (ML) model for predicting the risk of distant metastasis (DM) in medullary thyroid carcinoma (MTC). Methods: Demographic data of MTC patients were extracted from the Surveillance, Epidemiology, and End Results (SEER) database of the National Institutes of Health between 2004 and 2015 to develop six ML algorithm models. Models were evaluated based on accuracy, precision, recall rate, F1-score, and area under the receiver operating characteristic curve (AUC). The association between clinicopathological characteristics and target variables was interpreted. Analyses were performed using traditional logistic regression (LR). Results: In total, 2049 patients were included and 138 developed DM. Multivariable LR showed that age, sex, tumor size, extrathyroidal extension, and lymph node metastasis were predictive features for DM in MTC. Among the six ML models, the random forest (RF) had the best predictability in assessing the risk of DM in MTC, with an accuracy, precision, recall rate, F1-score, and AUC higher than those of the traditional binary LR model. Conclusion: RF was superior to traditional LR in predicting the risk of DM in MTC and can provide a valuable reference for clinicians in decision-making.

Identifying anal and cervical tumorigenesis-associated methylation signaling with machine learning methods.

Cervical and anal carcinoma are neoplastic diseases with various intraepithelial neoplasia stages. The underlying mechanisms for cancer initiation and progression have not been fully revealed. DNA methylation has been shown to be aberrantly regulated during tumorigenesis in anal and cervical carcinoma, revealing the important roles of DNA methylation signaling as a biomarker to distinguish cancer stages in clinics. In this research, several machine learning methods were used to analyze the methylation profiles on anal and cervical carcinoma samples, which were divided into three classes representing various stages of tumor progression. Advanced feature selection methods, including Boruta, LASSO, LightGBM, and MCFS, were used to select methylation features that are highly correlated with cancer progression. Some methylation probes including cg01550828 and its corresponding gene RNF168 have been reported to be associated with human papilloma virus-related anal cancer. As for biomarkers for cervical carcinoma, cg27012396 and its functional gene HDAC4 were confirmed to regulate the glycolysis and survival of hypoxic tumor cells in cervical carcinoma. Furthermore, we developed effective classifiers for identifying various tumor stages and derived classification rules that reflect the quantitative impact of methylation on tumorigenesis. The current study identified methylation signals associated with the development of cervical and anal carcinoma at qualitative and quantitative levels using advanced machine learning methods.

Subcellular Localization Prediction of Human Proteins Using Multifeature Selection Methods.

Subcellular localization attempts to assign proteins to one of the cell compartments that performs specific biological functions. Finding the link between proteins, biological functions, and subcellular localization is an effective way to investigate the general organization of living cells in a systematic manner. However, determining the subcellular localization of proteins by traditional experimental approaches is difficult. Here, protein-protein interaction networks, functional enrichment on gene ontology and pathway, and a set of proteins having confirmed subcellular localization were applied to build prediction models for human protein subcellular localizations. To build an effective predictive model, we employed a variety of robust machine learning algorithms, including Boruta feature selection, minimum redundancy maximum relevance, Monte Carlo feature selection, and LightGBM. Then, the incremental feature selection method with random forest and support vector machine was used to discover the essential features. Furthermore, 38 key features were determined by integrating results of different feature selection methods, which may provide critical insights into the subcellular location of proteins. Their biological functions of subcellular localizations were discussed according to recent publications. In summary, our computational framework can help advance the understanding of subcellular localization prediction techniques and provide a new perspective to investigate the patterns of protein subcellular localization and their biological importance.

Identification of protein-protein interaction associated functions based on gene ontology and KEGG pathway.

Protein-protein interactions (PPIs) are extremely important for gaining mechanistic insights into the functional organization of the proteome. The resolution of PPI functions can help in the identification of novel diagnostic and therapeutic targets with medical utility, thus facilitating the development of new medications. However, the traditional methods for resolving PPI functions are mainly experimental methods, such as co-immunoprecipitation, pull-down assays, cross-linking, label transfer, and far-Western blot analysis, that are not only expensive but also time-consuming. In this study, we constructed an integrated feature selection scheme for the large-scale selection of the relevant functions of PPIs by using the Gene Ontology and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway annotations of PPI participants. First, we encoded the proteins in each PPI with their gene ontologies and KEGG pathways. Then, the encoded protein features were refined as features of both positive and negative PPIs. Subsequently, Boruta was used for the initial filtering of features to obtain 5684 features. Three feature ranking algorithms, namely, least absolute shrinkage and selection operator, light gradient boosting machine, and max-relevance and min-redundancy, were applied to evaluate feature importance. Finally, the top-ranked features derived from multiple datasets were comprehensively evaluated, and the intersection of results mined by three feature ranking algorithms was taken to identify the features with high correlation with PPIs. Some functional terms were identified in our study, including cytokine-cytokine receptor interaction (hsa04060), intrinsic component of membrane (GO:0031224), and protein-binding biological process (GO:0005515). Our newly proposed integrated computational approach offers a novel perspective of the large-scale mining of biological functions linked to PPI.

Identification of cortical interneuron cell markers in mouse embryos based on machine learning analysis of single-cell transcriptomics.

Mammalian cortical interneurons (CINs) could be classified into more than two dozen cell types that possess diverse electrophysiological and molecular characteristics, and participate in various essential biological processes in the human neural system. However, the mechanism to generate diversity in CINs remains controversial. This study aims to predict CIN diversity in mouse embryo by using single-cell transcriptomics and the machine learning methods. Data of 2,669 single-cell transcriptome sequencing results are employed. The 2,669 cells are classified into three categories, caudal ganglionic eminence (CGE) cells, dorsal medial ganglionic eminence (dMGE) cells, and ventral medial ganglionic eminence (vMGE) cells, corresponding to the three regions in the mouse subpallium where the cells are collected. Such transcriptomic profiles were first analyzed by the minimum redundancy and maximum relevance method. A feature list was obtained, which was further fed into the incremental feature selection, incorporating two classification algorithms (random forest and repeated incremental pruning to produce error reduction), to extract key genes and construct powerful classifiers and classification rules. The optimal classifier could achieve an MCC of 0.725, and category-specified prediction accuracies of 0.958, 0.760, and 0.737 for the CGE, dMGE, and vMGE cells, respectively. The related genes and rules may provide helpful information for deepening the understanding of CIN diversity.

[Optimization of preparation technology for Acanthopanax senticosus total saponins microemulsion by D-optimal mixture design and intestinal absorption characteristics].

Following the preparation of Acanthopanax senticosus total saponins microemulsion, the formulation and preparation technology were optimized and the quality was evaluated. The absorption characteristics of A. senticosus total saponins microemulsion by the self-microemulsifying drug delivery system(SMEDDS) were investigated in the unidirectional intestinal perfusion model in vivo. The oil phase, mass ratio(K_m), number of revolutions, and drug concentration were subjected to single-factor investigation with the area of pseudo-ternary phase diagram as the index. The process was optimized by D-optimal mixture design with the particle size as the index, and then the appearance, morphology, and particle size were investigated. The mass concentrations of eleutherosides B and E in the microemulsion were determined. The results showed that the optimum formulation of A. senticosus total saponins microemulsion was determined as follows: 20.8% of water phase, 31.2% of isopropyl palmitate, and 48.0% of soybean phospholipid and absolute ethanol(K_m=1∶1). As revealed by the observation under a transmission electron microscope, the microemulsion exhibited homogeneous dispersion and was a spherical emulsion droplet in the water-in-oil type. At room temperature, the pH value was 5.19, the refractive index 1.416 5, the average particle size(26.47±0.04)nm, and the polydispersity index(PDI) 0.118±0.03. The content of the eleutherosides B and E was 0.038 9 and 0.166 4 mg·mL~(-1), respectively. The preliminary stability study showed that the solution was clear and transparent within 30 d, without stratification or content change, indicating good stability. The absorption of microemulsion in each intestinal segment was significantly improved as compared with that of the A. senticosus total saponins, with the best absorption effect detected in the ileum, which has laid a foundation for further development and utilization of A. senticosus.

Screening gene signatures for clinical response subtypes of lung transplantation.

Lung is the most important organ in the human respiratory system, whose normal functions are quite essential for human beings. Under certain pathological conditions, the normal lung functions could no longer be maintained in patients, and lung transplantation is generally applied to ease patients' breathing and prolong their lives. However, several risk factors exist during and after lung transplantation, including bleeding, infection, and transplant rejections. In particular, transplant rejections are difficult to predict or prevent, leading to the most dangerous complications and severe status in patients undergoing lung transplantation. Given that most common monitoring and validation methods for lung transplantation rejections may take quite a long time and have low reproducibility, new technologies and methods are required to improve the efficacy and accuracy of rejection monitoring after lung transplantation. Recently, one previous study set up the gene expression profiles of patients who underwent lung transplantation. However, it did not provide a tool to predict lung transplantation responses. Here, a further deep investigation was conducted on such profiling data. A computational framework, incorporating several machine learning algorithms, such as feature selection methods and classification algorithms, was built to establish an effective prediction model distinguishing patient into different clinical subgroups, corresponding to different rejection responses after lung transplantation. Furthermore, the framework also screened essential genes with functional enrichments and create quantitative rules for the distinction of patients with different rejection responses to lung transplantation. The outcome of this contribution could provide guidelines for clinical treatment of each rejection subtype and contribute to the revealing of complicated rejection mechanisms of lung transplantation.

Artificial intelligence for diagnosing microvessels of precancerous lesions and superficial esophageal squamous cell carcinomas: a multicenter study.

BACKGROUND: Intrapapillary capillary loop (IPCL) is an important factor for predicting invasion depth of esophageal squamous cell carcinoma (ESCC). The invasion depth is closely related to the selection of treatment strategy. However, diagnosis of IPCLs is complicated and subject to interobserver variability. This study aimed to develop an artificial intelligence (AI) system to predict IPCLs subtypes of precancerous lesions and superficial ESCC. METHODS: Images of magnifying endoscopy with narrow band imaging from three hospitals were collected retrospectively. IPCLs subtypes were annotated on images by expert endoscopists according to Japanese Endoscopic Society classification. The performance of the AI system was evaluated using internal and external validation datasets (IVD and EVD) and compared with that of the 11 endoscopists. RESULTS: A total of 7094 images from 685 patients were used to train and validate the AI system. The combined accuracy of the AI system for diagnosing IPCLs subtypes in IVD and EVD was 91.3% and 89.8%, respectively. The AI system achieved better performance than endoscopists in predicting IPCLs subtypes and invasion depth. The ability of junior endoscopists to diagnose IPCLs subtypes (combined accuracy: 84.7% vs 78.2%, P < 0.0001) and invasion depth (combined accuracy: 74.4% vs 67.9%, P < 0.0001) were significantly improved with AI system assistance. Although there was no significant differences, the performance of senior endoscopists was slightly elevated. CONCLUSIONS: The proposed AI system could improve the diagnostic ability of endoscopists to predict IPCLs classification of precancerous lesions and superficial ESCC.