Pathological images of gastric cancer serve as the gold standard for diagnosing this malignancy. However, the recurrence prediction task often encounters challenges such as insignificant morphological features of the lesions, insufficient fusion of multi-resolution features, and inability to leverage contextual information effectively. To address these issues, a three-stage recurrence prediction method based on pathological images of gastric cancer is proposed. In the first stage, the self-supervised learning framework SimCLR was adopted to train low-resolution patch images, aiming to diminish the interdependence among diverse tissue images and yield decoupled enhanced features. In the second stage, the obtained low-resolution enhanced features were fused with the corresponding high-resolution unenhanced features to achieve feature complementation across multiple resolutions. In the third stage, to address the position encoding difficulty caused by the large difference in the number of patch images, we performed position encoding based on multi-scale local neighborhoods and employed self-attention mechanism to obtain features with contextual information. The resulting contextual features were further combined with the local features extracted by the convolutional neural network. The evaluation results on clinically collected data showed that, compared with the best performance of traditional methods, the proposed network provided the best accuracy and area under curve (AUC), which were improved by 7.63% and 4.51%, respectively. These results have effectively validated the usefulness of this method in predicting gastric cancer recurrence.
As a kind of disease with high incidence rate, high mortality, high recurrence rate and high disability rate, stroke has become one of the most serious disease burdens in China. Rapid diagnosis and treatment of stroke can effectively improve the outcome of patients and reduce the psychological and economic burden of patients’ families and society. In recent years, with the rapid development of artificial intelligence technology,this technology can effectively improve daily diagnosis and treatment efficiency. This paper focuses on the application of artificial intelligence technology to the diagnosis, treatment and outcome prediction of stroke, aiming to provide ideas for further guiding precision medicine.
ObjectiveTo explore the risk factors affecting occurrence of arteriosclerosis obliterans (ASO) for patients with type 2 diabetes mellitus (T2DM) and to develop a nomogram predictive model using these risk factors. MethodsA case-control study was conducted. The patients with T2DM accompanied with ASO and those with T2DM alone, admitted to the First Affiliated Hospital of Xinjiang Medical University from January 2017 to December 2022, were retrospectively collected according to the inclusion and exclusion criteria. The basic characteristics, blood, thyroid hormones, and other relevant indicators of the paitents in two groups were compared. The multivariate logistic regression analysis was used to identify the risk factors for the occurrence of ASO in the patients with T2DM, and then a nomogram predictive model was developed. ResultsThere were 119 patients with T2DM alone and 114 patients with T2DM accompanied with lower extremity ASO in this study. The significant differences were observed between the two groups in terms of smoking history, white blood cell count, neutrophil count, lymphocyte count, platelet count, systemic immune-inflammation index, systemic inflammatory response index (SIRI), high-density lipoprotein cholesterol, apolipoprotein A1 (ApoA1), apolipoprotein α (Apoα), serum cystatin C, free-triiodothyronine (FT3), total triiodothyronine, FT3/total triiodothyronine ratio, fibrinogen (Fib), fibrinogen degradation products, and plasma D-dimer (P<0.05). Further the results of the multivariate logistic regression analysis revealed that the history of smoking, increased Fib level and SIRI value increased the probabilities of ASO occurrence in the patients with T2DM [OR (95%CI)=2.921 (1.023, 4.227), P=0.003; OR (95%CI)=2.641 (1.810, 4.327), P<0.001; OR (95%CI)=1.020 (1.004, 1.044), P=0.018], whereas higher levels of ApoA1 and FT3 were associated with reduced probabilities of ASO occurrence in the patients with T2DM [OR (95%CI)=0.231 (0.054, 0.782), P=0.021; OR (95%CI)=0.503 (0.352, 0.809), P=0.002]. The nomogram predictive model based on these factors demonstrated a good discrimination for predicting the ASO occurrence in the T2DM patients [area under the receiver operating characteristic curve (95%CI)=0.788 (0.730, 0.846)]. The predicted curve closely matched the ideal curve (Hosmer-Lemeshow goodness-of-fit test, χ2=5.952, P=0.653). The clinical decision analysis curve showed that the clinical net benefit of intervention based on the nomogram model was higher within a threshold probability range of 0.18 to 0.80 compared to no intervention or universal intervention. ConclusionsThe analysis results indicate that T2DM patients with a smoking history, elevated Fib level and SIRI value, as well as decreased ApoA1 and FT3 levels should be closely monitored for ASO risk. The nomogram predictive model based on these features has a good discriminatory power for ASO occurrence in T2DM patients, though its value warrants further investigation.
The drug-target protein interaction prediction can be used for the discovery of new drug effects. Recent studies often focus on the prediction of an independent matrix filling algorithm, which apply a single algorithm to predict the drug-target protein interaction. The single-model matrix-filling algorithms have low accuracy, so it is difficult to obtain satisfactory results in the prediction of drug-target protein interaction. AdaBoost algorithm is a strong multiple classifier combination framework, which is proved by the past researches in classification applications. The drug-target interaction prediction is a matrix filling problem. Therefore, we need to adjust the matrix filling problem to a classification problem before predicting the interaction among drug-target protein. We make full use of the AdaBoost algorithm framework to integrate several weak classifiers to improve performance and make accurate prediction of drug-target protein interaction. Experimental results based on the metric datasets show that our algorithm outperforms the other state-of-the-art approaches and classical methods in accuracy. Our algorithm can overcome the limitations of the single algorithm based on machine learning method, exploit the hidden factors better and improve the accuracy of prediction effectively.
Acute lung injury is one of the common and serious complications of acute aortic dissection, and it greatly affects the recovery of patients. Old age, overweight, hypoxemia, smoking history, hypotension, extensive involvement of dissection and pleural effusion are possible risk factors for the acute lung injury before operation. In addition, deep hypothermia circulatory arrest and blood product infusion can further aggravate the acute lung injury during operation. In this paper, researches on risk factors, prediction model, prevention and treatment of acute aortic dissection with acute lung injury were reviewed, in order to provide assistance for clinical diagnosis and treatment.
ObjectiveTo evaluate existing predictive models for surgical site infection (SSI) following colorectal cancer (CRC) surgery, aiming to provide a scientific basis for refining risk prediction models and developing clinically practical and widely applicable screening tools. MethodA comprehensive review of existing literature on predictive models for SSI following CRC surgery, both domestically and internationally, were conducted. ResultsThe determination of SSI following CRC surgery primarily relied on the Centers for Disease Control and Prevention standard of USA, which presented issues of consistency and accuracy. Various predictive models had been developed, including traditional statistical models and machine learning models, with 0.991 of an area under the operating characteristic curve of predictive model. However, most studies were based on retrospective and single-center data, which limited their applicability and accuracy. ConclusionsAlthough existing models provide strong support for predicting SSI following CRC surgery, there is a need for multi-center, prospective studies to enhance the generalizability and accuracy of these models. Additionally, future research should focus on improving model interpretability to better apply them in clinical practice, providing personalized risk assessments and intervention strategies for patients.
Aiming at the problem of scaffold degradation in bone tissue engineering, we studied the feasibility that controlls bone defect repair effect with the inhomogeneous structure of scaffold. The prediction model of bone defect repair which contains governing equations for bone formation and scaffold degradation was constructed on the basis of analyzing the process and main influence factors of bone repair in bone tissue engineering. The process of bone defect repair and bone structure after repairing can be predicted by combining the model with finite element method (FEM). Bone defect repair effects with homogenous and inhomogeneous scaffold were simulated respectively by using the above method. The simulation results illustrated that repair effect could be impacted by scaffold structure obviously and it can also be controlled via the inhomogeneous structure of scaffold with some feasibility.
ObjectiveTo establish a predictive model for neck lateral lymph node metastasis (LLNM) in unilateral papillary thyroid cancer (uni-PTC) with central lymph node metastasis (CLNM). MethodsThe uni-PTC patients with CLNM were included in this study. The patients underwent thyroid surgery in the 960th Hospital of the PLA Joint Logistics Support Force from May 2018 to December 2021, who were randomly divided into the modeling group and the validation group according to the ratio of 7∶3. The risk factors of neck LLNM were analyzed by univariate and multivariate logistic regression and the nomogram of prediction model was constructed. The receiver operating characteristic (ROC) curve and calibration curve were used to validate the prediction model. ResultsA total of 237 patients were included in this study, including 158 patients in the modeling group and 79 patients in the validation group. The LLNM occurred in the 84 patients of the modeling group and 43 patients of the validation group. The multivariate logistic regression analysis was performed according to the statistical indicators in the univariate analysis results of the modeling group and the risk factors considered in the previous studies. The results showed that the patients with maximum diameter of the lesions >1 cm, multiple lesions, extraglandular invasion, the rate of CLNM ≥0.414, and lesions located at the upper portion had higher probability of LLNM (OR>1, P<0.05). The area under ROC curve of the nomogram in predicting LLNM in the modeling group was 0.834 [95%CI (0.771, 0.896)], which in the validation group was 0.761 [95%CI (0.651, 0.871)]. The calibration curve showed a good calibration degree in the prediction model. ConclusionThe clinical risk prediction model established based on the risk factors can better predict the probability of LLNM.
This study introduced the construction of individualized risk assessment model based on Bayesian networks, comparing with traditional regression-based logistic models using practical examples. It evaluates the model's performance and demonstrates its implementation in the R software, serving as a valuable reference for researchers seeking to understand and utilize Bayesian network models.
ObjectiveTo explore the application value of machine learning models in predicting postoperative survival of patients with thoracic squamous esophageal cancer. MethodsThe clinical data of 369 patients with thoracic esophageal squamous carcinoma who underwent radical esophageal cancer surgery at the Department of Thoracic Surgery of Northern Jiangsu People's Hospital from January 2014 to September 2015 were retrospectively analyzed. There were 279 (75.6%) males and 90 (24.4%) females aged 41-78 years. The patients were randomly divided into a training set (259 patients) and a test set (110 patients) with a ratio of 7 : 3. Variable screening was performed by selecting the best subset of features. Six machine learning models were constructed on this basis and validated in an independent test set. The performance of the models' predictions was evaluated by area under the curve (AUC), accuracy and logarithmic loss, and the fit of the models was reflected by calibration curves. The best model was selected as the final model. Risk stratification was performed using X-tile, and survival analysis was performed using the Kaplan-Meier method with log-rank test. ResultsThe 5-year postoperative survival rate of the patients was 67.5%. All clinicopathological characteristics of patients between the two groups in the training and test sets were not statistically different (P>0.05). A total of seven variables, including hypertension, history of smoking, history of alcohol consumption, degree of tissue differentiation, pN stage, vascular invasion and nerve invasion, were included for modelling. The AUC values for each model in the independent test set were: decision tree (AUC=0.796), support vector machine (AUC=0.829), random forest (AUC=0.831), logistic regression (AUC=0.838), gradient boosting machine (AUC=0.846), and XGBoost (AUC=0.853). The XGBoost model was finally selected as the best model, and risk stratification was performed on the training and test sets. Patients in the training and test sets were divided into a low risk group, an intermediate risk group and a high risk group, respectively. In both data sets, the differences in surgical prognosis among three groups were statistically significant (P<0.001). ConclusionMachine learning models have high value in predicting postoperative prognosis of thoracic squamous esophageal cancer. The XGBoost model outperforms common machine learning methods in predicting 5-year survival of patients with thoracic squamous esophageal cancer, and it has high utility and reliability.