Jiancheng Dong

Diabetes, as a typical chronic disease, seriously affects the quality of life and economic burden of patients and their families. If the early and accurate prediction of diabetes can be realized according to patients’ health-data, the development of diabetes would be effectively prevented or mitigated. The existing diabetes prediction models have achieved better prediction results. However, noise and high-dimensional nonlinear features in health-data may still be important factors affecting the performance of prediction models and deserve further research. Aiming at these, we propose an integrated diabetes prediction model based on user-defined risk decision making strategies, named FDNN. The model tandemly integrates modules for data-preprocessing, diabetes prediction models based on different features, and user-defined risk decision making strategies. The problems of data-standardization, data noise filtering, and high-dimensional nonlinear features learning are solved hierarchically in the model. Finally, based on the PIMA dataset, FDNN is compared with baselines such as SVM, Logistic, random forest, and gradient boosting decision tree (GBDT). The experimental results show that noise does exist in the diabetes dataset and the noise is an important factor affecting the prediction performance. FDNN with feature selection is proposed to mitigate the impact of noise in diabetes dataset on prediction performance. In the dataset with noise, the sensitivity, specificity, accuracy and area under the receiver operating characteristic curve of FDNN are 0.9278, 0.6667, 0.8312, and 0.8475 respectively, which are better than the comparison algorithms. It is 6.75%, 11.92%, 5.79%, and 0.83% better than GBDT respectively, with more accurate and stable prediction performance.

Background: Growing evidence indicates that non-alcoholic fatty liver disease (NAFLD) is related to the occurrence and development of diabetic nephropathy (DN). This bioinformatics study aimed to explore optimal crosstalk genes and related pathways between NAFLD and DN.Methods: Gene expression profiles were downloaded from Gene Expression Omnibus. CIBERSORT algorithm was employed to analyze the similarity of infiltrating immunocytes between the two diseases. Protein–protein interaction (PPI) co-expression network and functional enrichment analysis were conducted based on the identification of common differentially expressed genes (DEGs). Least absolute shrinkage and selection operator (LASSO) regression and Boruta algorithm were implemented to initially screen crosstalk genes. Machine learning models, including support vector machine, random forest model, and generalized linear model, were utilized to further identify the optimal crosstalk genes between DN and NAFLD. An integrated network containing crosstalk genes, transcription factors, and associated pathways was developed.Results: Four gene expression datasets, including GSE66676 and GSE48452 for NAFLD and GSE30122 and GSE1009 for DN, were involved in this study. There were 80 common DEGs between the two diseases in total. The PPI network built with the 80 common genes included 77 nodes and 83 edges. Ten optimal crosstalk genes were selected by LASSO regression and Boruta algorithm, including CD36, WIPI1, CBX7, FCN1, SLC35D2, CP, ZDHHC3, PTPN3, LPL, and SPP1. Among these genes, LPL and SPP1 were the most significant according to NAFLD-transcription factor network. Five hundred twenty-nine nodes and 1,113 edges comprised the PPI network of activated pathway-gene. In addition, 14 common pathways of these two diseases were recognized using Gene Ontology (GO) analysis; among them, regulation of the lipid metabolic process is closely related to both two diseases.Conclusions: This study offers hints that NAFLD and DN have a common pathogenesis, and LPL and SPP1 are the most relevant crosstalk genes. Based on the common pathways and optimal crosstalk genes, our proposal carried out further research to disclose the etiology and pathology between the two diseases.

Background This is an updated version of the Cochrane Review first published in 2011, and most recently updated in 2019. Epilepsy is a chronic and disabling neurological disorder, affecting approximately 1% of the population. Up to 30% of people with epilepsy have seizures that are resistant to currently available antiepileptic drugs and require treatment with multiple antiepileptic drugs in combination. Felbamate is a second‐generation antiepileptic drug that can be used as add‐on therapy to standard antiepileptic drugs. Objectives To evaluate the efficacy and tolerability of felbamate versus placebo when used as an add‐on treatment for people with drug‐resistant focal‐onset epilepsy. Search methods For the latest update, we searched the Cochrane Register of Studies (CRS Web) and MEDLINE (Ovid, 1946 to 13 July 2021) on 15 July 2021. There were no language or time restrictions. We reviewed the reference lists of retrieved studies to search for additional reports of relevant studies. We also contacted the manufacturers of felbamate and experts in the field for information about any unpublished or ongoing studies. Selection criteria We searched for randomised placebo‐controlled add‐on studies of people of any age with drug‐resistant focal seizures. The studies could be double‐blind, single‐blind or unblinded and could be of parallel‐group or cross‐over design. Data collection and analysis Two review authors independently selected studies for inclusion and extracted information. In the case of disagreements, a third review author arbitrated. Review authors assessed the following outcomes: 50% or greater reduction in seizure frequency; absolute or percentage reduction in seizure frequency; treatment withdrawal; adverse effects; quality of life. Main results We included four randomised controlled trials, representing a total of 236 participants, in the review. Two trials had parallel‐group design, the third had a two‐period cross‐over design, and the fourth had a three‐period cross‐over design. We judged all four studies to be at an unclear risk of bias overall. Bias arose from the incomplete reporting of methodological details, the incomplete and selective reporting of outcome data, and from participants having unstable drug regimens during experimental treatment in one trial. Due to significant methodological heterogeneity, clinical heterogeneity and differences in outcome measures, it was not possible to perform a meta‐analysis of the extracted data. Only one study reported the outcome of 50% or greater reduction in seizure frequency, whilst three studies reported percentage reduction in seizure frequency compared to placebo. One study claimed an average seizure reduction of 35.8% with add‐on felbamate whilst another study claimed a more modest reduction of 4.2%. Both studies reported that seizure frequency increased with add‐on placebo and that there was a significant difference in seizure reduction between felbamate and placebo (P = 0.0005 and P = 0.018, respectively). The third study reported a 14% reduction in seizure frequency with add‐on felbamate but stated that the difference between treatments was not significant. There were conflicting results regarding treatment withdrawal. One study reported a higher treatment withdrawal for placebo‐randomised participants, whereas the other three studies reported higher treatment withdrawal rates for felbamate‐randomised participants. Notably, the treatment withdrawal rates for felbamate treatment groups across all four studies remained reasonably low (less than 10%), suggesting that felbamate may be well tolerated. Felbamate‐randomised participants most commonly withdrew from treatment due to adverse effects. The adverse effects consistently reported by all four studies were headache, dizziness and nausea. All three adverse effects were reported by 23% to 40% of felbamate‐treated participants versus 3% to 15% of placebo‐treated participants. We assessed the evidence for all outcomes using GRADE and rated the evidence as very low certainty, meaning that we have little confidence in the findings reported. We mainly downgraded evidence for imprecision due to the narrative synthesis conducted and the low number of events. We stress that the true effect of felbamate could likely be significantly different from that reported in this current review update. Authors' conclusions In view of the methodological deficiencies, the limited number of included studies and the differences in outcome measures, we have found no reliable evidence to support the use of felbamate as an add‐on therapy in people with drug‐resistant focal‐onset epilepsy. A large‐scale, randomised controlled trial conducted over a longer period of time is required to inform clinical practice.

Background: Diabetic kidney disease (DKD) patients are facing an extremely high risk of cardiovascular disease (CVD), which is a major cause of death for DKD patients. We aimed to build a deep learning model to predict CVD risk among DKD patients and perform risk stratifying, which could help them perform early intervention and improve personal health management. Methods: A retrospective cohort study was conducted to assess the risk of the occurrence of composite cardiovascular disease, which includes coronary heart disease, cerebrovascular diseases, congestive heart failure, and peripheral artery disease, in DKD patients. A least absolute shrinkage and selection operator (LASSO) regression was used to perform the variable selection. A deep learning-based survival model called DeepSurv, based on a feed-forward neural network was developed to predict CVD risk among DKD patients. We compared the model performance with the conventional Cox proportional hazards (CPH) model and the Random survival forest (RSF) model using the concordance index (C-index), the area under the curve (AUC), and integrated Brier scores (IBS). Results: We recruited 890 patients diagnosed with DKD in this retrospective study. During a median follow-up of 10.4 months, there are 289 patients who sustained a subsequent CVD. Seven variables, including age, high density lipoprotein (HDL), hemoglobin (Hb), systolic blood pressure (SBP), smoking status, 24 h urinary protein excretion, and total cholesterol (TC), chosen by LASSO regression were used to develop the predictive model. The DeepSurv model showed the best performance, achieved a C-index of 0.767(95% confidence intervals [CI]: 0.717–0.817), AUC of 0.780(95%CI: 0.721–0.839), and IBS of 0.067 in the validation set. Then we used the cut-off value determined by ROC (receiver operating characteristic) curve to divide the patients into different risk groups. Moreover, the DeepSurv model was also applied to develop an online calculation tool for patients to conduct risk monitoring. Conclusion: A deep-learning-based predictive model using seven clinical variables can effectively predict CVD risk among DKD patients and perform risk stratification. An online calculator allows its easy implementation.

COVID-19, caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), spread rapidly and affected most of the world since its outbreak in Wuhan, China, which presents a major challenge to the emergency response mechanism for sudden public health events and epidemic prevention and control in all countries. In the face of the severe situation of epidemic prevention and control and the arduous task of social management, the tremendous power of science and technology in prevention and control has emerged. The new generation of information technology, represented by big data and artificial intelligence (AI) technology, has been widely used in the prevention, diagnosis, treatment and management of COVID-19 as an important basic support. Although the technology has developed, there are still challenges with respect to epidemic surveillance, accurate prevention and control, effective diagnosis and treatment, and timely judgement. The prevention and control of sudden infectious diseases usually depend on the control of infection sources, interruption of transmission channels and vaccine development. Big data and AI are effective technologies to identify the source of infection and have an irreplaceable role in distinguishing close contacts and suspicious populations. Advanced computational analysis is beneficial to accelerate the speed of vaccine research and development and to improve the quality of vaccines. AI provides support in automatically processing relevant data from medical images and clinical features, tests and examination findings; predicting disease progression and prognosis; and even recommending treatment plans and strategies. This paper reviews the application of big data and AI in the COVID-19 prevention, diagnosis, treatment and management decisions in China to explain how to apply big data and AI technology to address the common problems in the COVID-19 pandemic. Although the findings regarding the application of big data and AI technologies in sudden public health events lack validation of repeatability and universality, current studies in China have shown that the application of big data and AI is feasible in response to the COVID-19 pandemic. These studies concluded that the application of big data and AI technology can contribute to prevention, diagnosis, treatment and management decision making regarding sudden public health events in the future.

Industry 4.0 places special emphasis on the use of intelligent models to discover patterns in data. In this study, we propose a novel TSK fuzzy system with low model complexity for multi-view data pattern discovery. Compared with the classic TSK fuzzy systems, the proposed one has two three merits: 1) we introduce a transformation matrix to relax the strict binary label matrix of the training set so that the margins between classes become more discriminative. 2) we introduce two kinds of sparsity regularizations, i.e., cross-rule and cross-view, to reduce indiscriminative fuzzy rules and consequent parameters so that the model complexity is significant reduced. 3) We introduce the alternating direction method of multipliers (ADMM) to optimize the objective function so that we have compact closed-form solutions in each iteration. Extensive experiments on different kinds of multi-view image datasets indicate the promising performance for data pattern discovery with low model complexity.

Previous studies have explored the association between retinal vascular changes and cognitive impairment. The retinal vasculature shares some characteristics with the cerebral vasculature, and quantitative changes in it could indicate cognitive impairment. Hence, a comprehensive meta-analysis was performed to clarify the potential relationship between retinal vascular geometric changes and cognitive impairment. Relevant databases were scrupulously and systematically searched for retinal vascular geometric changes including caliber, tortuosity, and fractal dimension (FD), and for cognitive impairment. The Newcastle-Ottawa Scale was used to evaluate the methodological quality of included studies. RevMan was used to perform the meta-analysis and detect publication bias. Sensitivity analyses were also performed. Five studies that involved 2,343 subjects were finally included in the meta-analysis. The results showed that there was no significant association between central retinal artery equivalents (Z=1.17) or central retinal venular equivalents (Z=1.74) and cognitive impairment (both p>0.05). Similarly, no significant difference was detected in retinal arteriolar tortuosity (Z=0.91) and venular tortuosity (Z=1.31) (both p>0.05). However, the retinal arteriolar FD (mean difference: -0.03, 95% CI: -0.05, -0.01) and venular FD (mean difference: -0.03, 95% CI: -0.05, -0.02) were associated with cognitive impairment. A smaller retinal microvascular FD might be associated with cognitive impairment. Further large-sample and well-controlled original studies are required to confirm the present findings.

AIM: To ensure the diagnostic value of computer aided techniques in diabetic retinopathy (DR) detection based on ophthalmic photography (OP). METHODS: PubMed, EMBASE, Ei village, IEEE Xplore and Cochrane Library database were searched systematically for literatures about computer aided detection (CAD) in DR detection. The methodological quality of included studies was appraised by the Quality Assessment Tool for Diagnostic Accuracy Studies (QUADAS-2). Meta-DiSc was utilized and a random effects model was plotted to summarize data from those included studies. Summary receiver operating characteristic curves were selected to estimate the overall test performance. Subgroup analysis was used to identify the efficiency of CAD in detecting DR, exudates (EXs), microaneurysms (MAs) as well as hemorrhages (HMs), and neovascularizations (NVs). Publication bias was analyzed using STATA. RESULTS: Fourteen articles were finally included in this Meta-analysis after literature review. Pooled sensitivity and specificity were 90% (95%CI, 85%-94%) and 90% (95% CI, 80%-96%) respectively for CAD in DR detection. With regard to CAD in EXs detecting, pooled sensitivity, specificity were 89% (95%CI, 88%-90%) and 99% (95%CI, 99%-99%) respectively. In aspect of MAs and HMs detection, pooled sensitivity and specificity of CAD were 42% (95%CI, 41%-44%) and 93% (95%CI, 93%-93%) respectively. Besides, pooled sensitivity and specificity were 94% (95%CI, 89%-97%) and 87% (95%CI, 83%-90%) respectively for CAD in NVs detection. No potential publication bias was observed. CONCLUSION: CAD demonstrates overall high diagnostic accuracy for detecting DR and pathological lesions based on OP. Further prospective clinical trials are needed to prove such effect.

Background Epilepsy is a common neurological condition, with an estimated incidence of 50 per 100,000 persons. People with epilepsy may present with various types of immunological abnormalities, such as low serum immunoglobulin A (IgA) levels, lack of the immunoglobulin G (IgG) subclass and identification of certain types of antibodies. Intravenous immunoglobulin (IVIg) treatment may represent a valuable approach and its efficacy has important implications for epilepsy management. This is an update of a Cochrane review first published in 2011 and last updated in 2017. Objectives To examine the effects of IVIg on the frequency and duration of seizures, quality of life and adverse effects when used as monotherapy or as add‐on treatment for people with epilepsy. Search methods For the latest update, we searched the Cochrane Register of Studies (CRS Web) (20 December 2018), MEDLINE (Ovid, 1946 to 20 December 2018), Web of Science (1898 to 20 December 2018), ISRCTN registry (20 December 2018), WHO International Clinical Trials Registry Platform (ICTRP, 20 December 2018), the US National Institutes of Health ClinicalTrials.gov (20 December 2018), and reference lists of articles. Selection criteria Randomised or quasi‐randomised controlled trials of IVIg as monotherapy or add‐on treatment in people with epilepsy. Data collection and analysis Two review authors independently assessed the trials for inclusion and extracted data. We contacted study authors for additional information. Outcomes included percentage of people rendered seizure‐free, 50% or greater reduction in seizure frequency, adverse effects, treatment withdrawal and quality of life. Main results We included one study (61 participants). The included study was a randomised, double‐blind, placebo‐controlled, multicentre trial which compared the treatment efficacy of IVIg as an add‐on with a placebo add‐on in patients with drug‐resistant epilepsy. Seizure freedom was not reported in the study. There was no significant difference between IVIg and placebo in 50% or greater reduction in seizure frequency (RR 1.89, 95% CI 0.85 to 4.21; one study, 58 participants; low‐certainty evidence). The study reported a statistically significant effect for global assessment in favour of IVIg (RR 3.29, 95% CI 1.13 to 9.57; one study, 60 participants; low‐certainty evidence). No adverse effects were demonstrated. We found no randomised controlled trials that investigated the effects of IVIg monotherapy for epilepsy. Overall, the included study was rated at low to unclear risk of bias. Using GRADE methodology, the certainty of the evidence was rated as low. Authors' conclusions We cannot draw any reliable conclusions regarding the efficacy of IVIg as a treatment for epilepsy. Further randomised controlled trials are needed.

Takagi-Sugeno-Kang (TSK) fuzzy systems are well known for their good balances between approximation accuracy and interpretability. Among a wide variety of existing TSK fuzzy systems, most of them are driven by special knowledge since the learned parameters of each fuzzy rule are totally different. However, common knowledge is equally important and useful in practice and hence a TSK fuzzy system embedded with common knowledge should be more intuitive and interpretable when tackling with real-world problems. In this paper, we propose a common and special knowledge-driven TSK fuzzy system (CSK-TSK-FS), in which the parameters corresponding to each feature in then-parts of fuzzy rules always keep invariant and these parameters are viewed as common knowledge. As for its modeling, except the gradient descent techniques and other existing training algorithms, we can obtain a trained CSK-TSK-FS from a trained GMM or a trained FLNN because the proposed fuzzy system CSK-TSK-FS is mathematically equivalent to a special GMM and a FLNN. CSK-TSK-FS has three characteristics: (1) with the classical centroid defuzzification strategy, the involved common knowledge can be separated from fuzzy rules such that the interpretability of CSK-TSK-FS can be enhanced; (2) it can be trained quickly by the proposed LLM-based training algorithm; (3) the equivalence relationships among CSK-TSK-FS, GMM and FLNN allow them to share some commonality in training such that the proposed LLM-based training algorithm provides a novel fast training tool for training GMM and FLNN. Experimental results on UCI, KEEL and epileptic EEG datasets demonstrate the promising classification of CSK-TSK-FS.