To develop deep learning models for predicting Interoperative hypotension (IOH) using waveforms from arterial blood pressure (ABP), electrocardiogram (ECG), and electroencephalogram (EEG), and to ...determine whether combination ABP with EEG or CG improves model performance. Data were retrieved from VitalDB, a public data repository of vital signs taken during surgeries in 10 operating rooms at Seoul National University Hospital from January 6, 2005, to March 1, 2014. Retrospective data from 14,140 adult patients undergoing non-cardiac surgery with general anaesthesia were used. The predictive performances of models trained with different combinations of waveforms were evaluated and compared at time points at 3, 5, 10, 15 minutes before the event. The performance was calculated by area under the receiver operating characteristic (AUROC), area under the precision-recall curve (AUPRC), sensitivity and specificity. The model performance was better in the model using both ABP and EEG waveforms than in all other models at all time points (3, 5, 10, and 15 minutes before an event) Using high-fidelity ABP and EEG waveforms, the model predicted IOH with a AUROC and AUPRC of 0.935 0.932 to 0.938 and 0.882 0.876 to 0.887 at 5 minutes before an IOH event. The output of both ABP and EEG was more calibrated than that using other combinations or ABP alone. The results demonstrate that a predictive deep neural network can be trained using ABP, ECG, and EEG waveforms, and the combination of ABP and EEG improves model performance and calibration.
Hepatic steatosis is a process of abnormal lipid deposition within the liver cells, often caused by excessive alcohol uptake or obesity. A conventional in vitro model for hepatic steatosis uses a ...liver cell culture, treated with fatty acids and measures accumulation of lipids within the cells. This model does not recapitulate the complex process of absorption and metabolism of digestive lipids. Here, we introduce a gut–liver chip, which mimics the gut absorption and hepatic metabolism in a microfluidic chip. Absorption of fatty acids through gut layer and subsequent deposition within liver cells was demonstrated. Tumor necrosis factor‐α, butyrate, and α‐lipoic acid were chosen as model molecules that can affect hepatic steatosis via different mechanisms, and their effects were evaluated. Our results suggest that the gut–liver chip can mimic the absorption and accumulation of fatty acids in the gut and the liver.
Hepatic steatosis is a process of abnormal lipid deposition within the liver cells, often caused by excessive alcohol uptake or obesity. A conventional in vitro model for hepatic steatosis uses a liver cell culture, treated with fatty acids and measures accumulation of lipids within the cells.
We investigated neural representations for visual perception of 10 handwritten digits and six visual objects from a convolutional neural network (CNN) and humans using functional magnetic resonance ...imaging (fMRI). Once our CNN model was fine‐tuned using a pre‐trained VGG16 model to recognize the visual stimuli from the digit and object categories, representational similarity analysis (RSA) was conducted using neural activations from fMRI and feature representations from the CNN model across all 16 classes. The encoded neural representation of the CNN model exhibited the hierarchical topography mapping of the human visual system. The feature representations in the lower convolutional (Conv) layers showed greater similarity with the neural representations in the early visual areas and parietal cortices, including the posterior cingulate cortex. The feature representations in the higher Conv layers were encoded in the higher‐order visual areas, including the ventral/medial/dorsal stream and middle temporal complex. The neural representations in the classification layers were observed mainly in the ventral stream visual cortex (including the inferior temporal cortex), superior parietal cortex, and prefrontal cortex. There was a surprising similarity between the neural representations from the CNN model and the neural representations for human visual perception in the context of the perception of digits versus objects, particularly in the primary visual and associated areas. This study also illustrates the uniqueness of human visual perception. Unlike the CNN model, the neural representation of digits and objects for humans is more widely distributed across the whole brain, including the frontal and temporal areas.
We investigated neural representations for visual perception of 10 handwritten digits and six natural objects using representational similarity analysis between features from a convolutional neural network (CNN) and neuronal activations from functional magnetic resonance imaging. There was a surprising similarity between the neural representations from the CNN model and neural representations for human visual perception, particularly in the primary visual and associated areas. Moreover, unlike CNN, the neural representation of digits and objects for humans is more widely distributed across the whole brain, including the frontal and temporal areas.
Functional connectivity (FC) patterns obtained from resting-state functional magnetic resonance imaging data are commonly employed to study neuropsychiatric conditions by using pattern classifiers ...such as the support vector machine (SVM). Meanwhile, a deep neural network (DNN) with multiple hidden layers has shown its ability to systematically extract lower-to-higher level information of image and speech data from lower-to-higher hidden layers, markedly enhancing classification accuracy. The objective of this study was to adopt the DNN for whole-brain resting-state FC pattern classification of schizophrenia (SZ) patients vs. healthy controls (HCs) and identification of aberrant FC patterns associated with SZ. We hypothesized that the lower-to-higher level features learned via the DNN would significantly enhance the classification accuracy, and proposed an adaptive learning algorithm to explicitly control the weight sparsity in each hidden layer via L1-norm regularization. Furthermore, the weights were initialized via stacked autoencoder based pre-training to further improve the classification performance. Classification accuracy was systematically evaluated as a function of (1) the number of hidden layers/nodes, (2) the use of L1-norm regularization, (3) the use of the pre-training, (4) the use of framewise displacement (FD) removal, and (5) the use of anatomical/functional parcellation. Using FC patterns from anatomically parcellated regions without FD removal, an error rate of 14.2% was achieved by employing three hidden layers and 50 hidden nodes with both L1-norm regularization and pre-training, which was substantially lower than the error rate from the SVM (22.3%). Moreover, the trained DNN weights (i.e., the learned features) were found to represent the hierarchical organization of aberrant FC patterns in SZ compared with HC. Specifically, pairs of nodes extracted from the lower hidden layer represented sparse FC patterns implicated in SZ, which was quantified by using kurtosis/modularity measures and features from the higher hidden layer showed holistic/global FC patterns differentiating SZ from HC. Our proposed schemes and reported findings attained by using the DNN classifier and whole-brain FC data suggest that such approaches show improved ability to learn hidden patterns in brain imaging data, which may be useful for developing diagnostic tools for SZ and other neuropsychiatric disorders and identifying associated aberrant FC patterns.
•Deep neural network (DNN) improves schizophrenia (SZ) classification performance.•Sparsity control of weights via L1-norm regularization increases performance.•Stacked autoencoder pre-training of DNN weights further enhances performance.•Lower/higher level features can be extracted from the DNN.•Lower/higher level features show aberrant FC in the pairs-of-nodes/networks in SZ.
Organic photovoltaics are an important part of a next‐generation energy‐harvesting technology that uses a practically infinite pollutant‐free energy source. They have the advantages of light weight, ...solution processability, cheap materials, low production cost, and deformability. However, to date, the moderate photovoltaic efficiencies and poor stabilities of organic photovoltaics impede their use as replacements for inorganic photovoltaics. Recent developments in bulk‐heterojunction organic photovoltaics mean that they have almost reached the lower efficiency limit for feasible commercialization. In this review article, the recent understanding of the ideal bulk‐heterojunction morphology of the photoactive layer for efficient exciton dissociation and charge transport is described, and recent attempts as well as early‐stage trials to realize this ideal morphology are discussed systematically from a morphological viewpoint. The various approaches to optimizing morphologies consisting of an interpenetrating bicontinuous network with appropriate domain sizes and mixed regions are categorized, and in each category, the recent trends in the morphology control on the multilength scale are highlighted and discussed in detail. This review article concludes by identifying the remaining challenges for the control of active layer morphologies and by providing perspectives toward real application and commercialization of organic photovoltaics.
The morphology of the bulk‐heterojunction photoactive layer critically influences charge‐carrier dynamics and therefore the overall photovoltaic performance of organic photovoltaics. The morphological issues reported so far are discussed and future directions toward further optimization are suggested.
Non‐negligible idiosyncrasy due to interindividual differences is an ongoing issue in resting‐state functional MRI (rfMRI) analysis. We show that a deep neural network (DNN) can be employed for ...individual identification by learning important features from the time‐varying functional connectivity (FC) of rfMRI in the Human Connectome Project. We employed the trained DNN to identify individuals from an independent dataset acquired at our institution. The results revealed that the DNN could successfully identify 300 individuals with an error rate of 2.9% using 15 s time‐window and 870 individuals with an error rate of 6.7%. A trained DNN with nonlinear hidden layers led to the proposal of the “fingerprint of FC” (fpFC) as representative edges of individual FC. The fpFCs for individuals exhibited commonly important and individual‐specific edges across time‐window lengths (from 5 min to 15 s). Furthermore, the utility of our model for another group of subjects was validated, supporting the feasibility of our technique in the context of transfer learning. In conclusion, our study offers an insight into the discovery of the intrinsic mode of the human brain using whole‐brain resting‐state FC and DNNs.
By using deep neural networks (DNNs), reliable and robust fingerprints of resting‐state functional connectivity for individuals were found. The trained DNN showed its efficacy in the identification of individuals from an independent dataset via transfer learning.
•Sampling-based algorithms are commonly used in motion planning problems.•The RRT* algorithm incrementally builds a tree of motion to find a solution.•Taking a shortcut to the ancestry increases the ...convergence rate to the optimal.•Combination with sampling strategies further improves the performance.
The Rapidly-exploring Random Tree (RRT) algorithm is a popular algorithm in motion planning problems. The optimal RRT (RRT*) is an extended algorithm of RRT, which provides asymptotic optimality. This paper proposes Quick-RRT* (Q-RRT*), a modified RRT* algorithm that generates a better initial solution and converges to the optimal faster than RRT*. Q-RRT* enlarges the set of possible parent vertices by considering not only a set of vertices contained in a hypersphere, as in RRT*, but also their ancestry up to a user-defined parameter, thus, resulting in paths with less cost than those of RRT*. It also applies a similar technique to the rewiring procedure resulting in acceleration of the tendency that near vertices share common parents. Since the algorithm proposed in this paper is a tree extending algorithm, it can be combined with other sampling strategies and graph-pruning algorithms. The effectiveness of Q-RRT* is demonstrated by comparing the algorithm with existing algorithms through numerical simulations. It is also verified that the performance can be further enhanced when combined with other sampling strategies.
Transcranial focused ultrasound (FUS) is making progress as a new non-invasive mode of regional brain stimulation. Current evidence of FUS-mediated neurostimulation for humans has been limited to the ...observation of subjective sensory manifestations and electrophysiological responses, thus warranting the identification of stimulated brain regions. Here, we report FUS sonication of the primary visual cortex (V1) in humans, resulting in elicited activation not only from the sonicated brain area, but also from the network of regions involved in visual and higher-order cognitive processes (as revealed by simultaneous acquisition of blood-oxygenation-level-dependent functional magnetic resonance imaging). Accompanying phosphene perception was also reported. The electroencephalo graphic (EEG) responses showed distinct peaks associated with the stimulation. None of the participants showed any adverse effects from the sonication based on neuroimaging and neurological examinations. Retrospective numerical simulation of the acoustic profile showed the presence of individual variability in terms of the location and intensity of the acoustic focus. With exquisite spatial selectivity and capability for depth penetration, FUS may confer a unique utility in providing non-invasive stimulation of region-specific brain circuits for neuroscientific and therapeutic applications.
Coronavirus disease 2019 (COVID-19) is a newly emerged human infectious disease caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). In a global pandemic, development of a cheap, ...rapid, accurate, and easy-to-use diagnostic test is necessary if we are to mount an immediate response to this emerging threat. Here, we report the development of a specific lateral flow immunoassay (LFIA)-based biosensor for COVID-19. We used phage display technology to generate four SARS-CoV-2 nucleocapsid protein (NP)-specific single-chain variable fragment-crystallizable fragment (scFv-Fc) fusion antibodies. The scFv-Fc antibodies bind specifically and with high affinity to the SARS-CoV-2 NP antigen, but not to NPs of other coronaviruses. Using these scFv-Fc antibodies, we screened three diagnostic antibody pairs for use on a cellulose nanobead (CNB)-based LFIA platform. The detection limits of the best scFv-Fc antibody pair, 12H1 as the capture probe and 12H8 as the CNB-conjugated detection probe, were 2 ng antigen protein and 2.5 × 104 pfu cultured virus. This LFIA platform detected only SARS-CoV-2 NP, not NPs from MERS-CoV, SARS-CoV, or influenza H1N1. Thus, we have successfully developed a SARS-CoV-2 NP-specific rapid diagnostic test, which is expected to be a simple and rapid diagnostic test for COVID-19.
•SARS-CoV-2 NP-specific scFv-Fc antibodies were generated.•Antibody pairs for the specific and sensitive detection of SARS-CoV-2 were found.•Lateral flow immunoassay-based rapid diagnostic tests sensitively detected NP antigen and SARS-CoV-2 virus.•The COVID-19 biosensor showed no cross-reactivity with other coronaviruses, SARS-CoV and MERS-CoV.
Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) causes coronavirus disease 2019 (COVID-19), a newly emerging human infectious disease. Because no specific antiviral drugs or vaccines are ...available to treat COVID-19, early diagnostics, isolation, and prevention are crucial for containing the outbreak. Molecular diagnostics using reverse transcription polymerase chain reaction (RT-PCR) are the current gold standard for detection. However, viral RNAs are much less stable during transport and storage than proteins such as antigens and antibodies. Consequently, false-negative RT-PCR results can occur due to inadequate collection of clinical specimens or poor handling of a specimen during testing. Although antigen immunoassays are stable diagnostics for detection of past infection, infection progress, and transmission dynamics, no matched antibody pair for immunoassay of SARS-CoV-2 antigens has yet been reported. In this study, we designed and developed a novel rapid detection method for SARS-CoV-2 spike 1 (S1) protein using the SARS-CoV-2 receptor ACE2, which can form matched pairs with commercially available antibodies. ACE2 and S1-mAb were paired with each other for capture and detection in a lateral flow immunoassay (LFIA) that did not cross-react with SARS-CoV Spike 1 or MERS-CoV Spike 1 protein. The SARS-CoV-2 S1 (<5 ng of recombinant proteins/reaction) was detected by the ACE2-based LFIA. The limit of detection of our ACE2-LFIA was 1.86 × 105 copies/mL in the clinical specimen of COVID-19 Patients without no cross-reactivity for nasal swabs from healthy subjects. This is the first study to detect SARS-CoV-2 S1 antigen using an LFIA with matched pair consisting of ACE2 and antibody. Our findings will be helpful to detect the S1 antigen of SARS-CoV-2 from COVID-19 patients.
•The human ACE2 and commercial antibody were paired with each other as capture and detection probes in a lateral flow immunoassay that was not cross-reactive with SARS-CoV S1 and MERS-CoV S1 proteins.•The limit of detection of our ACE2-LFIA was 1.86 × 105 copies/mL in the clinical specimen of COVID-19 patients without no cross-reactivity for nasal swabs from healthy subjects.•This is the first study to detect SARS-CoV-2 S1 antigen using an LFIA with matched pair consisting of ACE2 and antibody.
PDF
