Search

Background Ensuring high accuracy in multimodal image fusion for oral and maxillofacial tumors is crucial before further application. The aim of this study was to explore the factors influencing the accuracy of multimodal image fusion for oral and maxillofacial tumors. Methods Pairs of single-modality images were obtained from oral and maxillofacial tumor patients, and were fused using a proprietary navigation system by using three algorithms (automatic fusion, manual fusion, and registration point-based fusion). Fusion accuracy was evaluated including two aspects—overall fusion accuracy and tumor volume fusion accuracy—and were indicated by mean deviation and fusion index, respectively. Image modality, fusion algorithm, and other characteristics of multimodal images that may have potential influence on fusion accuracy were recorded. Univariate and multivariate analysis were used to identify relevant affecting factors. Results Ninety-three multimodal images were generated by fusing 31 pairs of single-modality images. The interaction effect of image modality and fusion algorithm (P = 0.02, P = 0.003) and thinner slice thickness (P = 0.006) were shown to significantly influence the overall fusion accuracy. The tumor volume (P P = 0.007), and image modality (P = 0.01) were significant influencing factors for tumor volume fusion accuracy. Conclusions To ensure high overall fusion accuracy, manual fusion was not preferred in CT/MRI image fusion, and neither was automatic fusion in image fusion containing PET modality. Using image sets with thinner slice thickness could increase overall fusion accuracy. CT/MRI fusion yielded higher tumor volume fusion accuracy than fusion containing PET modality. The tumor volume fusion accuracy should be taken into consideration during image fusion when the tumor volume is small and the tumor is located in the mandible.

Mixture cure models have been developed as an effective tool to analyze failure time data with a cure fraction. Used in conjunction with the logistic regression model, this model allows covariate-adjusted inference of an exposure effect on the cured probability and on the hazard of failure for the uncured subjects. However, the covariate-adjusted inference for the overall exposure effect is not directly provided. In this paper, we describe a Cox proportional hazards cure model to analyze interval-censored survival data in the presence of a cured fraction and then apply a post-estimation approach by using model-predicted estimates difference to assess the overall exposure effect on the restricted mean survival time scale. For baseline hazard/survival function estimation, simple parametric models as fractional polynomials or restricted cubic splines are utilized to approximate the baseline logarithm cumulative hazard function, or alternatively, the full likelihood is specified through a piecewise linear approximation for the cumulative baseline hazard function. Simulation studies were conducted to demonstrate the unbiasedness of both estimation methods for the overall exposure effect estimates over various baseline hazard distribution shapes. The methods are applied to analyze the interval-censored relapse time data from a smoking cessation study.

After wars, some unexploded bombs remained underground, and these faulty bombs seriously threaten the safety of people. The ability to accurately identify targets is crucial for subsequent mining work. A deep learning algorithm is used to recognize targets, which significantly improves recognition accuracy compared with the traditional recognition algorithm for measuring the magnetic moment of the target and the included geomagnetism angle. In this paper, a ResNet-18-based recognition system is presented for classifying metallic object types. First, a fluxgate magnetometer cube arrangement structure (FMCAS) magnetic field feature collector is constructed, utilizing an eight-fluxgate magnetometer sensor array structure that provides a 400 mm separation between each sensitive unit. Magnetic field data are acquired, along an east–west survey line on the northern side of the measured target using the FMCAS. Next, the location and type of targets are modified to create a database of magnetic target models, increasing the diversity of the training dataset. The experimental dataset is constructed by constructing the magnetic flux density tensor matrix. Finally, the enhanced ResNet-18 is used to train the data for the classification recognition recognizer. According to the test findings of 107 validation set groups, this method’s recognition accuracy is 84.1 percent. With a recognition accuracy rate of 96.3 percent, a recall rate of 96.4 percent, and a precision rate of 96.4 percent, the target with the largest magnetic moment has the best recognition impact. Experimental findings demonstrate that our enhanced RestNet-18 network can efficiently classify metallic items. This provides a new idea for underground metal target identification and classification.

Path planning for wheeled mobile robots on partially known uneven terrain is an open challenge since robot motions can be strongly influenced by terrain with incomplete environmental information such as locally detected obstacles and impassable terrain areas. This paper proposes a hierarchical path planning approach for a wheeled robot to move in a partially known uneven terrain. We first model the partially known uneven terrain environment respecting the terrain features, including the slope, step, and unevenness. Second, facilitated by the terrain model, we use A⋆ algorithm to plan a global path for the robot based on the partially known map. Finally, the Q-learning method is employed for local path planning to avoid locally detected obstacles in close range as well as impassable terrain areas when the robot tracks the global path. The simulation and experimental results show that the designed path planning approach provides satisfying paths that avoid locally detected obstacles and impassable areas in a partially known uneven terrain compared with the classical A⋆ algorithm and the artificial potential field method.

The application of artificial intelligence has realized the transformation of people's production and lifestyle, and also promoted the progress of physical education and comprehensive health quality. The application of artificial intelligence in the current physical education movement is increasing. By utilizing its advanced method of virtual simulation technology, the purpose of this paper is to realize the interventional research on the physical education movement and comprehensive health quality in the environment of artificial intelligence. This paper proposes to use the virtual simulation technology and Kinect algorithm in artificial intelligence to design the virtual sports simulation teaching mode. The functional module design part where the Kinect algorithm helps the teaching of virtual sports simulation experiments, which is helpful to analyze and solve the objective system imbalance and ecological imbalance in online physical education teaching. By using the principles and rules of the Mean Shift image segmentation algorithm for reference, the investigation and research on the comprehensive health quality of students are carried out, so as to realize the ecologicalization of the virtual sports school. In the investigation and research on the comprehensive quality of students, the results show that the overall quality of these students who has reached the level of qualified or unqualified is accounting for about 30% of the total number. It is worth noting that in terms of scientific and cultural quality, only 43.34% of all students have excellent grades. It can be seen that the important training goal of current school research is how to use reasonable and effective methods and strategies to improve students' scientific and cultural level, and improve students' other comprehensive scores at the same time.

Atrial fibrillation (AF) is a typical category of arrhythmia. Clinical diagnosis of AF is based on the detection of abnormal R-R intervals (RRIs) with an electrocardiogram (ECG). Previous studies considered this detection problem as a classification problem and focused on extracting a number of features. In this study we demonstrate that instead of using any specific numerical characteristic as the input feature, the probability density of RRIs from ECG conserves comprehensive statistical information; hence, is a natural and efficient input feature for AF detection. Incorporated with a support vector machine as the classifier, results on the MIT-BIH database indicates that the proposed method is a simple and accurate approach for AF detection in terms of accuracy, sensitivity, and specificity.

Background Gene fusion events are significant sources of somatic variation across adult and pediatric cancers and are some of the most clinically-effective therapeutic targets, yet low consensus of RNA-Seq fusion prediction algorithms makes therapeutic prioritization difficult. In addition, events such as polymerase read-throughs, mis-mapping due to gene homology, and fusions occurring in healthy normal tissue require informed filtering, making it difficult for researchers and clinicians to rapidly discern gene fusions that might be true underlying oncogenic drivers of a tumor and in some cases, appropriate targets for therapy. Results We developed annoFuse, an R package, and shinyFuse, a companion web application, to annotate, prioritize, and explore biologically-relevant expressed gene fusions, downstream of fusion calling. We validated annoFuse using a random cohort of TCGA RNA-Seq samples (N = 160) and achieved a 96% sensitivity for retention of high-confidence fusions (N = 603). annoFuse uses FusionAnnotator annotations to filter non-oncogenic and/or artifactual fusions. Then, fusions are prioritized if previously reported in TCGA and/or fusions containing gene partners that are known oncogenes, tumor suppressor genes, COSMIC genes, and/or transcription factors. We applied annoFuse to fusion calls from pediatric brain tumor RNA-Seq samples (N = 1028) provided as part of the Open Pediatric Brain Tumor Atlas (OpenPBTA) Project to determine recurrent fusions and recurrently-fused genes within different brain tumor histologies. annoFuse annotates protein domains using the PFAM database, assesses reciprocality, and annotates gene partners for kinase domain retention. As a standard function, reportFuse enables generation of a reproducible R Markdown report to summarize filtered fusions, visualize breakpoints and protein domains by transcript, and plot recurrent fusions within cohorts. Finally, we created shinyFuse for algorithm-agnostic interactive exploration and plotting of gene fusions. Conclusions annoFuse provides standardized filtering and annotation for gene fusion calls from STAR-Fusion and Arriba by merging, filtering, and prioritizing putative oncogenic fusions across large cancer datasets, as demonstrated here with data from the OpenPBTA project. We are expanding the package to be widely-applicable to other fusion algorithms and expect annoFuse to provide researchers a method for rapidly evaluating, prioritizing, and translating fusion findings in patient tumors.

To address the problem of microstructural analysis of titania nanoparticles with high cytotoxicity, the authors propose X-ray phase-comparative CT imaging studies. In this method, the HE-stained section samples were compared with the X-ray phase-contrast CT imaging microscopic images, and 3D texture analysis was used to observe the changes in the preparation of hepatocyte microstructures in the two groups. The results show that X-ray phase-contrast CT imaging microscopic images and their larger image size are closely related to HE staining images, and X-ray phase-contrast CT microscopic images can observe important data of hepatocytes from multiple angles. The ship skeleton extraction method based on the endpoint limit also has advantages over traditional algorithms in extraction accuracy and can provide more 3D feature files, confirming the growth and transformation of normal hepatocytes into hepatocyte cytotoxic microstructures. The distribution effect of using the ensemble process is better than the simple 2D feature set and 3D feature set, and the overall accuracy is improved; the result distribution of the tree determination and random forest methods is also better than that of the support vector machine method. The experimental results show that the X-ray phase-contrast CT images can highlight the 2D and 3D imaging features of the hepatotoxic microstructure of TiO2 nanoparticles and provide data for quantitative analysis.

The timely detection of equipment failure can effectively avoid industrial safety accidents. The existing equipment fault diagnosis methods based on single-mode signal not only have low accuracy, but also have the inherent risk of being misled by signal noise. In this paper, we reveal the possibility of using multi-modal monitoring data to improve the accuracy of equipment fault prediction. The main challenge of multi-modal data fusion is how to effectively fuse multi-modal data to improve the accuracy of fault prediction. We propose a multi-modal learning framework for fusion of low-quality monitoring data and high-quality monitoring data. In essence, low-quality monitoring data are used as a compensation for high-quality monitoring data. Firstly, the low-quality monitoring data is optimized, and then the features are extracted. At the same time, the high-quality monitoring data is dealt with by a low complexity convolutional neural network. Moreover, the robustness of the multi-modal learning algorithm is guaranteed by adding noise to the high-quality monitoring data. Finally, different dimensional features are projected into a common space to obtain accurate fault sample classification. Experimental results and performance analysis confirm the superiority of the proposed algorithm. Compared with the traditional feature concatenation method, the prediction accuracy of the proposed multi-modal learning algorithm can be improved by up to 7.42%.

Immune-related long noncoding RNAs (irlncRNAs) are actively involved in regulating the immune status. This study aimed to establish a risk model of irlncRNAs and further investigate the roles of irlncRNAs in predicting prognosis and the immune landscape in pancreatic cancer. The transcriptome profiles and clinical information of 176 pancreatic cancer patients were retrieved from The Cancer Genome Atlas (TCGA). Immune-related genes (irgenes) downloaded from ImmPort were used to screen 1903 immune-related lncRNAs (irlncRNAs) using Pearson’s correlation analysis (R > 0.5; p, Graphical abstract

Due to the complexity of myocardial infarction (MI) waveform, most traditional automatic diagnosis models rarely detect it, while those able to detect MI often require high computing and storage capacity, rendering them unsuitable for portable devices. Therefore, in order for convenient real-time MI detection, it is essential to design lightweight models suitable for resource-limited portable devices. This paper proposes a novel multi-channel lightweight model (ML-Net), that provides a new solution for portable detection devices with limited resources. In ML-Net, each electrocardiogram (ECG) lead is assigned an independent channel, ensuring data independence and preserve the ECG characteristics of different angles represented by different leads. Moreover, convolution kernels of heterogeneous sizes are utilized to achieve accurate classification with only a small amount of lead data. Extensive experiments over actual ECG data from the PTB diagnostic database are conducted to evaluate ML-Net. The results show that ML-Net outperforms comparable schemes in diagnosing MI, and it requires lower computational cost and less memory, so that portable devices can be more widely used in the field of Internet of Medical Things(IoMT).

Rotational malalignment between the femoral and tibial components in total knee arthroplasty (TKA) can affect clinical outcomes, but there is no consensus on how to best determine tibia tray orientation. The posterolateral corner-locked (PLCL) technique may be a new method. This study aims to assess the applicability of this technique in a Chinese population. Forty normal Chinese volunteers were recruited and underwent computed tomography (CT) of the lower limbs. Knee model reconstructions and simulated standard tibial osteotomy were conducted digitally. The transepicondylar axis (TEA), the Akagi line, and the line connecting the medial third of the tibial tubercle with the midpoint of the posterior cruciate ligament (PCL) were projected to the tibial cross-section and marked. The PLCL technique was applied using either symmetrical or asymmetrical tibial tray templates, and the anteroposterior (AP) axis of the tibial tray was marked. The angles between the TEA and these lines were calculated, and the statistical differences were analyzed. The angle between the TEA and the Akagi line and between the TEA and the line connecting the medial third of the tibial tubercle with the midpoint of the PCL were 96.90 ± 5.57 and 107.31 ± 5.95 degrees, respectively. The angles between the TEA and the AP axis of the symmetrical and the asymmetrical design tibial trays were 94.01 ± 4.21 and 96.65 ± 4.70 degrees, respectively. Except for the Akagi line and AP axis of the asymmetrical tibial tray, statistical differences were found between all lines (p

Visual question answering (VQA) has gained increasing attention in both natural language processing and computer vision. The attention mechanism plays a crucial role in relating the question to meaningful image regions for answer inference. However, most existing VQA methods: 1) learn the attention distribution either from free-form regions or detection boxes in the image, which is intractable in answering questions about the foreground object and background form, respectively and 2) neglect the prior knowledge of human attention and learn the attention distribution with an unguided strategy. To fully exploit the advantages of attention, the learned attention distribution should focus more on the question-related image regions, such as human attention for both the questions, about the foreground object and background form. To achieve this, this article proposes a novel VQA model, called adversarial learning of supervised attentions (ALSAs). Specifically, two supervised attention modules: 1) free form-based and 2) detection-based, are designed to exploit the prior knowledge for attention distribution learning. To effectively learn the correlations between the question and image from different views, that is, free-form regions and detection boxes, an adversarial learning mechanism is implemented as an interplay between two supervised attention modules. The adversarial learning reinforces the two attention modules mutually to make the learned multiview features more effective for answer inference. The experiments performed on three commonly used VQA datasets confirm the favorable performance of ALSA.

In deep learning tasks, the update step size determined by the learning rate at each iteration plays a critical role in gradient-based optimization. However, determining the appropriate learning rate in practice typically relies on subjective judgment. In this work, we propose a novel optimization method based on local quadratic approximation (LQA). In each update step, we locally approximate the loss function along the gradient direction by using a standard quadratic function of the learning rate. Subsequently, we propose an approximation step to obtain a nearly optimal learning rate in a computationally efficient manner. The proposed LQA method has three important features. First, the learning rate is automatically determined in each update step. Second, it is dynamically adjusted according to the current loss function value and parameter estimates. Third, with the gradient direction fixed, the proposed method attains a nearly maximum reduction in the loss function. Extensive experiments were conducted to prove the effectiveness of the proposed LQA method.

At the beginning of 2020, the COVID-19 pandemic was able to spread quickly in Wuhan and in the province of Hubei due to a lack of experience with this novel virus. Additionally, authories had no proven experience with applying insufficient medical, communication and crisis management tools. For a considerable period of time, the actual number of people infected was unknown. There were great uncertainties regarding the dynamics and spread of the Covid-19 virus infection. In this paper, we develop a system dynamics model for the three connected regions (Wuhan, Hubei excl. Wuhan, China excl. Hubei) to understand the infection and spread dynamics of the virus and provide a more accurate estimate of the number of infected people in Wuhan and discuss the necessity and effectivity of protective measures against this epidemic, such as the quarantines imposed throughout China. We use the statistics of confirmed cases of China excl. Hubei. Also the daily data on travel activity within China was utilized, in order to determine the actual numerical development of the infected people in Wuhan City and Hubei Province. We used a multivariate Monte Carlo optimization to parameterize the model to match the official statistics. In particular, we used the model to calculate the infections, which had already broken out, but were not diagnosed for various reasons. publishedVersion

Deformities of the maxillofacial region following trauma and ablative surgery are devastating and not uncommon. Reconstruction of such defects is a surgically challenging procedure. Conventionally, reconstruction of dental arch defects lacks preoperative customised planning and relies heavily on the surgeon experience to ensure optimum surgical outcomes. The restoration of the dental arch shape and function has taken precedence after an extensive tumour resection surgery, especially in the current age of technological advancement. Thus, personalised and accurate reconstruction of dental arch defects has become a new goal. Computer-assisted surgery, especially navigation-assisted surgery, has gained popularity of late, in reconstructing deformities and restoring facial symmetry, appearance and function in the maxillofacial region. This technology provides a clearer three-dimensional visualisation of the area of interest and its relationship with the adjacent vital structures. Together with preoperative virtual surgical planning, it allows more specific and accurate osteotomies, thus reducing the ischemia and total operating times substantially. The risk of complications is also minimised whilst improving the final surgical outcomes. The use of the intraoperative navigation system and other computer-assisted surgical techniques during surgery can significantly improve the precision of the reconstruction of dental arch deformities, and achieve personalised and functional reconstructive goals while enhancing the quality of life of patients postoperatively. The Society of Oral and Maxillofacial Surgery, Chinese Stomatological Association provides the present professional perspective and treatment protocol for navigation-guided reconstruction of dental arch defects, to allow standardisation of the technique while promoting its application among oral and maxillofacial surgeons.

Objective To develop a revised evaluation method for accuracy of multimodal image fusion for oral and maxillofacial tumors and explore its application for comparing the accuracy of three commonly used fusion algorithms, automatic fusion, manual fusion, and registration point–based fusion. Materials and methods Image sets of patients with oral and maxillofacial tumor were fused using the iPlan 3.0 navigation system. Fusion accuracy included two aspects: (1) overall fusion accuracy: represented by the mean value of the coordinate differences along the x-, y-, and z- axes (Δx, Δy, and Δz), mean deviation (MD), and root mean square (RMS) of six pairs of landmarks on the two image sets; (2) tumor volume fusion accuracy: represented by Fusion Index (FI), which was calculated based on the volume of tumor delineated on the two image sets. Results Eighteen pairs of image sets of 17 patients were enrolled in this study. The Δx and Δy values for the three algorithms were less than 1.5 mm. The Δz values for automatic fusion, manual fusion and registration point–based fusion was 1.049 mm, 1.864 mm and 1.254 mm. The MD for automatic fusion, manual fusion and registration point–based fusion was 1.978 mm, 2.788 mm and 1.926 mm. Significant differences existed in Δz for manual fusion and that for automatic fusion (P = 0.058), in MD for manual fusion and that for automatic fusion (P = 0.087), and in MD for manual fusion and that for registration point–based fusion (P = 0.069). The FI for automatic fusion, manual fusion, and registration point–based fusion was 0.594, 0.520, and 0.549; the inter-algorithm differences were not significant (P = 0.290). Conclusion The automatic fusion and the registration point–based fusion were more accurate than manual fusion, and therefore were recommended to be used in multimodal image fusion for oral and maxillofacial tumors.

Drosophila melanogaster is a well-established model organism that is widely used in genetic studies. This species enjoys the availability of a wide range of research tools, well-annotated reference databases and highly similar gene circuitry to other insects. To facilitate molecular mechanism studies in Drosophila, we present the Predicted Drosophila Interactome Resource (PDIR), a database of high-quality predicted functional gene interactions. These interactions were inferred from evidence in 10 public databases providing information for functional gene interactions from diverse perspectives. The current version of PDIR includes 102 835 putative functional associations with balanced sensitivity and specificity, which are expected to cover 22.56% of all Drosophila protein interactions. This set of functional interactions is a good reference for hypothesis formulation in molecular mechanism studies. At the same time, these interactions also serve as a high-quality reference interactome for gene set linkage analysis (GSLA), which is a web tool for the interpretation of the potential functional impacts of a set of changed genes observed in transcriptomics analyses. In a case study, we show that the PDIR/GSLA system was able to produce a more comprehensive and concise interpretation of the collective functional impact of multiple simultaneously changed genes compared with the widely used gene set annotation tools, including PANTHER and David. PDIR and its associated GSLA service can be accessed at http://drosophila.biomedtzc.cn.

Background and Objectives Glycemic control with unannounced meals is the major challenge for artificial pancreas. In this study, we described the performance and safety of learning-type model predictive control (L-MPC) for artificial pancreas challenged by an unannounced meal in type 1 diabetes (T1D). Methods This closed-loop (CL) system was tested in 29 T1D patients at one site in a 4 h inpatient open-label study. Participants used an L-MPC CL system for 6 days after 2-day system identification using open-loop (OL) insulin system. During the CL period, the L-MPC system was started from 8:00 am to noon each day. At 9:00 am, each participant consumed 50 g of carbohydrates with no prandial insulin bolus. At 9:30 am on CL-Day 4 or CL-Day 6, participants rode bicycles for 20 minutes or drank 50 ml of beer, in a random order. Results As the primary outcome, TIR on CL-Day 3 was 65.2±23.3%, which was 9.8 points higher (95% CI 1.8 to 17.8; P = 0.019) than that on CL-Day 1. The time of glucose >10 mmol/L was decreased by 11.0% (95% CI -18.7 to 3.3; P = 0.007), and mean glucose level was decreased by 1.1 mmol/L (95% CI -1.1 to 0.5; P = 0.000). The total daily insulin dosage showed no significant difference (-0.1U, 95% CI -1.34 to 1.32; P = 0.982). Compared with OL-Day1 with a postprandial bolus, the TIR was increased by 13.7 points (95% CI 1.4 to 26.0; P = 0.030), the time of glucose >10 mmol/L and the mean glucose level were also decreased. Compared with the exercise day (CL-Day E, 62.0 ± 23.3%; P = 0.347) or alcohol day (CL-Day A, 64.0 ± 23.6%; P = 0.756), there was no statistically significant difference in terms of TIR, time of glucose >10 mmol/L and mean glucose level. No severe hypoglycemic events occurred and hypoglycemic episodes were not increased by using closed-loop insulin system. Conclusion The L-MPC CL insulin system achieved good glycemic control challenged by an unannounced meal.

Taking the Xiushui township of Baisha county in Hainan province as the research area,the random forest algorithm with obvious advantages in feature selection and classification extraction was used to extract the information of the Callicarpa nudiflora planting in the study area. Firstly,four kinds of different characteristic variables were generated based on World View-3 data,including spectral features,principal component features,vegetation index and texture features. Secondly,the spatial distribution of the C. nudiflora in the study area was extracted by remote sensing by random forest classification algorithm. Finally,the feature space of the random forest classification algorithm was optimized based on the feature importance to obtain the best random forest classification results,and this result is compared with the classification result of the random forest algorithm of the unoptimized feature space. The results showed that:①The overall accuracy of the C. nudiflora extracted by World View-3 image was 89. 97%,and the Kappa coefficient was 0. 84,which indicates that the random forest algorithm had higher classification accuracy and better applicability in Hainan C. nudiflora recognition.② The overall accuracy of extracting C. nudiflora with the dimension reduction feature was 90. 4,and the Kappa coefficient was 0. 85,which indicates that the random forest algorithm can effectively select features. At the same time as the feature variable data mining,the precision of the information extraction of the C. nudiflora was still guaranteed,and the operation efficiency was improved. This study provides a new idea,method and technical means for information extraction of cultivated medicinal plant resources in terms of feature selection and method selection.

The fields of human activity analysis have recently begun to diversify. Many researchers have taken much interest in developing action recognition or action prediction methods. The research on human action evaluation differs by aiming to design computation models and evaluation approaches for automatically assessing the quality of human actions. This line of study has become popular because of its explosively emerging real-world applications, such as physical rehabilitation, assistive living for elderly people, skill training on self-learning platforms, and sports activity scoring. This paper presents a comprehensive survey of approaches and techniques in action evaluation research, including motion detection and preprocessing using skeleton data, handcrafted feature representation methods, and deep learning-based feature representation methods. The benchmark datasets from this research field and some evaluation criteria employed to validate the algorithms’ performance are introduced. Finally, the authors present several promising future directions for further studies.

Background Making accurate patient care decision, as early as possible, is a constant challenge, especially for physicians in the emergency department. The increasing volumes of electronic medical records (EMRs) open new horizons for automatic diagnosis. In this paper, we propose to use machine learning approaches for automatic infection detection based on EMRs. Five categories of information are utilized for prediction, including personal information, admission note, vital signs, diagnose test results and medical image diagnose. Results Experimental results on a newly constructed EMRs dataset from emergency department show that machine learning models can achieve a decent performance for infection detection with area under the receiver operator characteristic curve (AUC) of 0.88. Out of all the five types of information, admission note in text form makes the most contribution with the AUC of 0.87. Conclusions This study provides a state-of-the-art EMRs processing system to automatically make medical decisions. It extracts five types of features associated with infection and achieves a decent performance on automatic infection detection based on machine learning models.

Automatic and accurate segmentation of osteosarcoma region in CT images can help doctor make a reasonable treatment plan, thus improving cure rate. In this paper, a multiple supervised residual network (MSRN) was proposed for osteosarcoma image segmentation. Three supervised side output modules were added to the residual network. The shallow side output module could extract image shape features, such as edge features and texture features. The deep side output module could extract semantic features. The side output module could compute the loss value between output probability map and ground truth and back-propagate the loss information. Then, the parameters of residual network could be modified by gradient descent method. This could guide the multi-scale feature learning of the network. The final segmentation results were obtained by fusing the results output by the three side output modules. A total of 1900 CT images from 15 osteosarcoma patients were used to train the network and a total of 405 CT images from another 8 osteosarcoma patients were used to test the network. Results indicated that MSRN enabled a dice similarity coefficient (DSC) of 89.22%, a sensitivity of 88.74% and a F1-measure of 0.9305, which were larger than those obtained by fully convolutional network (FCN) and U-net. Thus, MSRN for osteosarcoma segmentation could give more accurate results than FCN and U-Net.

The feasibility of near-infrared spectroscopy (NIRS) combined with chemometrics for the rapid detection of the cellulose and hemicellulose contents in corn stover is discussed. Competitive adaptive reweighted sampling (CARS) and genetic simulated annealing algorithm (GSA) were combined (CARS-GSA) to select the characteristic wavelengths of cellulose and hemicellulose and to reduce the dimensionality and multicollinearity of the NIRS data. The whole spectra contained 1845 wavelength variables. After CARS-GSA optimization, the number of characteristic wavelengths of cellulose (hemicellulose) was reduced to 152 (260), accounting for 8.24% (14.09%) of all wavelengths. The coefficients of determination of the regression models for predicting the cellulose and hemicellulose contents were 0.968 and 0.996, the root mean square errors of prediction (RMSEPs) were 0.683 and 0.648, and the residual predictive deviations (RPDs) were 5.213 and 16.499, respectively. The RMSEP of the cellulose and hemicellulose regression models was 0.152 and 0.190 lower for CARS-GSA than for the full-spectrum, and the RPD was increased by 0.949 and 3.47, respectively. The results showed that the CARS-GSA model substantially reduced the number of characteristic wavelengths and significantly improved the predictive ability of the regression model.

Background and objective Automatic osteosarcoma tumor segmentation on computed tomography (CT) images is a challenging problem, as tumors have large spatial and structural variabilities. In this study, an automatic tumor segmentation method, which was based on a fully convolutional networks with multiple supervised side output layers (MSFCN), was presented. Methods Image normalization is applied as a pre-processing step for decreasing the differences among images. In the frame of the fully convolutional networks, supervised side output layers were added to three layers in order to guide the multi-scale feature learning as a contracting structure, which was then able to capture both the local and global image features. Multiple feature channels were used in the up-sampling portion to capture more context information, for the assurance of accurate segmentation of the tumor, with low contrast around the soft tissue. The results of all the side outputs were fused to determine the final boundaries of the tumors. Results A quantitative comparison of the 405 osteosarcoma manual segmentation results from the CT images showed that the average Dice similarity coefficient (DSC), average sensitivity, average Hammoude distance (HM) and F1-measure were 87.80%, 86.88%, 19.81% and 0.908, respectively. It was determined that, when compared with the other learning-based algorithms (for example, the fully convolution networks (FCN), U-Net method, and holistically-nested edge detection (HED) method), the MSFCN had the best performances in terms of DSC, sensitivity, HM and F1-measure. Conclusion The results indicated that the proposed algorithm contributed to the fast and accurate delineation of tumor boundaries, which could potentially assist doctors in making more precise treatment plans.

Three-dimension path planning of uninhabited combat aerial vehicle (UCAV) is a complicated optimal problem, which mainly focused on optimizing the flight route considering the different types of constrains under complex combating environment. A novel predator-prey pigeon-inspired optimization (PPPIO) is proposed to solve the UCAV three-dimension path planning problem in dynamic environment. Pigeon-inspired optimization (PIO) is a new bio-inspired optimization algorithm. In this algorithm, map and compass operator model and landmark operator model are used to search the best result of a function. The prey-predator concept is adopted to improve global best properties and enhance the convergence speed. The characteristics of the optimal path are presented in the form of a cost function. The comparative simulation results show that our proposed PPPIO algorithm is more efficient than the basic PIO, particle swarm optimization (PSO), and different evolution (DE) in solving UCAV three-dimensional path planning problems.

The aim of this study was to validate the feasibility of a novel structural and computational fluid dynamics-based fractional flow reserve (FFR) algorithm for coronary computed tomography angiography (CTA), using alternative boundary conditions to detect lesion-specific ischemia.A new model of computed tomographic (CT) FFR relying on boundary conditions derived from structural deformation of the coronary lumen and aorta with transluminal attenuation gradient and assumptions regarding microvascular resistance has been developed, but its accuracy has not yet been validated.A total of 338 consecutive patients with 422 vessels from 9 Chinese medical centers undergoing CTA and invasive FFR were retrospectively analyzed. CT FFR values were obtained on a novel on-site computational fluid dynamics-based CT FFR (uCT-FFR [version 1.5, United-Imaging Healthcare, Shanghai, China]). Performance characteristics of uCT-FFR and CTA in detecting lesion-specific ischemia in all lesions, intermediate lesions (luminal stenosis 30% to 70%), and "gray zone" lesions (FFR 0.75 to 0.80) were calculated with invasive FFR as the reference standard. The effect of coronary calcification on uCT-FFR measurements was also assessed.Per vessel sensitivities, specificities, and accuracies of 0.89, 0.91, and 0.91 with uCT-FFR, 0.92, 0.34, and 0.55 with CTA, and 0.94, 0.37, and 0.58 with invasive coronary angiography, respectively, were found. There was higher specificity, accuracy, and AUC for uCT-FFR compared with CTA and qualitative invasive coronary angiography in all lesions, including intermediate lesions (p 0.001 for all). No significant difference in diagnostic accuracy was observed in the "gray zone" range versus the other 2 lesion groups (FFR ≤0.75 and0.80; p = 0.397) and in patients with "gray zone" versus FFR ≤0.75 (p = 0.633) and versus FFR0.80 (p = 0.364), respectively. No significant difference in the diagnostic performance of uCT-FFR was found between patients with calcium scores ≥400 and 400 (p = 0.393).This novel computational fluid dynamics-based CT FFR approach demonstrates good performance in detecting lesion-specific ischemia. Additionally, it outperforms CTA and qualitative invasive coronary angiography, most notably in intermediate lesions, and may potentially have diagnostic power in gray zone and highly calcified lesions.

Background Wearable smart watches provide large amount of real-time data on the environmental state of the users and are useful to determine risk factors for onset and progression of myopia. We aim to evaluate the efficacy of machine learning algorithm in differentiating indoor and outdoor locations as collected by use of smart watches. Methods Real time data on luminance, ultraviolet light levels and number of steps obtained with smart watches from dataset A: 12 adults from 8 scenes and manually recorded true locations. 70% of data was considered training set and support vector machine (SVM) algorithm generated using the variables to create a classification system. Data collected manually by the adults was the reference. The algorithm was used for predicting the location of the remaining 30% of dataset A. Accuracy was defined as the number of correct predictions divided by all. Similarly, data was corrected from dataset B: 172 children from 3 schools and 12 supervisors recorded true locations. Data collected by the supervisors was the reference. SVM model trained from dataset A was used to predict the location of dataset B for validation. Finally, we predicted the location of dataset B using the SVM model self-trained from dataset B. We repeated these three predictions with traditional univariate threshold segmentation method. Results In both datasets, SVM outperformed the univariate threshold segmentation method. In dataset A, the accuracy and AUC of SVM were 99.55% and 0.99 as compared to 95.11% and 0.95 with the univariate threshold segmentation (p Conclusions Machine learning algorithm allows for discrimination of outdoor versus indoor environments with high accuracy and provides an opportunity to study and determine the role of environmental risk factors in onset and progression of myopia. The accuracy of machine learning algorithm could be improved if the model is trained with the dataset itself.

Synonymous mutations do not change the encoded amino acids but may alter the structure or function of an mRNA in ways that impact gene function. Advances in next generation sequencing technologies have detected numerous synonymous mutations in the human genome. Several computational models have been proposed to predict deleterious synonymous mutations, which have greatly facilitated the development of this important field. Consequently, there is an urgent need to assess the state-of-the-art computational methods for deleterious synonymous mutation prediction to further advance the existing methodologies and to improve performance. In this regard, we systematically compared a total of 10 computational methods (including specific method for deleterious synonymous mutation and general method for single nucleotide mutation) in terms of the algorithms used, calculated features, performance evaluation and software usability. In addition, we constructed two carefully curated independent test datasets and accordingly assessed the robustness and scalability of these different computational methods for the identification of deleterious synonymous mutations. In an effort to improve predictive performance, we established an ensemble model, named Prediction of Deleterious Synonymous Mutation (PrDSM), which averages the ratings generated by the three most accurate predictors. Our benchmark tests demonstrated that the ensemble model PrDSM outperformed the reviewed tools for the prediction of deleterious synonymous mutations. Using the ensemble model, we developed an accessible online predictor, PrDSM, available at http://bioinfo.ahu.edu.cn:8080/PrDSM/. We hope that this comprehensive survey and the proposed strategy for building more accurate models can serve as a useful guide for inspiring future developments of computational methods for deleterious synonymous mutation prediction.

Near-infrared(NIR)As a fast and non-destructive testing technology, spectroscopy techniques is very suitable for pharmaceutical discrimination. Autoencoder network, as a hot research topic, has drawn widespread attention in machine learning research in recent years. Compared with traditional surface learning algorithm models, Autoencoder network has more powerful modeling capability as a typical deep networks model. Based on the unsupervised greedy layer-wise pre-training, autoencoder trains the network layer by layer while minimizing the error in reconstructing. Each layer is pre-trained with an unsupervised learning algorithm, learning a nonlinear transformation of the input of each layer which is the output of the previous layer. Pre-whitening process could get the inner structural features of the data more effectively. The supervised fine-tuning is followed with the unsupervised pre-training which sets the stage for a final training phase. The deep architecture is fine-tuned with respect to a supervised training criterion with gradient-based optimization. In this paper, firstly, the preprocessing step and pre-whitening transformation were used to treat near-infrared spectroscopy data of erythromycin ethylsuccinate, The pre-whitening transformation would reduce the correlation of the features, which gave each feature the same variance. Experimental results showed that the pre-whitening process had improved the classification accuracy of Sparse Denoising Autoencoder (SDAE) effectively. The SDAE with two hidden layers combined with pre-whitening was used to build the classification model for the identification of counterfeit pharmaceutical. The BP neural networks was compared with SVM algorithm for the classification accuracy and mean absolute difference (MAD). SDAE algorithm had higher classification accuracy than BP neural networks which had the same network structure with the SDAE networks, and SDAE algorithm also performed better than the SVM algorithm when the train datasets achieved a certain amount. As to the generalization performances, SDAE algorithm had less mean absolute difference of classification accuracy than SVM and BP Neural Networks. This result showed that SDAE algorithm could be effectively used to discriminate the counterfeit pharmaceutical.

Measuring the similarity between pairs of biological entities is important in molecular biology. The introduction of Gene Ontology (GO) provides us with a promising approach to quantifying the semantic similarity between two genes or gene products. This kind of similarity measure is closely associated with the GO terms annotated to biological entities under consideration and the structure of the GO graph. However, previous works in this field mainly focused on the upper part of the graph, and seldom concerned about the lower part. In this study, we aim to explore information from the lower part of the GO graph for better semantic similarity. We proposed a framework to quantify the similarity measure beneath a term pair, which takes into account both the information two ancestral terms share and the probability that they co-occur with their common descendants. The effectiveness of our approach was evaluated against seven typical measurements on public platform CESSM, protein–protein interaction and gene expression datasets. Experimental results consistently show that the similarity derived from the lower part contributes to better semantic similarity measure. The promising features of our approach are the following: (1) it provides a mirror model to characterize the information two ancestral terms share with respect to their common descendant; (2) it quantifies the probability that two terms co-occur with their common descendant in an efficient way; and (3) our framework can effectively capture the similarity measure beneath two terms, which can serve as an add-on to improve traditional semantic similarity measure between two GO terms. The algorithm was implemented in Matlab and is freely available from http://ejl.org.cn/bio/GOBeneath/ .

With the availability of high-throughput technologies, a huge number of biological data (e.g., somatic mutation, DNA methylation and gene expression) in multiple cancers have been generated. A major challenge is to identify functional and vital driver mutation import for the initiation and progression of cancer. In this paper, we introduce a novel method, named Co-occurring mutated metagene Genetic Algorithm (CoGA), to solve the maximum weight submatrix problem, with the aim of distinguishing mutated driver pathways in cancer. The algorithm relies on the combinatorial properties of mutations in the same pathways: high coverage and mutual exclusivity, and the possible properties of mutations in different pathways: co-occurring pattern. We carried out the experiment with glioblastoma multiform (GBM) data. The experimental results show that compared with the original model, our algorithm has more potential to identify driver pathways in cancer with biological significance.

One of the most challenging areas of sensor development for nuclear medicine is the design of proton therapy monitoring systems. Sensors are operated in a high detection rate regime in beam-on conditions. We realized a prototype of a monitoring system for proton therapy based on the technique of positron emission tomography. We used the Plug and Imaging (P&amp, I) technology in this application. This sensing system includes LYSO/silicon photomultiplier (SiPM) detection elements, fast digital multi voltage threshold (MVT) readout electronics and dedicated image reconstruction algorithms. In this paper, we show that the P&amp, I sensor system has a uniform response and is controllable in the experimental conditions of the proton therapy room. The prototype of PET monitoring device based on the P&amp, I sensor system has an intrinsic experimental spatial resolution of approximately 3 mm (FWHM), obtained operating the prototype both during the beam irradiation and right after it. The count-rate performance of the P&amp, I sensor approaches 5 Mcps and allows the collection of relevant statistics for the nuclide analysis. The measurement of both the half life and the relative abundance of the positron emitters generated in the target volume through irradiation of 10 10 protons in approximately 15 s is performed with 0.5% and 5 % accuracy, respectively.

Thyroid ultrasonography is a widely used clinical technique for nodule diagnosis in thyroid regions. However, it remains difficult to detect and recognize the nodules due to low contrast, high noise, and diverse appearance of nodules. In today's clinical practice, senior doctors could pinpoint nodules by analyzing global context features, local geometry structure, and intensity changes, which would require rich clinical experience accumulated from hundreds and thousands of nodule case studies. To alleviate doctors' tremendous labor in the diagnosis procedure, we advocate a machine learning approach to the detection and recognition tasks in this paper. In particular, we develop a multitask cascade convolution neural network (MC-CNN) framework to exploit the context information of thyroid nodules. It may be noted that our framework is built upon a large number of clinically confirmed thyroid ultrasound images with accurate and detailed ground truth labels. Other key advantages of our framework result from a multitask cascade architecture, two stages of carefully designed deep convolution networks in order to detect and recognize thyroid nodules in a pyramidal fashion, and capturing various intrinsic features in a global-to-local way. Within our framework, the potential regions of interest after initial detection are further fed to the spatial pyramid augmented CNNs to embed multiscale discriminative information for fine-grained thyroid recognition. Experimental results on 4309 clinical ultrasound images have indicated that our MC-CNN is accurate and effective for both thyroid nodules detection and recognition. For the correct diagnosis rate of malignant and benign thyroid nodules, its mean Average Precision (mAP) performance can achieve up to [Formula: see text] accuracy, which outperforms the common CNNs by [Formula: see text] on average. In addition, we conduct rigorous user studies to confirm that our MC-CNN outperforms experienced doctors, yet only consuming roughly [Formula: see text] ( 1/48) of doctors' examination time on average. Therefore, the accuracy and efficiency of our new method exhibit its great potential in clinical applications.

Understanding the role of DNA methylation often requires accurate assessment and comparison of these modifications in a genome-wide fashion. Sequencing-based DNA methylation profiling provides an unprecedented opportunity to map and compare complete DNA CpG methylomes. These include whole genome bisulfite sequencing (WGBS), Reduced-Representation Bisulfite-Sequencing (RRBS), and enrichment-based methods such as MeDIP-seq, MBD-seq, and MRE-seq. An investigator needs a method that is flexible with the quantity of input DNA, provides the appropriate balance among genomic CpG coverage, resolution, quantitative accuracy, and cost, and comes with robust bioinformatics software for analyzing the data. In this chapter, we describe four protocols that combine state-of-the-art experimental strategies with state-of-the-art computational algorithms to achieve this goal. We first introduce two experimental methods that are complementary to each other. MeDIP-seq, or methylation-dependent immunoprecipitation followed by sequencing, uses an anti-methylcytidine antibody to enrich for methylated DNA fragments, and uses massively parallel sequencing to reveal identity of enriched DNA. MRE-seq, or methylation-sensitive restriction enzyme digestion followed by sequencing, relies on a collection of restriction enzymes that recognize CpG containing sequence motifs, but only cut when the CpG is unmethylated. Digested DNA fragments enrich for unmethylated CpGs at their ends, and these CpGs are revealed by massively parallel sequencing. The two computational methods both implement advanced statistical algorithms that integrate MeDIP-seq and MRE-seq data. M&M is a statistical framework to detect differentially methylated regions between two samples. methylCRF is a machine learning framework that predicts CpG methylation levels at single CpG resolution, thus raising the resolution and coverage of MeDIP-seq and MRE-seq to a comparable level of WGBS, but only incurring a cost of less than 5% of WGBS. Together these methods form an effective, robust, and affordable platform for the investigation of genome-wide DNA methylation.

Background In a typical electrophysiological experiment, especially one that includes studying animal behavior, the data collected normally contain spikes, local field potentials, behavioral responses and other associated data. In order to obtain informative results, the data must be analyzed simultaneously with the experimental settings. However, most open-source toolboxes currently available for data analysis were developed to handle only a portion of the data and did not take into account the sorting of experimental conditions. Additionally, these toolboxes require that the input data be in a specific format, which can be inconvenient to users. Therefore, the development of a highly integrated toolbox that can process multiple types of data regardless of input data format and perform basic analysis for general electrophysiological experiments is incredibly useful. Results Here, we report the development of a Python based open-source toolbox, referred to as NeoAnalysis, to be used for quick electrophysiological data processing and analysis. The toolbox can import data from different data acquisition systems regardless of their formats and automatically combine different types of data into a single file with a standardized format. In cases where additional spike sorting is needed, NeoAnalysis provides a module to perform efficient offline sorting with a user-friendly interface. Then, NeoAnalysis can perform regular analog signal processing, spike train, and local field potentials analysis, behavioral response (e.g. saccade) detection and extraction, with several options available for data plotting and statistics. Particularly, it can automatically generate sorted results without requiring users to manually sort data beforehand. In addition, NeoAnalysis can organize all of the relevant data into an informative table on a trial-by-trial basis for data visualization. Finally, NeoAnalysis supports analysis at the population level. Conclusions With the multitude of general-purpose functions provided by NeoAnalysis, users can easily obtain publication-quality figures without writing complex codes. NeoAnalysis is a powerful and valuable toolbox for users doing electrophysiological experiments. Electronic supplementary material The online version of this article (10.1186/s12938-017-0419-7) contains supplementary material, which is available to authorized users.

Blumea balsamifera (Ai-na-xiang) is used as an important plant source of natural borneol, which is widely used in the pharmaceutical industry. The aim of this study was to establish the methods based on near infrared (NIR) spectroscopy for determining the geographical origin of B. balsamifera, as well as developing a method for the quantitative rapid analysis of the active pharmaceutical ingredients (APIs). A total of 109 samples were collected from China in 2013 and arbitrarily divided into calibration and prediction sets using the Kennard–Stone algorithm. The l-borneol and total flavone contents of the samples were measured by gas chromatography and ultraviolet-visible spectroscopy, respectively. The NIR spectra were acquired using an integrating sphere and a partial least squares (PLS) model was built using the optimum wavelength regions, which were selected using a synergy interval partial least-squares (SiPLS) algorithm. The root mean square errors of prediction of the l-borneol and total flavone models were 0.0779 and 2.2694 mg/g, with R2 of 0.9069 and 0.8013, respectively. A discriminant model to determine the geographical origin of B. balsamifera (Guizhou and Hainan) was also established using a partial least squares discriminant analysis method with an optimum pretreatment method. The prediction accuracy rate of the model was 100%. NIR spectroscopy can be used as a reliable and environmentally friendly method to determine the API and the origin of different B. balsamifera samples.

The diagnostic performance of Fibroscan might be improved when combined with other serum fibrosis related markers. Previous study has demonstrated that S100A4 expression is associated with liver fibrosis in humans with hepatitis. This study aimed to clarify diagnostic accuracy of serum S100A4 levels for significant liver fibrosis in patients with chronic hepatitis B (CHB), and develop a combined algorithm of liver stiffness measurement (LSM) and S100A4 to predict significant liver fibrosis in CHB.One hundred and seventy-five CHB patients who had performed liver biopsy were consecutively included. We evaluated serum S100A4 levels, LSM values and other clinically-approved fibrosis scores.Serum S100A4 level was higher in CHB patients with significant fibrosis, compared to those without [199.58 (33.31-1971.96) vs. 107.15 (2.10-1038.94), P0.001]. Using receiver-operating characteristic (ROC) analyses, the area under the curves (AUC), sensitivity, specificity and accuracy of S100A4 were found to be 0.749, 62.7%, 75.9% and 0.70 for significant fibrosis (≥Stage 2), respectively. Although not superior to LSM, these results were better than the fibrosis index based on the 4 factor (FIB-4) and the aspartate aminotransferase-to-platelet ratio index (APRI) for significant fibrosis detection. An algorithm consisting of S100A4 and LSM was derived. The AUC, sensitivity, specificity and accuracy of model based on serum S100A4 level and LSM were 0.866, 86.6%, 77.8% and 0.79 for significant fibrosis detection, superior to those based on LSM alone (0.834, 76.1%, 80.7% and 0.76, P=0.041).Serum S100A4 level was identified as a fibrosis marker of liver fibrosis in patients with CHB. Combining serum S100A4 with LSM improved the accuracy of transient elastography for hepatitis B significant fibrosis detection.

Conditional two-part random-effects models have been proposed for the analysis of healthcare cost panel data that contain both zero costs from the non-users of healthcare facilities and positive costs from the users. These models have been extended to accommodate more flexible data structures when using the generalized Gamma distribution to model the positive healthcare expenditures. However, a major drawback with the extended model, which is inherited from the conditional models, is that it is fairly difficult to make direct marginal inference with respect to overall healthcare costs that includes both zeros and non-zeros, or even on positive healthcare costs. In this article, we first propose two types of marginalized two-part random-effects generalized Gamma models (m2RGGMs): Type I m2RGGMs for the inference on positive healthcare costs and Type II m2RGGMs for the inference on overall healthcare costs. Then, the concepts of marginal effect and incremental effect of a covariate on overall and positive healthcare costs are introduced, and estimation of these effects is carefully discussed. Especially, we derive the variance estimates of these effects by following the delta methods and Taylor series approximations for the purpose of making marginal inference. Parameter estimates of Type I and Type II m2RGGMs are obtained through maximum likelihood estimation. An empirical analysis of longitudinal healthcare costs collected in the China Health and Nutrition Survey is conducted using the proposed methodologies.

Zao5241 is an elite soybean [Glycine max (L.) Merr.] line and backbone parent. In this study, we employed iTRAQ to analyze the proteomes and protein expression profiles of Zao5241 during leaf development. We identified 1,245 proteins in all experiments, of which only 45 had been previously annotated. Among overlapping proteins between three biological replicates, 598 proteins with 2 unique peptides identified were reliably quantified. The protein datasets were classified into 36 GO functional terms, and the photosynthesis term was most significantly enriched. A total of 113 proteins were defined as being differentially expressed during leaf development; 41 proteins were found to be differently expressed between two and four week old leaves, and 84 proteins were found to be differently expressed between two and six week old leaves, respectively. Cluster analysis of the data revealed dynamic proteomes. Proteins annotated as electron carrier activity were greatly enriched in the peak expression profiles, and photosynthesis proteins were negatively modulated along the whole time course. This dataset will serve as the foundation for a systems biology approach to understanding photosynthetic development.

DNA methylation plays key roles in diverse biological processes such as X chromosome inactivation, transposable element repression, genomic imprinting, and tissue-specific gene expression. Sequencing-based DNA methylation profiling provides an unprecedented opportunity to map and compare complete DNA methylomes. This includes one of the most widely applied technologies for measuring DNA methylation: methylated DNA immunoprecipitation followed by sequencing (MeDIP-seq), coupled with a complementary method, methylation-sensitive restriction enzyme sequencing (MRE-seq). A computational approach that integrates data from these two different but complementary assays and predicts methylation differences between samples has been unavailable. Here, we present a novel integrative statistical framework M&M (for integration of MeDIP-seq and MRE-seq) that dynamically scales, normalizes, and combines MeDIP-seq and MRE-seq data to detect differentially methylated regions. Using sample-matched whole-genome bisulfite sequencing (WGBS) as a gold standard, we demonstrate superior accuracy and reproducibility of M&M compared to existing analytical methods for MeDIP-seq data alone. M&M leverages the complementary nature of MeDIP-seq and MRE-seq data to allow rapid comparative analysis between whole methylomes at a fraction of the cost of WGBS. Comprehensive analysis of nineteen human DNA methylomes with M&M reveals distinct DNA methylation patterns among different tissue types, cell types, and individuals, potentially underscoring divergent epigenetic regulation at different scales of phenotypic diversity. We find that differential DNA methylation at enhancer elements, with concurrent changes in histone modifications and transcription factor binding, is common at the cell, tissue, and individual levels, whereas promoter methylation is more prominent in reinforcing fundamental tissue identities.

Fatigue is reported as the most common symptom by patients with multiple sclerosis (MS). The physiological and functional parameters related to fatigue in MS patients are currently not well established. A new wearable wireless body measurement system, named Fatigue Monitoring System (FAMOS), was developed to study fatigue in MS. It can continuously measure electrocardiogram, body-skin temperature, electromyogram and motions of feet. The goal of this study is to test the ability of distinguishing fatigued MS patients from healthy subjects by the use of FAMOS. This paper presents the realization of the measurement system including the design of both hardware and dedicated signal processing algorithms. Twenty-six participants including 17 MS patients with fatigue and 9 sex- and age-matched healthy controls were included in the study for continuous 24 h monitoring. The preliminary results show significant differences between fatigued MS patients and healthy controls. In conclusion, the FAMOS enables continuous data acquisition and estimation of multiple physiological and functional parameters. It provides a new, flexible and objective approach to study fatigue in MS, which can distinguish between fatigued MS patients and healthy controls. The usability and reliability of the FAMOS should however be further improved and validated through larger clinical trials.

We evaluated the application of three machine learning algorithms, including logistic regression, support vector machine and back-propagation neural network, for diagnosing congenital heart disease and colorectal cancer. By inspecting related serum tumor marker levels in colorectal cancer patients and healthy subjects, early diagnosis models for colorectal cancer were built using three machine learning algorithms to assess their corresponding diagnostic values. Except for serum alpha-fetoprotein, the levels of 11 other serum markers of patients in the colorectal cancer group were higher than those in the benign colorectal cancer group (P < 0.05). The results of logistic regression analysis indicted that individual detection of serum carcinoembryonic antigens, CA199, CA242, CA125, and CA153 and their combined detection was effective for diagnosing colorectal cancer. Combined detection had a better diagnostic effect with a sensitivity of 94.2% and specificity of 97.7%; combining serum carcinoembryonic antigens, CA199, CA242, CA125, and CA153, with the support vector machine diagnosis model and back-propagation, a neural network diagnosis model was built with diagnostic accuracies of 82 and 75%, sensitivities of 85 and 80%, and specificities of 80 and 70%, respectively. Colorectal cancer diagnosis models based on the three machine learning algorithms showed high diagnostic value and can help obtain evidence for the early diagnosis of colorectal cancer.