Impaired olfactory identification in dementia-free individuals is associated with the functional abnormality of the precuneus.

OBJECTIVE: Olfactory dysfunction indicates a higher risk of developing dementia. However, the potential structural and functional changes are still largely unknown. METHODS: A total of 236 participants were enrolled, including 45 Alzheimer's disease (AD) individuals and 191dementia-free individuals. Detailed study methods, comprising neuropsychological assessment and olfactory identification test (University of Pennsylvania smell identification test, UPSIT), as well as structural and functional magnetic resonance imaging (MRI) were applied in this research. The dementia-free individuals were divided into two sub-groups based on olfactory score: dementia-free with olfactory dysfunction (DF-OD) sub-group and dementia-free without olfactory dysfunction (DF-NOD) sub-group. The results were analyzed for subsequent intergroup comparisons and correlations. The cognitive assessment was conducted again three years later. RESULTS: (i) At dementia-free stage, there was a positive correlation between olfactory score and cognitive function. (ii) In dementia-free group, the volume of crucial brain structures involved in olfactory recognition and processing (such as amygdala, entorhinal cortex and basal forebrain volumes) are positively associated with olfactory score. (iii) Compared to the DF-NOD group, the DF-OD group showed a significant reduction in olfactory network (ON) function. (iv) Compared to DF-NOD group, there were significant functional connectivity (FC) decline between PCun_L(R)_4_1 in the precuneus of posterior default mode network (pDMN) and the salience network (SN) in DF-OD group, and the FC values decreased with falling olfactory scores. Moreover, in DF-OD group, the noteworthy reduction in FC were observed between PCun_L(R)_4_1 and amygdala, which was a crucial component of ON. (v) The AD conversion rate of DF-OD was 29.41%, while the DF-NOD group was 12.50%. The structural and functional changes in the precuneus were also observed in AD and were more severe. CONCLUSIONS: In addition to the olfactory circuit, the precuneus is a critical structure in the odor identification process, whose abnormal function underlies the olfactory identification impairment of dementia-free individuals.

Detection and characterization of novel luchacoviruses, genus Alphacoronavirus, in saliva and feces of meso-carnivores in the northeastern United States.

IMPORTANCE: Several coronaviruses (CoVs) have been detected in domesticated, farmed, and wild meso-carnivores, causing a wide range of diseases and infecting diverse species, highlighting their important but understudied role in the epidemiology of these viruses. Assessing the viral diversity hosted in wildlife species is essential to understand their significance in the cross-species transmission of CoVs. Our focus here was on CoV discovery in meso-carnivores in the Northeast United States as a potential "hotspot" area with high density of humans and urban wildlife. This study identifies novel alphacoronaviruses circulating in multiple free-ranging wild and domestic species in this area and explores their potential epidemiological importance based on regions of the Spike gene, which are relevant for virus-host interactions.

Verteporfin regulates corneal neovascularization through inhibition of YAP protein activation.

Corneal neovascularization (CNV) is a vision-threatening disease that is becoming a growing public health concern. While Yes-associated protein (YAP) plays a critical role in neovascular disease and allow for the sprouting angiogenesis. Verteporfin (VP) is a classical inhibitor of the YAP-TEAD complex, which is used for clinical treatment of neovascular macular degeneration through photodynamic therapy. The purpose of this study is to explore the effect of verteporfin (VP) on the inhibition of CNV and its potential mechanism. Rat CNV model were established by suturing in the central cornea and randomly divided into three groups (control, CNV and VP group). Neovascularization was observed by slit lamp to extend along the corneal limbus to the suture line. RNA-sequencing was used to reveal the related pathways on the CNV and the results revealed the vasculature development process and genes related with angiogenesis in CNV. In CNV group, we detected the nuclear translocation of YAP and the expression of CD31 in corneal neovascular endothelial cells through immunofluorescence. After the application of VP, the proliferation, migration and the tube formation of HUVECs were significantly inhibited. Furthermore, VP showed the CNV inhibition by tail vein injection without photoactivation. Then we found that the expression of phosphorylated YAP significantly decreased, and its downstream target protein connective tissue growth factor (CTGF) increased in the CNV group, while the expression was just opposite in other groups. Besides, both the expression of vascular endothelial growth factor receptor 2 (VEGFR2) and cofilin significantly increased in CNV group, and decreased after VP treatment. Therefore, we conclude that Verteporfin could significantly inhibited the CNV without photoactivation by regulating the activation of YAP.

Individualized small bowel preparation for computed tomography enterography: A prospective randomized controlled trial.

BACKGROUND AND AIM: The study aims to evaluate the feasibility of body mass index (BMI)-based individualized small bowel preparation for computed tomography enterography (CTE). METHODS: In this prospective randomized controlled study, patients undergoing CTE were randomly assigned to the individualized group or standardized group. Those in individualized group were given different volumes of mannitol solution based on BMI (1000 mL for patients with BMI < 18.5 kg/m2, 1500 mL for patients with 18.5 kg/m2 ≤ BMI < 25 kg/m2 and 2000 mL for patients with BMI ≥ 25 kg/m2) while patients in the standardized group were all asked to consume 1500-mL mannitol solution. CTE images were reviewed by two experienced radiologists blindly. Each segment of the small bowel was assessed for small bowel image quality and disease detection rates. Patients were invited to record a diary regarding adverse events and acceptance. RESULTS: A total of 203 patients were enrolled and randomly divided into two groups. For patients with BMI < 18.5 kg/m2, 1000-mL mannitol solution permitted a significantly lower rate of flatulence (P = 0.045) and defecating frequency (P = 0.011) as well as higher acceptance score (P = 0.015), but did not affect bowel image quality and diseases detection compared with conventional dosage. For patients with BMI ≥ 25 kg/m2, 2000-mL mannitol solution provided better overall image quality (P = 0.033) but comparable rates of adverse events and patients' acceptance compared with conventional dosage. CONCLUSIONS: Individualized bowel preparation could achieve both satisfactory image quality and patients' acceptance thus might be an acceptable alternative in CTE.

Genome-wide identification and expression profiles of the Phytophthora infestans responsive CYPome (cytochrome P450 complement) in Solanum tuberosum.

Cytochrome P450s represent one of the largest protein families across all domains of life. In plants, biotic stress can regulate the expression of some P450 genes. However, the CYPome (cytochrome P450 complement) in Solanum tuberosum and its response to Phytophthora infestans infection remains unrevealed. In this study, 488 P450 genes were identified from potato genome, which can be divided into 41 families and 57 subfamilies. Responding to the infection of P. infestans, 375 potato P450 genes were expressed in late blight resistant or susceptible cultivars. A total of 14 P450 genes were identified as resistant related candidates, and 81 P450 genes were identified as late blight responsive candidates. Several phytohormone biosynthesis, brassinosteroid biosynthesis, and phenylpropanoid biosynthesis involved P450 genes were differentially expressed during the potato-pathogen interactions. This study firstly reported the CYPome in S. tuberosum, and characterized the expression patterns of these P450 genes during the infection of P. infestans.

Ankle reflex and neurological symptom score: a primary level screening method for diabetic peripheral neuropathy.

Herein, we aimed to develop an easily available and efficient screening method for diabetic peripheral neuropathy (DPN) suitable for primary care settings, emphasizing simplicity, speed, and accuracy. Nerve conduction studies were conducted on 214 patients with diabetes, encompassing the outcomes of five distinct assessments: diabetic neuropathy symptom (DNS), vibration perception threshold (VPT), and nerve screening. The diagnostic accuracy of the VPT and nerve screening was evaluated by comparing them with that of the nerve conduction study. To assess diagnostic efficacy, various combinations were examined, including DNS combined with VPT, pain, temperature, touch, and ankle reflex. The diagnostic performance of DNS was superior to that of the five neurological screening items and VPT, with sensitivity, specificity, and accuracy of 0.68, 0.81, and 0.73, respectively. Among the two combined methods, "DNS + ankle reflex" was identified as having the highest diagnostic value, with an area under the curve, a sensitivity, a specificity, and an accuracy of 0.81, 0.89, 0.70, and 0.80, respectively. Furthermore, a combination of "DNS + ankle reflex + touch + pain + VPT" achieved the best performance among the five combinations, with an area under the curve, sensitivity, specificity, and accuracy of 0.85, 0.93, 0.68, and 0.81, respectively. The combination of DNS, ankle reflex, touch, pain, and VPT methods showed the highest diagnostic value for DPN. However, considering factors including accuracy, time, and economic cost, we recommend using a simpler combination of DNS and ankle reflex for large-scale screening of patients with DPN.

Based on HPLC and HS-GC-IMS Techniques, the Changes in the Internal Chemical Components of Schisandra chinensis (Turcz.) Baill. Fruit at Different Harvesting Periods Were Analyzed.

Schisandra chinensis, as a traditional Chinese herbal medicine, has clear pharmacological effects such as treating asthma, protecting nerves and blood vessels, and having anti-inflammatory properties. Although the Schisandra chinensis fruit contain multiple active components, the lignans have been widely studied as the primary pharmacologically active compound. The volatile chemical components of Schisandra chinensis include a large amount of terpenes, which have been proven to have broad pharmacological activities. However, when to harvest to ensure the highest accumulation of pharmacologically active components in Schisandra chinensis fruits is a critical issue. The Schisandra chinensis fruit trees in the resource nursery were all planted in 2019 and began bearing fruit in 2021. Their nutritional status and tree vigor remain consistently good. The content of lignans and organic acids in the fruits of Schisandra chinensis over seven different harvest periods was tested, and the results of high-performance liquid chromatography (HPLC) indicated that the lignan content was higher, at 35 mg/g, in late July, and the organic acid content was higher, at 72.34 mg/g, in early September. If lignans and organic acids are to be selected as raw materials for pharmacological development, the harvest can be carried out at this stage. Using HS-GC-IMS technology, a total of 67 volatile flavor substances were detected, and the fingerprint of the volatile flavor substances in the different picking periods was established. It was shown by the results that the content of volatile flavor substances was the highest in early August, and 16 flavor substances were selected by odor activity value (OAV). The variable importance in projection (VIP) values of 16 substances were further screened, and terpinolene was identified as the key volatile flavor substance that caused the aroma characteristics of Schisandra chinensis fruit at different harvesting periods. If the aroma component content of Schisandra chinensis fruit is planned to be used as raw material for development and utilization, then early August, when the aroma component content is higher, should be chosen as the time for harvest. This study provides a theoretical basis for the suitable harvesting time of Schisandra chinensis for different uses, and promotes the high-quality development of the Schisandra chinensis industry.

Coreference Resolution Based on High-Dimensional Multi-Scale Information.

Coreference resolution is a key task in Natural Language Processing. It is difficult to evaluate the similarity of long-span texts, which makes text-level encoding somewhat challenging. This paper first compares the impact of commonly used methods to improve the global information collection ability of the model on the BERT encoding performance. Based on this, a multi-scale context information module is designed to improve the applicability of the BERT encoding model under different text spans. In addition, improving linear separability through dimension expansion. Finally, cross-entropy loss is used as the loss function. After adding BERT and span BERT to the module designed in this article, F1 increased by 0.5% and 0.2%, respectively.

Tunneling nanotubes-based intercellular mitochondrial trafficking as a novel therapeutic target in dry eye.

Cell-to-cell mitochondria transfer via tunneling nanotubes (TNTs) has recently been revealed as a spontaneous way to protect damaged cells. Previously, we have reported mesenchymal stem cells (MSCs) can rescue retinal ganglion cell and corneal epithelium through intercellular mitochondrial trafficking. Mitochondrial damage and oxidative stress in corneal epithelial cells are vital in dry eye disease (DED). However, whether intercellular mitochondrial transfer is involved in the pathological and repair process of DED is currently unknown. Therefore, in this study, we designed a coculture system to evaluate the role of intercellular mitochondrial transfer between human corneal epithelial cells (CEC) in DED. In addition, we successfully discovered the ROCK inhibitor, Y-27632 as an intensifier to improve the efficiency of intercellular mitochondrial transport. As expected, the enhanced mitochondrial transfer promotes the regeneration of CECs. Moreover, through further exploration of mechanisms, it was demonstrated that F-actin-mediated cell morphological changes and cytoskeletal remodeling may be potential mechanisms for Y-27632 to induce mitochondrial metastasis. In conclusion, we established a new method for cell repair in DED that healthy CEC offered mitochondria to damaged CEC, providing a new insight into the cellular mechanism of corneal epithelium homeostatic regenerative therapeutics in DED.

Does multidimensional daily information predict the onset of myopia? A 1-year prospective cohort study.

PURPOSE: This study aimed to develop an interpretable machine learning model to predict the onset of myopia based on individual daily information. METHOD: This study was a prospective cohort study. At baseline, non-myopia children aged 6-13 years old were recruited, and individual data were collected through interviewing students and parents. One year after baseline, the incidence of myopia was evaluated based on visual acuity test and cycloplegic refraction measurement. Five algorithms, Random Forest, Support Vector Machines, Gradient Boosting Decision Tree, CatBoost and Logistic Regression were utilized to develop different models and their performance was validated by area under curve (AUC). Shapley Additive exPlanations was applied to interpret the model output on the individual and global level. RESULT: Of 2221 children, 260 (11.7%) developed myopia in 1 year. In univariable analysis, 26 features were associated with the myopia incidence. Catboost algorithm had the highest AUC of 0.951 in the model validation. The top 3 features for predicting myopia were parental myopia, grade and frequency of eye fatigue. A compact model using only 10 features was validated with an AUC of 0.891. CONCLUSION: The daily information contributed reliable predictors for childhood's myopia onset. The interpretable Catboost model presented the best prediction performance. Oversampling technology greatly improved model performance. This model could be a tool in myopia preventing and intervention that can help identify children who are at risk of myopia, and provide personalized prevention strategies based on contributions of risk factors to the individual prediction result.

Modeling of the Rating of Perceived Exertion Based on Heart Rate Using Machine Learning Methods.

Rating of perceived exertion (RPE) can serve as a more convenient and economical alternative to heart rate (HR) for exercise intensity control. This study aims to explore the influence of factors, such as indicators of demographic, anthropometric, body composition, cardiovascular function and basic exercise ability on the relationship between HR and RPE, and to develop the model predicting RPE from HR. 48 healthy participants were recruited to perform an incrementally 6-stage pedaling test. HR and RPE were collected during each stage. The influencing factors were identified with the forward selection method to train Gaussian Process regression (GPR), support vector machine (SVM) and linear regression models. Metrics of R2, adjusted R2 and RMSE were calculated to evaluate the performance of the models. The GPR model outperformed the SVM and linear regression models, and achieved an R2 of 0.95, adjusted R2 of 0.89 and RMSE of 0.52. Indicators of age, resting heart rate (RHR), Central arterial pressure (CAP), body fat rate (BFR) and body mass index (BMI) were identified as factors that best predicted the relationship between RPE and HR. It is possible to use GPR model to estimate RPE from HR accurately, after adjusting for age, RHR, CAP, BFR and BMI.

Reducing Power Line Interference from sEMG Signals Based on Synchrosqueezed Wavelet Transform.

Power line interference (PLI) is a major source of noise in sEMG signals. As the bandwidth of PLI overlaps with the sEMG signals, it can easily affect the interpretation of the signal. The processing methods used in the literature are mostly notch filtering and spectral interpolation. However, it is difficult for the former to reconcile the contradiction between completely filtering and avoiding signal distortion, while the latter performs poorly in the case of a time-varying PLI. To solve these, a novel synchrosqueezed-wavelet-transform (SWT)-based PLI filter is proposed. The local SWT was developed to reduce the computation cost while maintaining the frequency resolution. A ridge location method based on an adaptive threshold is presented. In addition, two ridge extraction methods (REMs) are proposed to fit different application requirements. Parameters were optimized before further study. Notch filtering, spectral interpolation, and the proposed filter were evaluated on the simulated signals and real signals. The output signal-to-noise ratio (SNR) ranges of the proposed filter with two different REMs are 18.53-24.57 and 18.57-26.92. Both the quantitative index and the time-frequency spectrum diagram show that the performance of the proposed filter is significantly better than that of the other filters.

Flavor Quality Analysis of Ten Actinidia arguta Fruits Based on High-Performance Liquid Chromatography and Headspace Gas Chromatography-Ion Mobility Spectrometry.

Actinidia arguta is a fruit crop with high nutritional and economic value. However, its flavor quality depends on various factors, such as variety, environment, and post-harvest handling. We analyzed the composition of total soluble sugars, titratable acids, organic acids, and flavor substances in the fruits of ten A. arguta varieties. The total soluble sugar content ranged from 4.22 g/L to 12.99 g/L, the titratable acid content ranged from 52.55 g/L to 89.9 g/L, and the sugar-acid ratio ranged from 5.39 to 14.17 at the soft ripe stage. High-performance liquid chromatography (HPLC) showed that citric, quinic, and malic acids were the main organic acids in the A. arguta fruits. Headspace gas chromatography-ion mobility spectrometry (HS-GC-IMS) detected 81 volatile compounds in 10 A. arguta varieties, including 24 esters, 17 alcohols, 23 aldehydes, 7 ketones, 5 terpenes, 2 acids, 1 Pyrazine, 1 furan, and 1 benzene. Esters and aldehydes had the highest relative content of total volatile compounds. An orthogonal partial least squares discriminant analysis (OPLS-DA) based on the odor activity value (OAV) revealed that myrcene, benzaldehyde, methyl isobutyrate, α-phellandrene, 3-methyl butanal, valeraldehyde, ethyl butyrate, acetoin, (E)-2-octenal, hexyl propanoate, terpinolene, 1-penten-3-one, and methyl butyrate were the main contributors to the differences in the aroma profiles of the fruits of different A. arguta varieties. Ten A. arguta varieties have different flavors. This study can clarify the differences between varieties and provide a reference for the evaluation of A. arguta fruit flavor, variety improvement and new variety selection.

Retroflexed view for reexamination of the right colon after forward view examination: Systematic review and meta-analysis.

BACKGROUND: The effect of retroflexed view (RV) for the reexamination of the right colon after forward view (FV) examination has not been fully understood. METHODS: We searched multiple databases including PubMed, Embase, and the Cochrane Library for prospective studies exploring the role of RV for reexamination of the right colon. A meta-analysis was performed on outcomes including lesion detection rates, lesion miss rates, and withdrawal time. RESULTS: Four randomized controlled trials aimed to compare the impact of the second withdrawal from the right colon in RV vs. FV following a standard colonoscopy. Both the additional adenoma detection rate (AADR) and additional polyp detection rate (APDR) of the right colon were lower in the RV group compared with the FV group (risk ratio [RR] 0.73 for AADR; RR 0.76 for APDR); similar results were noted in comparisons of the adenoma miss rate and polyp miss rate. Six prospective cohort studies aimed to describe the effect of the RV examination of the right colon after one or two rounds of FV examination. Both the adenoma detection rate (ADR) and polyp detection rate (PDR) of the right colon were slightly higher in combined examinations with RV examination than single FV examination (RR 1.11 for ADR; RR 1.16 for PDR) or two FV examinations (RR 1.21 for ADR; RR 1.22 for PDR). CONCLUSIONS: FV may detect more adenomas and polyps than RV during the second withdrawal from the right colon. RV may detect additional adenomas and polyps in the right colon after two FV examinations.

[Research on muscle fatigue recognition model based on improved wavelet denoising and long short-term memory].

The automatic recognition technology of muscle fatigue has widespread application in the field of kinesiology and rehabilitation medicine. In this paper, we used surface electromyography (sEMG) to study the recognition of leg muscle fatigue during circuit resistance training. The purpose of this study was to solve the problem that the sEMG signals have a lot of noise interference and the recognition accuracy of the existing muscle fatigue recognition model is not high enough. First, we proposed an improved wavelet threshold function denoising algorithm to denoise the sEMG signal. Then, we build a muscle fatigue state recognition model based on long short-term memory (LSTM), and used the Holdout method to evaluate the performance of the model. Finally, the denoising effect of the improved wavelet threshold function denoising method proposed in this paper was compared with the denoising effect of the traditional wavelet threshold denoising method. We compared the performance of the proposed muscle fatigue recognition model with that of particle swarm optimization support vector machine (PSO-SVM) and convolutional neural network (CNN). The results showed that the new wavelet threshold function had better denoising performance than hard and soft threshold functions. The accuracy of LSTM network model in identifying muscle fatigue was 4.89% and 2.47% higher than that of PSO-SVM and CNN, respectively. The sEMG signal denoising method and muscle fatigue recognition model proposed in this paper have important implications for monitoring muscle fatigue during rehabilitation training and exercise.

[Simulation Study on the Influence of Sampling Delay on the Accuracy of Energy Metabolism Measurement].

Indirect energy metabolism measurement is the gold standard for providing nutritional support for critical illness. The accuracy of the measurement data directly affects the outcome of the disease. In order to study the influence of sampling delay on the accuracy of energy metabolism measurement under mechanical ventilation, the Matlab/Simulink platform and respiratory electrical model were used for simulation and quantitative analysis. The results show that the error of indirect energy metabolism measurement increases with the increase of sampling delay, the error of sampling delay in mechanical ventilation mode is larger than that of spontaneous breathing, and the error of sampling delay in PCV mode of mechanical ventilation is larger than that in VCV mode. Therefore, there should be different sampling delay compensation strategies under severe mechanical ventilation and its different control modes.

[Design and Verification of Lung Diffusion Function Detection System].

A lung diffusion function detection system is designed. Firstly, the controllable collection of air, test gas source and calibration gas source was based on single-breath method measurement principle. Secondly, pulmonary diffusing capacity for carbon monoxide (DlCO) was calculated by gas concentration measured by the non-dispersive infrared sensor to measure, the gas flow measured by the differential pressure sensor, and the temperature, humidity and atmospheric pressure sensors to test and evaluate the quantitative detection and evaluation of lung diffusion function. Moreover, a preliminary verification of the lung diffusion function detection system was implemented, and the results showed that the error of the lung carbon monoxide diffusion and the alveolar volume did not exceed 5%. Therefore, the system has high accuracy and is of great value for early screening and accurate assessment of COPD.

[Design and Verification of Human Metabolic Measurement System Based on STM32].

A high-precision human metabolic measurement system is designed. The system uses STM32F103 as the main control chip to acquire oxygen, carbon dioxide and flow signals to calculate four quantitative indicators: oxygen consumption(VO2), carbon dioxide production(VCO2), respiratory entropy(RQ) and resting energy metabolism(REE), and finally uses an upper computer to display the calculation results.In this paper, the signal acquisition circuit design was carried out for the oxygen sensor, carbon dioxide sensor and flow sensor, and the validity of the device was verified with the American machine MGCDiagnositcs using Bland-Altman analysis method, and the results showed that the four parameters of VO2,VCO2, RQ and REE of both devices fell in the agreement interval of more than 95%. The device thus provides accurate metabolic measurements and offers an effective tool for the field of general health and clinical nutrition support in China.

Cytomegalovirus reactivation in immunocompetent mechanical ventilation patients: a prospective observational study.

BACKGROUND: Cytomegalovirus (CMV) reactivation is associated with adverse prognoses of critically ill patients. However, the epidemiology and predictors of CMV reactivation in immunocompetent patients receiving mechanical ventilation (MV) are not clear. The aim of this study was to investigate the epidemiology and predictors of CMV reactivation in immunocompetent patients requiring MV. METHODS: A single-center, prospective observational study (conducted from June 30, 2017 to July 01, 2018) with a follow-up of 90 days (September 29, 2018) that included 71 CMV-seropositive immunocompetent patients with MV at a 37-bed university hospital general intensive care unit (ICU) in China. Routine detection of CMV DNAemia was performed once a week for 28 days (Days 1, 7, 14, 21, and 28). CMV serology, laboratory findings, and clinical data were obtained during hospitalization. RESULTS: Among 71 patients, 13 (18.3%) showed CMV reactivation within 28 days in the ICU. The median time to reactivation was 7 days. CMV reactivation was related to various factors, including body mass index (BMI), sepsis, N-terminal pro-b-type natriuretic peptide (NT-proBNP), blood urea nitrogen (BUN), and hemoglobin (Hb) levels (P < 0.05). In the multivariate regression model, BMI, Hb level, and sepsis were independently associated with CMV reactivation patients (P < 0.05). Moreover, the area under the receiver operating characteristic (AUROC) of BMI, Hb, and BMI combined with Hb was 0.69, 0.70, and 0.76, respectively. The duration of MV, hospitalization expense, length of ICU stay, and 90 day all-cause mortality rate in patients with CMV reactivation was significantly higher than in those without CMV reactivation (P < 0.05). CONCLUSIONS: Among immunocompetent patients with MV, the incidence of CMV reactivation was 18.3%. CMV reactivation was associated with several adverse prognoses. BMI, Hb, and sepsis were independent risk factors for CMV reactivation. BMI and Hb may predict CMV reactivation.

Real-time artificial intelligence for endoscopic diagnosis of early esophageal squamous cell cancer (with video).

BACKGROUND AND AIMS: Endoscopic diagnosis of early esophageal squamous cell cancer (ESCC) is complicated and dependent on operators' experience. This study aimed to develop an artificial intelligence (AI) model for automatic diagnosis of early ESCC. METHODS: Non-magnifying and magnifying endoscopic images of normal/noncancerous lesions, early ESCC, and advanced esophageal cancer (AEC) were retrospectively obtained from Qilu Hospital of Shandong University. A total of 10,988 images from 5075 cases were chosen for training and validation. Another 2309 images from 1055 cases were collected for testing. One hundred and four real-time videos were also collected to evaluate the diagnostic performance of the AI model. The diagnostic performance of the AI model was compared with endoscopists by magnifying images and the assistant efficiency of the AI model for novices was evaluated. RESULTS: The AI diagnosis for non-magnifying images showed a per-patient accuracy, sensitivity, and specificity of 99.5%, 100%, 99.5% for white light imaging, and 97.0%, 97.2%, 96.4% for optical enhancement/iodine straining images. Regarding diagnosis for magnifying images, the per-patient accuracy, sensitivity, and specificity were 88.1%, 90.9%, and 85.0%. The diagnostic accuracy of the AI model was similar to experts (84.5%, P = 0.205) and superior to novices (68.5%, P = 0.005). The diagnostic performance of novices was significantly improved by AI assistance. When it comes to the diagnosis for real-time videos, the AI model showed acceptable performance as well. CONCLUSIONS: The AI model could accurately recognize early ESCC among noncancerous mucosa and AEC. It could be a potential assistant for endoscopists, especially for novices.