Deep Learning Models for the Screening of Cognitive Impairment Using Multimodal Fundus Images.

OBJECTIVE: We aimed to develop a deep learning system capable of identifying subjects with cognitive impairment quickly and easily based on multimodal ocular images. DESIGN: Cross sectional study. SUBJECTS: Participants of Beijing Eye Study 2011 and patients attending Beijing Tongren Eye Center and Beijing Tongren Hospital Physical Examination Center. METHODS: We trained and validated a deep learning algorithm to assess cognitive impairment using retrospectively collected data from the Beijing Eye Study 2011. Cognitive impairment was defined as a Mini-Mental State Examination score < 24. Based on fundus photographs and OCT images, we developed 5 models based on the following sets of images: macula-centered fundus photographs, optic disc-centered fundus photographs, fundus photographs of both fields, OCT images, and fundus photographs of both fields with OCT (multimodal). The performance of the models was evaluated and compared in an external validation data set, which was collected from patients attending Beijing Tongren Eye Center and Beijing Tongren Hospital Physical Examination Center. MAIN OUTCOME MEASURES: Area under the curve (AUC). RESULTS: A total of 9424 retinal photographs and 4712 OCT images were used to develop the model. The external validation sets from each center included 1180 fundus photographs and 590 OCT images. Model comparison revealed that the multimodal performed best, achieving an AUC of 0.820 in the internal validation set, 0.786 in external validation set 1, and 0.784 in external validation set 2. We evaluated the performance of the multi-model in different sexes and different age groups; there were no significant differences. The heatmap analysis showed that signals around the optic disc in fundus photographs and the retina and choroid around the macular and optic disc regions in OCT images were used by the multimodal to identify participants with cognitive impairment. CONCLUSIONS: Fundus photographs and OCT can provide valuable information on cognitive function. Multimodal models provide richer information compared with single-mode models. Deep learning algorithms based on multimodal retinal images may be capable of screening cognitive impairment. This technique has potential value for broader implementation in community-based screening or clinic settings. FINANCIAL DISCLOSURE(S): Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article.

Tobacco exposure primes the secretion of CCL21 positively associated with tertiary lymphoid structure and response to immunotherapy.

BACKGROUND: It has been reported that smoking history as a predictor of immunotherapy efficacy in patients with advanced lung cancer, however, the underlying mechanisms of this phenomenon remain largely unknown. METHODS: The patients with lung adenocarcinoma's (LUAD) cohort and the orthotopical transplanted mouse model were used to explore the correlation between smoking status and tertiary lymphoid structure (TLS) and chemokine CCL21, respectively. Cell adhesion and co-immunoprecipitation assays were performed to explore the interaction between CD4+T cells and CD20+B cells under tobacco exposure. Chromatin immunoprecipitation-PCR was used to dissect the mechanism of upregulated CCL21 secretion in tobacco treatment. Serum CCL21 level was recorded in patients with LUAD treated with immunotherapy. RESULTS: Here we observed that individuals with a smoking history exhibit an increased quantity and maturation level of TLS compared with non-smokers, along with higher levels of CCL21 secretion. Tobacco exposure promoted CCL21 expression in an epithelial cell-intrinsic manner, of which BaP, the main component of tobacco, facilitated the nuclear retention of the aryl hydrocarbon receptor that occupied the promoter of CCL21. Additionally, the activated CCL21/CCR7 axis increased the CD11a expression of CD4+T cells, boosting the interaction with CD20+B cells dependent on ICAM1, which potentially induced the TLSs formation. Patients with elevated serum levels of CCL21 benefited more from immunotherapy. CONCLUSIONS: Patients with a smoking history exhibited higher levels of TLS via the CCL21-dependent mechanism, serum CCL21 was identified as a reliable biomarker for predicting the efficacy of immunotherapy.

Definition of a new blood cell count score for early survival prediction for non-small cell lung cancer patients treated with atezolizumab: Integrated analysis of four multicenter clinical trials.

Importance: Blood cell count test (BCT) is a robust method that provides direct quantification of various types of immune cells to reveal the immune landscape to predict atezolizumab treatment outcomes for clinicians to decide the next phase of treatment. Objective: This study aims to define a new BCTscore model to predict atezolizumab treatment benefits in non-small lung cell cancer (NSCLC) patients. Design Setting and Participants: This study analyzed four international, multicenter clinical trials (OAK, BIRCH, POPLAR, and FIR trials) to conduct post-hoc analyses of NSCLC patients undergoing atezolizumab (anti-PD-L1) single-agent treatment (n = 1,479) or docetaxel single-agent treatment (n = 707). BCT was conducted at three time points: pre-treatment (T1), the first day of treatment cycle 3 (T2), and first day of treatment cycle 5 (T3). Univariate and multivariate Cox regression analyses were conducted to identify early BCT biomarkers to predict atezolizumab treatment outcomes in NSCLC patients. Main Outcomes and Measures: Overall survival (OS) was used as the primary end point, whereas progression-free survival (PFS) according to Response Evaluation Criteria in Solid Tumors (RECIST), clinical benefit (CB), and objective response rate (ORR) were used as secondary end points. Results: The BCT biomarkers of neutrophil-to-lymphocyte ratio (NLR) and platelet-to-lymphocyte ratio (PLR) at time point T3 and neutrophil-to-monocyte ratio (NMR) at time point T2 with absolute cutoff values of NLR_T3 = 5, PLR_T3 = 180, and NMR_T2 = 6 were identified as strong predictive biomarkers for atezolizumab (Ate)-treated NSCLC patients in comparison with docetaxel (Dtx)-treated patients regarding OS (BCTscore low risk: HR Ate vs. Dtx = 1.54 (95% CI: 1.04-2.27), P = 0.031; high risk: HR Ate vs. Dtx = 0.84 (95% CI: 0.62-1.12), P = 0.235). The identified BCTscore model showed better OS AUC in the OAK (AUC12month = 0.696), BIRCH (AUC12month = 0.672) and POPLAR+FIR studies (AUC12month = 0.727) than that of each of the three single BCT biomarkers. Conclusion and Relevance: The BCTscore model is a valid predictive and prognostic biomarker for early survival prediction in atezolizumab-treated NSCLC patients.

Artificial Intelligence Model for Antiinterference Cataract Automatic Diagnosis: A Diagnostic Accuracy Study.

Background: Cataract is the leading cause of blindness worldwide. In order to achieve large-scale cataract screening and remarkable performance, several studies have applied artificial intelligence (AI) to cataract detection based on fundus images. However, the fundus images they used are original from normal optical circumstances, which is less impractical due to the existence of poor-quality fundus images for inappropriate optical conditions in actual scenarios. Furthermore, these poor-quality images are easily mistaken as cataracts because both show fuzzy imaging characteristics, which may decline the performance of cataract detection. Therefore, we aimed to develop and validate an antiinterference AI model for rapid and efficient diagnosis based on fundus images. Materials and Methods: The datasets (including both cataract and noncataract labels) were derived from the Chinese PLA general hospital. The antiinterference AI model consisted of two AI submodules, a quality recognition model for cataract labeling and a convolutional neural networks-based model for cataract classification. The quality recognition model was performed to distinguish poor-quality images from normal-quality images and further generate the pseudo labels related to image quality for noncataract. Through this, the original binary-class label (cataract and noncataract) was adjusted to three categories (cataract, noncataract with normal-quality images, and noncataract with poor-quality images), which could be used to guide the model to distinguish cataract from suspected cataract fundus images. In the cataract classification stage, the convolutional-neural-network-based model was proposed to classify cataracts based on the label of the previous stage. The performance of the model was internally validated and externally tested in real-world settings, and the evaluation indicators included area under the receiver operating curve (AUC), accuracy (ACC), sensitivity (SEN), and specificity (SPE). Results: In the internal and external validation, the antiinterference AI model showed robust performance in cataract diagnosis (three classifications with AUCs >91%, ACCs >84%, SENs >71%, and SPEs >89%). Compared with the model that was trained on the binary-class label, the antiinterference cataract model improved its performance by 10%. Conclusion: We proposed an efficient antiinterference AI model for cataract diagnosis, which could achieve accurate cataract screening even with the interference of poor-quality images and help the government formulate a more accurate aid policy.

Slowly Digestible Carbohydrate Diet Ameliorates Hyperglycemia and Hyperlipidemia in High-Fat Diet/Streptozocin-Induced Diabetic Mice.

Objective: Given that the prevalence rate of type 2 diabetes mellitus (T2DM) continues to increase, it is important to find an effective method to prevent or treat this disease. Previous studies have shown that dietary intervention with a slowly digestible carbohydrate (SDC) diet can improve T2DM with almost no side effects. However, the underlying mechanisms of SDC protect against T2DM remains to be elucidated. Methods: The T2DM mice model was established with a high-fat diet and streptozocin injection. Then, SDC was administered for 6 weeks. Bodyweight, food intake, organ indices, fasting blood glucose (FBG), oral glucose tolerance test (OGTT), homeostasis model assessment for insulin resistance (HOMA-IR), and other biochemical parameters were measured. Histopathological and lipid accumulation analyses were performed, and the glucose metabolism-related gene expressions in the liver and skeletal muscle were determined. Lastly, colonic microbiota was also analyzed. Results: SDC intervention alleviated the weight loss in the pancreas, lowered blood glucose and glycosylated hemoglobin levels, and improved glucose tolerance and HOMA-IR. SDC intervention improved serum lipid profile, adipocytokines levels, and lowered the lipid accumulation in the liver, subcutaneous adipose tissue, and epididymal visceral adipose tissue. In addition, SDC intervention increased the expression levels of IRS-2 and GLUT-2 in liver tissues and elevated GLUT-4 expression levels in skeletal muscle tissues. Notably, SDC intervention decreased the Bacteroidetes/Firmicutes ratio, increased Desulfovibrio and Lachnospiraceae genus levels, and inhibited the relative abundance of potentially pathogenic bacteria. Conclusions: SDC intervention can improve hyperglycemia and hyperlipidemia status in diabetic mice, suggesting that this intervention might be beneficial for T2DM.

Preventive effect of sanguinarine on intestinal injury in mice exposed to whole abdominal irradiation.

Intestinal injury is one of the major side effects that are induced by medical radiation exposure, and has limited effective therapies. In this study, we investigated the beneficial effects of sanguinarine (SAN) on intestinal injury induced by ionizing radiation (IR) both in vitro and in vivo. Mice were exposed to whole abdominal irradiation (WAI) to mimic clinical scenarios. SAN was injected intraperitoneally to mitigate IR-induced injury. Histological examination was performed to assess the tissue injuries of the spleen and small intestine. A small intestinal epithelial cell line-6 (IEC-6) was analyzed for its viability and apoptosis in vitro under different treatments. Inflammation-related pathways and serum inflammatory cytokines were detected via Western blot analysis and ELISA, respectively. High-throughput sequencing was used to characterize the gut microbiota profile. High-performance liquid chromatography was performed to assess short-chain fatty acid contents in the colon. In vitro, SAN pretreatment protected cell viability and reduced apoptosis in IEC-6 cells. In vivo, SAN pretreatment protected immune organs, alleviated intestinal injury, and promoted intestinal recovery. SAN also reduced the levels of inflammatory cytokines, suppressed high mobility group box 1 (HMGB1)/ Toll-like receptor 4 (TLR4) pathway activation, and modulated gut microbiota composition. Our findings demonstrate that the beneficial properties of SAN alleviated intestinal radiation injury. Thus, SAN represents a therapeutic option for protecting against IR-induced intestinal injury in preclinical settings.

p-Coumaric acid ameliorates ionizing radiation-induced intestinal injury through modulation of oxidative stress and pyroptosis.

AIMS: Intestinal injury is a clinical problem related to radiotherapy or accidental exposure to ionizing radiation. This study aimed to investigate the protective effect of p-coumaric acid (CA) against radiation induced intestinal injury. MAIN METHODS: The present study orally administered CA to C57BL/6 male mice at 30 min before total body irradiation and continued for 3 days post irradiation. Then, the mice were sacrificed at day 3.5 or 14 after irradiation, respectively. The blood was collected to analyze the inflammatory cytokines. The antioxidant indexes of jejunum tissues were determined. Hematoxylin and eosin staining and apoptosis analysis was studied to investigate the pathological changes of the jejunum tissues. In addition, quantitative real-time polymerase chain reaction (qRT-PCR) and western blot were carried out to determine the changes in mRNA and protein levels of jejunum tissues. KEY FINDINGS: Compared with the only irradiated group, treatment with CA improved intestinal morphology and apoptosis, increased the villus height and the ratio of villus height to crypt depth. It also reduced the oxidative stress and inflammatory response. The molecular mechanism analysis showed that CA significantly inhibited the pyroptosis genes (Caspase-1, NLRP3 and AIM2) mRNA expression and improved the intestinal barrier genes expression. SIGNIFICANCE: The results suggested that CA ameliorates ionizing radiation-induced intestinal injury by inhibition of oxidative stress, inflammatory response and pyroptosis.

Anti-Oxidative and Immuno-Protective Effect of Camel Milk on Radiation-Induced Intestinal Injury in C57BL/6 J Mice.

PURPOSE: The main objective is to investigate the protective effect of camel milk (CM) on radiation-induced intestinal injury. METHODS: The C57BL/6 J mice in 2 experiments were assigned into control group (Con), irradiation group (IR), and CM+irradiation group (CM+IR). After receiving the CM via gavage for 14 days, the mice in the first experiment were exposed to 6 Gy X-ray whole body irradiation, and survival rate was compared among the groups. Mice in the second experiment were exposed to 4 Gy irradiation and sacrificed at day 7. The small intestines were collected to examine the histopathological changes and to determine the anti-oxidative index and HMGB1/TLR4 inflammatory pathway. Fasting blood was used to measure serum pro-inflammatory factors. RESULTS: Compared with the IR group, the survival time was prolonged, and survival rate was increased in the CM+IR group. CM increased levels of SOD and GSH and decreased MDA in the jejunum. Furthermore, intestinal protein expression of HMGB1/TLR4 pathway (TLR4, NF-κB, and HMGB1) was up-regulated by CM intervention. CM decreased the serum levels of TNF-α and IL-1ß and increased IL-10 level. CONCLUSIONS: CM extended the survival time and had a protective effect against radiation-induced jejunum injury by regulation of antioxidant capacity and HMGB1/TLR4/NF-κB/MyD88 inflammatory signaling pathway.

At What Dose Can Total Body and Whole Abdominal Irradiation Cause Lethal Intestinal Injury Among C57BL/6J Mice?

PURPOSE AND METHODS: To investigate the doses of total body (TBI) and whole abdominal irradiation (WAI) induced lethal intestinal injury, healthy C57BL/6 J mice were divided randomly into 7 groups: control group; 6, 7, and 8 Gy TBI groups; and 5, 10, and 15 Gy WAI groups. The survival length, general conditions, body weight, daily food and water intake of the mice and the histopathological changes of small intestine were observed. RESULTS: Lethal injury among C57BL/6 J mice was caused by ≥6 Gy TBI and 15 Gy WAI. Their body weight and food intake decreased, the structure of their small intestinal villi was destroyed, and the number of surviving crypts per circumference of the jejunum decreased in ≥6 Gy TBI groups and 15 Gy WAI group. The mice in the 10 Gy WAI group significantly lost weight within 5 days but recovered slowly thereafter. They also had poor appetite and reversibly damaged intestinal mucosa. CONCLUSIONS: Nonlethal intestinal injury could be induced by 10 Gy WAI, whereas lethal intestinal injury could be triggered by ≥6 Gy TBI and >15 Gy WAI in mice. Our results provided a basis for establishing radiation-induced intestinal injury models with C57BL/6 J mice.

Simultaneous determination of multiclass illegal dyes with different acidic-basic properties in foodstuffs by LC-MS/MS via polarity switching mode.

Simultaneous determination of multiclass illegal dyes possessing different chemical properties is difficult. By using LC-MS/MS via negative/positive ion switching mode, an efficient and fast multi-residual method for simultaneous determination of multiclass 52 illegal dyes with different acidic-basic properties in foodstuffs was developed and validated during one single run, including 23 fat-soluble neutral azo dyes, 8 acidic sulfonated azo dyes, 12 triphenylmethane basic dyes, three basic indole dyes, three xanthene dyes, one quinoline dye, and two anthraquinones dyes. The illegal dyes were extracted with methanol-acetonitrile and further purified with d-SPE procedure to reduce interference. Sample dilution with 100-fold was used for the elimination of matrix effects of the quantitation of LC-MS/MS analysis. Validation data showed the good recoveries in the range of 71.2-111.2%, with relative standard deviations less than 20%, suggesting the developed method is suitable for the identification and quantitation of multiclass illegal dyes at trace levels in foods.

The Constipation-Relieving Property of d-Tagatose by Modulating the Composition of Gut Microbiota.

d-tagatose, a monosaccharide as well as a dietary supplement, has been reported as having a wide range of applicability in the food industry, however, the prebiotic activity, anticonstipation effects, and related mechanisms are still unclear. In this study, using the loperamide-induced constipation Kunming mice as the animal model, the effects of d-tagatose for the prevention of constipation were evaluated by gastrointestinal transit experiment and defecation experiment. Furthermore, the underlying mechanism was clarified by evaluating the change of the biochemical indicators and analyzing 16S rRNA amplicon of gut microbiota among the different mice groups. The results showed that the gastrointestinal transit rate, fecal number, and weight in six hours were significantly enhanced after the administration of d-tagatose. In addition, d-tagatose significantly increased the serum levels of acetylcholine (Ach) and substance P (SP), whereas the serum levels of nitric oxide (NO) were significantly decreased. Moreover, the 16S rRNA sequencing analysis revealed that the changes in the gut microbiota caused by constipation were restored by d-tagatose treatment. In conclusion, this study indicated that the administration of d-tagatose as a dietary supplement can effectively prevent and relieve constipation in Kunming mice, and it is a promising prebiotic candidate with constipation-relieving properties.

Sparse dynamical Boltzmann machine for reconstructing complex networks with binary dynamics.

Revealing the structure and dynamics of complex networked systems from observed data is a problem of current interest. Is it possible to develop a completely data-driven framework to decipher the network structure and different types of dynamical processes on complex networks? We develop a model named sparse dynamical Boltzmann machine (SDBM) as a structural estimator for complex networks that host binary dynamical processes. The SDBM attains its topology according to that of the original system and is capable of simulating the original binary dynamical process. We develop a fully automated method based on compressive sensing and a clustering algorithm to construct the SDBM. We demonstrate, for a variety of representative dynamical processes on model and real world complex networks, that the equivalent SDBM can recover the network structure of the original system and simulates its dynamical behavior with high precision.

Dexmedetomidine mitigates isoflurane-induced neurodegeneration in fetal rats during the second trimester of pregnancy.

Dexmedetomidine has significant neuroprotective effects. However, whether its protective effects can reduce neurotoxicity caused by isoflurane in fetal brain during the second trimester of pregnancy remains unclear. In this study, timed-pregnancy rats at gestational day 14 spontaneously inhaled 1.5% isoflurane for 4 hours, and were intraperitoneally injected with dexmedetomidine at dosages of 5, 10, 20, and 20 µg/kg 15 minutes before inhalation and after inhalation for 2 hours. Our results demonstrate that 4 hours after inhaling isoflurane, 20 µg/kg dexmedetomidine visibly mitigated isoflurane-induced neuronal apoptosis, reversed downregulation of brain-derived neurotrophic factor expression, and lessened decreased spatial learning and memory ability in adulthood in the fetal rats. Altogether, these findings indicate that dexmedetomidine can reduce neurodegeneration induced by isoflurane in fetal rats during the second trimester of pregnancy. Further, brain-derived neurotrophic factor participates in this process.

Physical controllability of complex networks.

A challenging problem in network science is to control complex networks. In existing frameworks of structural or exact controllability, the ability to steer a complex network toward any desired state is measured by the minimum number of required driver nodes. However, if we implement actual control by imposing input signals on the minimum set of driver nodes, an unexpected phenomenon arises: due to computational or experimental error there is a great probability that convergence to the final state cannot be achieved. In fact, the associated control cost can become unbearably large, effectively preventing actual control from being realized physically. The difficulty is particularly severe when the network is deemed controllable with a small number of drivers. Here we develop a physical controllability framework based on the probability of achieving actual control. Using a recently identified fundamental chain structure underlying the control energy, we offer strategies to turn physically uncontrollable networks into physically controllable ones by imposing slightly augmented set of input signals on properly chosen nodes. Our findings indicate that, although full control can be theoretically guaranteed by the prevailing structural controllability theory, it is necessary to balance the number of driver nodes and control cost to achieve physical control.

Energy scaling and reduction in controlling complex networks.

Recent works revealed that the energy required to control a complex network depends on the number of driving signals and the energy distribution follows an algebraic scaling law. If one implements control using a small number of drivers, e.g. as determined by the structural controllability theory, there is a high probability that the energy will diverge. We develop a physical theory to explain the scaling behaviour through identification of the fundamental structural elements, the longest control chains (LCCs), that dominate the control energy. Based on the LCCs, we articulate a strategy to drastically reduce the control energy (e.g. in a large number of real-world networks). Owing to their structural nature, the LCCs may shed light on energy issues associated with control of nonlinear dynamical networks.

Extreme events in multilayer, interdependent complex networks and control.

We investigate the emergence of extreme events in interdependent networks. We introduce an inter-layer traffic resource competing mechanism to account for the limited capacity associated with distinct network layers. A striking finding is that, when the number of network layers and/or the overlap among the layers are increased, extreme events can emerge in a cascading manner on a global scale. Asymptotically, there are two stable absorption states: a state free of extreme events and a state of full of extreme events, and the transition between them is abrupt. Our results indicate that internal interactions in the multiplex system can yield qualitatively distinct phenomena associated with extreme events that do not occur for independent network layers. An implication is that, e.g., public resource competitions among different service providers can lead to a higher resource requirement than naively expected. We derive an analytical theory to understand the emergence of global-scale extreme events based on the concept of effective betweenness. We also articulate a cost-effective control scheme through increasing the capacity of very few hubs to suppress the cascading process of extreme events so as to protect the entire multi-layer infrastructure against global-scale breakdown.

Spatiotemporal patterns and predictability of cyberattacks.

A relatively unexplored issue in cybersecurity science and engineering is whether there exist intrinsic patterns of cyberattacks. Conventional wisdom favors absence of such patterns due to the overwhelming complexity of the modern cyberspace. Surprisingly, through a detailed analysis of an extensive data set that records the time-dependent frequencies of attacks over a relatively wide range of consecutive IP addresses, we successfully uncover intrinsic spatiotemporal patterns underlying cyberattacks, where the term "spatio" refers to the IP address space. In particular, we focus on analyzing macroscopic properties of the attack traffic flows and identify two main patterns with distinct spatiotemporal characteristics: deterministic and stochastic. Strikingly, there are very few sets of major attackers committing almost all the attacks, since their attack "fingerprints" and target selection scheme can be unequivocally identified according to the very limited number of unique spatiotemporal characteristics, each of which only exists on a consecutive IP region and differs significantly from the others. We utilize a number of quantitative measures, including the flux-fluctuation law, the Markov state transition probability matrix, and predictability measures, to characterize the attack patterns in a comprehensive manner. A general finding is that the attack patterns possess high degrees of predictability, potentially paving the way to anticipating and, consequently, mitigating or even preventing large-scale cyberattacks using macroscopic approaches.

Controlling extreme events on complex networks.

Extreme events, a type of collective behavior in complex networked dynamical systems, often can have catastrophic consequences. To develop effective strategies to control extreme events is of fundamental importance and practical interest. Utilizing transportation dynamics on complex networks as a prototypical setting, we find that making the network "mobile" can effectively suppress extreme events. A striking, resonance-like phenomenon is uncovered, where an optimal degree of mobility exists for which the probability of extreme events is minimized. We derive an analytic theory to understand the mechanism of control at a detailed and quantitative level, and validate the theory numerically. Implications of our finding to current areas such as cybersecurity are discussed.

Optimizing cooperation on complex networks in the presence of failure.

Cooperation has been recognized as a fundamental driving force in many natural, social, and economic systems. We investigate whether, given a complex-networked system in which agents (nodes) interact with one another according to the rules of evolutionary games and are subject to failure or death, cooperation can prevail and be optimized. We articulate a control scheme to maximize cooperation by introducing a time tolerance, a time duration that sustains an agent even if its payoff falls below a threshold. Strikingly, we find that a significant cooperation cluster can emerge when the time tolerance is approximately uniformly distributed over the network. A heuristic theory is derived to understand the optimization mechanism, which emphasizes the role played by medium-degree nodes. Implications for policy making to prevent or mitigate large-scale cascading breakdown are pointed out.