AFTGAN: prediction of multi-type PPI based on attention free transformer and graph attention network.

MOTIVATION: Protein-protein interaction (PPI) networks and transcriptional regulatory networks are critical in regulating cells and their signaling. A thorough understanding of PPIs can provide more insights into cellular physiology at normal and disease states. Although numerous methods have been proposed to predict PPIs, it is still challenging for interaction prediction between unknown proteins. In this study, a novel neural network named AFTGAN was constructed to predict multi-type PPIs. Regarding feature input, ESM-1b embedding containing much biological information for proteins was added as a protein sequence feature besides amino acid co-occurrence similarity and one-hot coding. An ensemble network was also constructed based on a transformer encoder containing an AFT module (performing the weight operation on vital protein sequence feature information) and graph attention network (extracting the relational features of protein pairs) for the part of the network framework. RESULTS: The experimental results showed that the Micro-F1 of the AFTGAN based on three partitioning schemes (BFS, DFS and the random mode) on the SHS27K and SHS148K datasets was 0.685, 0.711 and 0.867, as well as 0.745, 0.819 and 0.920, respectively, all higher than that of other popular methods. In addition, the experimental comparisons confirmed the performance superiority of the proposed model for predicting PPIs of unknown proteins on the STRING dataset. AVAILABILITY AND IMPLEMENTATION: The source code is publicly available at https://github.com/1075793472/AFTGAN. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

TRPM7 mediates endoplasmic reticulum stress and ferroptosis in sepsis-induced myocardial injury.

Transient receptor potential melastatin 7 (TRPM7), a non-selective cation channel, was significantly upregulated in the blood of patients with sepsis. This study focuses on the preliminary exploration of the probable regulatory mechanism of TRPM7 in sepsis-induced myocardial injury (SIMI). HL-1 cardiac muscle cell line was treated with lipopolysaccharide (LPS) to mimic SIMI in vitro, and TRPM7 level was assessed. The impacts of TRPM7 knockdown on cellular inflammation response, oxidative stress, apoptosis, endoplasmic reticulum (ER) stress, and ferroptosis were identified. In order to explore the mechanism, ER stress agonist tunicamycin (TM) or ferroptosis inducer erastin was applied to treat HL-1 cells. The influences of TM and erastin on the aforementioned aspects were evaluated. TRPM7 was elevated in response to LPS stimulation, and its knockdown reduced the secretion of inflammatory factors and oxidative stress degree. Moreover, TRPM7 knockdown significantly suppressed cell apoptosis, ER stress, and ferroptosis. TM and erastin reversed the functions of TRPM7 knockdown, indicating ER stress and ferroptosis mediated in the regulation of TRPM7. This research proposes the possibility of TRPM7 as a marker or target for SIMI, and provides theoretical support for follow-up research.

Few-cycle THz wave manipulation with a high degree of freedom via f-t modulation.

THz waves have been intensively applied in many fields, e.g., spectroscopy, imaging, and communications. However, owing to the rarity of available techniques for manipulating circularly polarized few-cycle THz waves on picosecond time scales, most of the current studies are conducted with linearly polarized THz waves. Here we demonstrate circularly polarized (CP) THz (dual) pulses generated by a polarization-twisting pulse/dual pulse (PTP/PTDP). The polarization-twisting optical dual pulses can be generated via a modified Michelson interferometer (MI) system, which provides the ability to control the frequency, helicity, and time interval of the dual pulses arbitrarily and individually. Such a novel, to the best of our knowledge, modulation technique shows huge potential for applications, not only in imaging and spectroscopy but also in next-generation communications.

Extraocular muscle volume index at the orbital apex with optic neuritis: a combined parameter for diagnosis of dysthyroid optic neuropathy.

OBJECTIVES: To evaluate the diagnostic performance of the extraocular muscle volume index at the orbital apex (AMI) and the signal intensity ratio (SIR) of the optic nerve in dysthyroid optic neuropathy (DON). METHODS: Clinical data and magnetic resonance imaging were collected retrospectively from 63 Graves' ophthalmopathy patients, including 24 patients with DON and 39 without DON. The volume of these structures was obtained by reconstructing their orbital fat and extraocular muscles. The SIR of the optic nerve and axial length of eyeball were also measured. The posterior 3/5 of the retrobulbar space volume was used as the orbital apex to compare parameters in patients with or without DON. Area under the receiver operating characteristic curve (AUC) analysis was used to select the morphological and inflammatory parameters with the highest diagnostic value. A logistic regression was performed to identify the risk factors of DON. RESULTS: One hundred twenty-six orbits (35 with DON and 91 without DON) were analyzed. Most of the parameters in DON patients were significantly higher than in non-DON patients. However, the SIR 3 mm behind the eyeball of the optic nerve and AMI had the highest diagnostic value in these parameters and are independent risk factors of DON by stepwise multivariate logistic regression analysis. Combining AMI and SIR had a higher diagnostic value than a single index. CONCLUSIONS: Combining AMI with SIR 3 mm behind the eyeball's orbital nerve can be a potential parameter for diagnosing DON. CLINICAL RELEVANCE STATEMENT: The present study provided a quantitative index based on morphological and signal changes to assess the DON, allowing clinicians and radiologists to monitor DON patients timely. KEY POINTS: The extraocular muscle volume index at the orbital apex (AMI) has excellent diagnostic performance for dysthyroid optic neuropathy. A signal intensity ratio (SIR) of 3 mm behind the eyeball has a higher AUC compared to other slices. Combining AMI and SIR has a higher diagnostic value than a single index.

Artificial Intelligence-Based Psoriasis Severity Assessment: Real-world Study and Application.

BACKGROUND: Psoriasis is one of the most frequent inflammatory skin conditions and could be treated via tele-dermatology, provided that the current lack of reliable tools for objective severity assessments is overcome. Psoriasis Area and Severity Index (PASI) has a prominent level of subjectivity and is rarely used in real practice, although it is the most widely accepted metric for measuring psoriasis severity currently. OBJECTIVE: This study aimed to develop an image-artificial intelligence (AI)-based validated system for severity assessment with the explicit intention of facilitating long-term management of patients with psoriasis. METHODS: A deep learning system was trained to estimate the PASI score by using 14,096 images from 2367 patients with psoriasis. We used 1962 patients from January 2015 to April 2021 to train the model and the other 405 patients from May 2021 to July 2021 to validate it. A multiview feature enhancement block was designed to combine vision features from different perspectives to better simulate the visual diagnostic method in clinical practice. A classification header along with a regression header was simultaneously applied to generate PASI scores, and an extra cross-teacher header after these 2 headers was designed to revise their output. The mean average error (MAE) was used as the metric to evaluate the accuracy of the predicted PASI score. By making the model minimize the MAE value, the model becomes closer to the target value. Then, the proposed model was compared with 43 experienced dermatologists. Finally, the proposed model was deployed into an app named SkinTeller on the WeChat platform. RESULTS: The proposed image-AI-based PASI-estimating model outperformed the average performance of 43 experienced dermatologists with a 33.2% performance gain in the overall PASI score. The model achieved the smallest MAE of 2.05 at 3 input images by the ablation experiment. In other words, for the task of psoriasis severity assessment, the severity score predicted by our model was close to the PASI score diagnosed by experienced dermatologists. The SkinTeller app has been used 3369 times for PASI scoring in 1497 patients from 18 hospitals, and its excellent performance was confirmed by a feedback survey of 43 dermatologist users. CONCLUSIONS: An image-AI-based psoriasis severity assessment model has been proposed to automatically calculate PASI scores in an efficient, objective, and accurate manner. The SkinTeller app may be a promising alternative for dermatologists' accurate assessment in the real world and chronic disease self-management in patients with psoriasis.

Construction and verification of a quantitative index system to evaluate the medical and nursing care for the elderly service needs. / è€å¹´äººåŒ»å »æœåŠ¡éœ€æ±‚é‡åŒ–è¯„ä¼°æŒ‡æ ‡ä½“ç³»çš„æž„å»ºä¸ŽéªŒè¯.

OBJECTIVES: To construct a quantitative index system with the integrated medical and nursing care assessment for the elderly service needs, this system can assess the cost of medical and care services accurately and objectively, so as to provide scientific basis for the allocation of old-age service resources in China. METHODS: Based on the survival needs of the Existence, Relation and Growth theory, an index system is constructed through literature analysis, group discussion, and expert correspondence. Analytic hierarchy process was used to determine the weights of indicators at all levels. The 3-grades service items corresponding to each index were quantified through the measurement of working hours, and the medical and nursing care needs of 624 disabled/demented elderly people over 60 years old in Changsha were investigated to evaluate their reliability and validity. RESULTS: The authoritative coefficients of the 2 rounds of expert correspondence were 88.5% and 88.6%, respectively, and the opinion coordination coefficients were 0.159 and 0.167, respectively. The final quantitative evaluation index system included 4 first-level indicators, 17 second-level indicators, and 105 third-level indicators. The service time of doctor ranged from 6.01 to 22.64 min, the service time of nurses ranged from 0.77 to 24.79 min, and the service time of caregiver ranged from 0.12 to 51.88 min. The Cronbach's αcoefficient was 0.73, the split-half reliability was 0.74, the content validity was 0.93, and the calibration validity was 0.781. CONCLUSIONS: The quantitative evaluation index system of medical and nursing service need for the elderly can be used to accurately evaluate the medical and nursing service need.

Highly sensitive electrochemical BPA sensor based on titanium nitride-reduced graphene oxide composite and core-shell molecular imprinting particles.

A sensitive electrochemical sensor was proposed via combining molecular imprinting technique with the graphene material-doped titanium nitride. The novel graphene with 3-dimensional structure displayed more binding sites and better electrochemical properties. Moreover, this study focused on coating pyrrole with electrical conductivity on the surface of silica as a monomer, and BPA as the template. The interaction made specific detection possible, between monomer and template. With a series of characterizations and electrochemical measurements, CPE (carbon paste electrode)-contained TiN-rGO composite was proved to have conductivity improved. Also, the modified polymer performed well selectivity which reflected in that it was almost impervious to distractions. Under optimized conditions, a linear dependence was observed from 0.5 to 100 nmol L-1 with a detection limit of 0.19 nmol L-1. The sensor explicated outstanding repeatability via repetitive experiment with the RSD of 0.02%, while the results of stability experiment reached the RSD of 1.90%. Eventually, it was used to analyze BPA residues in 3 kinds of daily supplies. The results indicated the potential of the sensor in environmental detection prospectively.

Photoresponsive molecularly imprinted polymers based on 4-[(4-methacryloyloxy)phenylazo] benzenesulfonic acid for the determination of sulfamethazine.

Core-shell structured photoresponsive molecularly imprinted polymers were developed for the determination of sulfamethazine in milk samples. The photoresponsive imprinted polymers were prepared with polymethyl methacrylate containing a mass of ester groups as core, sulfamethazine as template molecules, self-synthesized water-soluble 4-[(4-methacryloyloxy)phenylazo] benzenesulfonic acid as a photoresponsive monomer, and ethylene dimethacrylate as cross-linker. Interestingly, the imprinted polymer can specifically adsorb sulfamethazine under dark and 440 nm irradiation, and release it at 365 nm. A series of adsorption experiments showed that the maximum adsorption capacity reached 12.5 mg⋅g-1 , and the adsorption equilibrium was achieved within 80 min. Moreover, the imprinted polymers display excellent reusability, with almost no performance loss after four times photo-controlled adsorption-release cycles, and the imprinted polymers have excellent selectively for sulfamethazine (imprinting factor  = 3.01). In the end, the imprinted polymers realized effective separation and enrichment of sulfamethazine in milk, with a recovery rate of over 97.5%. The material can be used as a solid-phase extractant in the process of enrichment and separation for the quantitative detection of sulfamethazine in milk samples.

Selective enrichment-release of trace dibutyl phthalate via molecular-imprinting based photo-controlled switching followed by high-performance liquid chromatography analysis.

A novel intelligent photo-controlled molecularly imprinted polymers were synthesized, based on the magnetic core-shell structure, with 4-[(4-methacryloyloxy) phenylazo] benzenesulfonic acid as the functional monomer and ethylene glycol dimethyl acrylate as the cross-linking agent. Subsequently, a series of light-controlled enrichment-release performance showed that it only took about 30 and 10 min to reach the equilibrium photosensitive characteristic peak, respectively. The photo-controlled polymers could intelligently select target molecules, the maximum adsorption capacity for dibutyl phthalate was 3.88 mg/g. However, the adsorption capacity for its structural analogue dicyclohexyl phthalate was only 0.88 mg/g. The Freundlich and Langmuir isothermal equations were discussed for the specific enrichment process. Finally, the photo-controlled molecularly imprinted polymers were successfully applied to the selective detection of dibutyl phthalate, with the recovery rate of 95.4-98.4%. It could be used for the analysis of trace dibutyl phthalate in actual samples.

Development and Clinical Uses of a Mobile Application for Smart Wound Nursing Management.

OBJECTIVE: Establishing a smart platform for wound nursing management is imperative to accelerating the development of this specialty. A mobile application (app) was created based on the clinical needs of wound care to improve working efficiency and wound management, as well as enable information sharing and categorical retrieval of patient records. METHODS: Supported by the National Mobile Laboratory of China, the authors' wound care center collaborated with information technology engineers and user interface designers to establish a multidisciplinary team of experts in wound nursing information management and develop a digital app. Electronic tablets were chosen as a convenient clinical method to collect complete information in a standardized format. In the app, case records are synchronously cataloged for rapid retrieval. The app was designed such that practicing nurses could quickly learn how to use the various functions of the app. Through constant iterative optimization, the relevant functional modules have been further improved. RESULTS AND CONCLUSIONS: After a 1-year clinical trial, app functions have stabilized, fulfilling the expectations of technicians and boosting the operating efficiency of clinical practices. In comparison with traditional methods, the app allows for easy access to patient information; rapidly reviewed, synchronous cataloguing; and convenient retrieval. Further exploration is warranted.

Construction and verification of an intelligent measurement model for diabetic foot ulcer. / ä¸€ç§ç³–å°¿ç— è¶³æºƒç–¡æ™ºèƒ½æµ‹é‡æ¨¡åž‹çš„æž„å»ºä¸ŽéªŒè¯.

OBJECTIVES: The measurement of diabetic foot ulcers is important for the success in diabetic foot ulcer management. At present, it lacks the accurate and convenient measurement tools in clinical. In recent years, artificial intelligence technology has demonstrated the potential application value in the field of image segmentation and recognition. This study aims to construct an intelligent measurement model of diabetic foot ulcers based on the deep learning method, and to conduct preliminary verification. METHODS: The data of 1 042 diabetic foot ulcers clinical samples were collected. The ulcers and color areas were manually labeled, of which 782 were used as the training data set and 260 as the test data set. The Mask RCNN ulcer tissue color semantic segmentation and RetinaNet scale digital scale target detection were used to build a model. The training data set was input into the model and iterated. The test data set was used to verify the intelligent measurement model. RESULTS: This study established an intelligent measurement model of diabetic foot ulcers based on deep learning. The mean average precision@.5 intersection over union (mAP@.5IOU) of the color region segmentation in the training set and the test set were 87.9% and 63.9%, respectively; the mAP@.5IOU of the ruler scale digital detection in the training set and the test set were 96.5% and 83.4%, respectively. Compared with the manual measurement result of the test sample, the average error of the intelligent measurement result was about 3 mm. CONCLUSIONS: The intelligent measurement model has good accuracy and robustness in measuring the diabetic foot ulcers. Future research can further optimize the model with larger-scale data samples.

Generation of sub-half-cycle 10 µm pulses through filamentation at kilohertz repetition rates.

We have experimentally demonstrated the generation of sub-half-cycle phase-stable pulses with the carrier wavelength of 10.2 µm through two-color filamentation in nitrogen. The carrier-envelope phase (CEP) of the MIR pulse is passively stabilized and controlled by the attosecond time delay between the two-color input pulses. The duration of the MIR pulse is 13.7 fs, which corresponds to 0.402 cycles. The absolute value of the CEP of the generated sub-half-cycle pulse is consistent with a simple four-wave difference frequency generation model. We have also found that the 10 kHz repetition rate of the light source causes the fluctuation of the pulse energy on a few hundred millisecond time scale.

Efficiently accumulating germ-like stem cells from mouse postnatal ovary by in situ tissue culture.

Although recent studies have revealed that germline stem cells (GSCs) exist in the mouse postnatal ovary, how to efficiently obtain GSCs for regenerating neo-oogenesis is still a technical challenge. Here, we report that using in situ tissue culture we can efficiently accumulate large amounts of proliferating germ-like cells from mouse postnatal ovaries. Usually, more than 10,000 germ-like cells can be derived from one ovary by this method, and over 20% of these cells can grow into germ-like cells with self-renewal, which thus can serve as a good cell pool to isolate GSCs by other cell assorting methods such as FACS. This method is simple and time-saving, which should be useful for in future studies on mouse GSCs.

Photo-stimulated "turn-on/off" molecularly imprinted polymers based on magnetic mesoporous silicon surface for efficient detection of sulfamerazine.

In this study, novel photo-stimulated molecularly imprinted polymers based on magnetic mesoporous carrier surface were developed for selective identification and intelligent separation of sulfamerazine in complex samples. The photosensitive monomer of the molecularly imprinted polymers was azobenzene derivative 5-[(4-(methacryloyloxy)phenyl) diazenyl] isophthalic acid with stimulus reaction mechanisms, which has photoisomerization between trans and cis for N=N bonds. Further, the properties of the photo-stimulated molecularly imprinted polymers were further evaluated through several sets of adsorption experiments. It illustrated that the maximum adsorption amount is 0.45 mmol/L. By ultraviolet spectrophotometry, the material reaches typical characteristic peaks of photo sensitivity, and the cycle time is 16 min. Three adsorption and desorption processes were repeated, the adsorption rate reached 34.4%. Overall, the photo-stimulated molecularly imprinted polymers can enrich and separate determine sulfamerazine with high selectivity, which have good recovery for real samples.

A sensitive electrochemical sensor modified with multi-walled carbon nanotubes doped molecularly imprinted silica nanospheres for detecting chlorpyrifos.

A highly sensitive and convenient electrochemical sensor, based on surface molecularly imprinted polymers and multiwalled carbon nanotubes, was successfully developed to detect chlorpyrifos in real samples. In order to solve the problems like uneven shapes, poor size accessibility, and low imprinting capacity, the layer of the molecularly imprinted polymer was prepared on the surface of silica nanospheres. Moreover, the doping of multiwalled carbon nanotubes greatly improved the electrical properties of developed sensor. Under the optimal conductions, the electrochemical response of the sensor is linearly proportional to the concentration of chlorpyrifos in the range of 5.0 × 10-12 -5.0 × 10-8  mol/L with a low detection limit of 8.1 × 10-13  mol/L. The prepared sensor exhibited multiple advantages such as low cost, simple preparation, convenient use, excellent selectivity, and good reproducibility. Finally, the prepared sensor was successfully used to detect chlorpyrifos in vegetable and fruit.

Development of prognostic model for patients at CKD stage 3a and 3b in South Central China using computational intelligence.

BACKGROUND: Chronic kidney disease (CKD) stage 3 was divided into two subgroups by eGFR (45 mL/ min 1.73 m2). There is difference in prevalence of CKD, racial differences, economic development, genetic, and environmental backgrounds between China and Western countries. METHODS: We used a computational intelligence model (CKD stage 3 Modeling, CSM) to distinguish CKD stage 3 with CKD stage 3a/3b by data distribution rules, pearson correlation coefficient (PCC), spearman correlation (SCC) analysis, logistic regression (LR), random forest (RF), support vector machine (SVM), and neural network (Nnet) to develop Prognostic Model for patients with CKD stage 3a/3b in South Central China. Furthermore, we used RF to discover risk factors of progression of CKD stage 3a and 3b to CKD stage 5. 1090 cases of CKD stage 3 patients in Xiangya Hospital were collected. Among them, 455 patients progressed to CKD stage 5 in a median follow-up of 4 years (IQR 4.295, 4.489). RESULTS: We found that the common risk factors for progression of CKD stage 3a/3b to CKD stage 5 included albumin, creatinine, total protein, etc. Proteinuria, direct bilirubin, hemoglobin, etc. accounted for the progression from stage CKD stage 3a to stage 5. The risk factors for CKD stage 3b progression to stage 5 included low-density lipoprotein cholesterol, diabetes, eosinophil percentage, etc. CONCLUSIONS: CSM could be used as a point-of-care test to screen patients at high risk for disease progression, might allowing individualized therapeutic management.

Hausdorff Distance Model-Based Identity Authentication for IP Circuits in Service-Centric Internet-of-Things Environment.

Rapid advances in the Internet-of-Things (IoT) have exposed the underlying hardware devices to security threats. As the major component of hardware devices, the integrated circuit (IC) chip also suffers the threat of illegal, malicious attacks. To protect against attacks and vulnerabilities of a chip, a credible authentication is of fundamental importance. In this paper, we propose a Hausdorff distance-based method to authenticate the identity of IC chips in IoT environments, where the structure is analyzed, and the lookup table (LUT) resources are treated as a set of reconfigurable nodes in field programmable gate array (FPGA)-based IC design. Unused LUT resources are selected for insertion of the copyright information by using the depth-first search algorithm, and the random positions are reordered with the Hausdorff distance matching function next, so these positions are mapped to satisfy the specific constraints of the optimal watermark positions. If the authentication process is activated, virtual positions are mapped to the initial key file, yet the identity of the IC designed can be authenticated using the mapping relationship of the Hausdorff distance function. Experimental results show that the proposed method achieves good randomness and secrecy in watermark embedding, as well the extra resource overhead caused by watermarks are promising.

Cortical Microtubule Organization during Petal Morphogenesis in Arabidopsis.

Cortical microtubules guide the direction and deposition of cellulose microfibrils to build the cell wall, which in turn influences cell expansion and plant morphogenesis. In the model plant Arabidopsis thaliana (Arabidopsis), petal is a relatively simple organ that contains distinct epidermal cells, such as specialized conical cells in the adaxial epidermis and relatively flat cells with several lobes in the abaxial epidermis. In the past two decades, the Arabidopsis petal has become a model experimental system for studying cell expansion and organ morphogenesis, because petals are dispensable for plant growth and reproduction. Recent advances have expanded the role of microtubule organization in modulating petal anisotropic shape formation and conical cell shaping during petal morphogenesis. Here, we summarize recent studies showing that in Arabidopsis, several genes, such as SPIKE1, Rho of plant (ROP) GTPases, and IPGA1, play critical roles in microtubule organization and cell expansion in the abaxial epidermis during petal morphogenesis. Moreover, we summarize the live-confocal imaging studies of Arabidopsis conical cells in the adaxial epidermis, which have emerged as a new cellular model. We discuss the microtubule organization pattern during conical cell shaping. Finally, we propose future directions regarding the study of petal morphogenesis and conical cell shaping.

The Application of Deep Learning in the Risk Grading of Skin Tumors for Patients Using Clinical Images.

According to diagnostic criteria, skin tumors can be divided into three categories: benign, low degree and high degree malignancy. For high degree malignant skin tumors, if not detected in time, they can do serious harm to patients' health. However, in clinical practice, identifying malignant degree requires biopsy and pathological examination which is time costly. Furthermore, in many areas, due to the severe shortage of dermatologists, it's inconvenient for patients to go to hospital for examination. Therefore, an easy to access screening method of malignant skin tumors is needed urgently. Firstly, we spend 5 years to build a dataset which includes 4,500 images of 10 kinds of skin tumors. All instances are verified pathologically thus trustworthy; Secondly, we label each instance to be either low-risk, high-risk or dangerous in which Junctional nevus, Intradermal nevus, Dermatofibroma, Lipoma and Seborrheic keratosis are low-risk, Basal cell carcinoma, Bowen's disease and Actinic keratosis are high-risk, Squamous cell carcinoma and Malignant melanoma are dangerous; Thirdly, we apply the Xception architecture to build the risk degree classifier. The area under the curve (AUC) for three risk degrees reach 0.959, 0.919 and 0.947 respectively. To further evaluate the validity of the proposed risk degree classifier, we conduct a competition with 20 professional dermatologists. The results showed the proposed classifier outperforms dermatologists. Our system is helpful to patients in preliminary screening. It can identify the patients who are at risk and alert them to go to hospital for further examination.

[Clinical image identification of basal cell carcinoma and pigmented nevi based on convolutional neural network].

OBJECTIVE: To construct an intelligent assistant diagnosis model based on the clinical images of basal cell carcinoma (BCC) and pigmented nevi in Chinese by using the advanced convolutional neural network (CNN).
 Methods: Based on the Xiangya Medical Big Data Platform, we constructed a large-scale clinical image dataset of skin diseases according to Chinese ethnicity and the Xiangya Skin Disease Dataset. We evaluated the performance of 5 mainstream CNN models (ResNet50, InceptionV3, InceptionResNetV2, DenseNet121, and Xception) on a subset of BCC and pigmented nevi of this dataset. We also analyzed the basis of the diagnosis results in the form of heatmaps. We compared the optimal CNN classification model with 30 professional dermatologists.
 Results: The Xiangya Skin Disease Dataset contains 150 223 clinical images with lesion annotations, covering 543 skin diseases, and each image in the dataset contains support for pathological gold standards and the patient's overall medical history. On the test set of 349 BCC and 497 pigmented nevi, the optimal CNN model was Xception, and its classification accuracy can reach 93.5%, of which the area under curve (AUC) values were 0.974 and 0.969, respectively. The results of the heatmap showed that the CNN model can indeed learn the characteristics associated with disease identification. The ability of the Xception model to identify clinical images of BCC and Nevi was basically comparable to that of professional dermatologists.
 Conclusion: This study is the first assistant diagnosis study for skin tumor based on Chinese ethnic clinical dataset. It proves that CNN model has the ability to distinguish between Chinese ethnicity's BCC and Nevi, and lays a solid foundation for the following application of artificial intelligence in the diagnosis and treatment for skin tumors.