Congestive heart failure is a progressive and complex syndrome resulted from ventricular dysfunction which is difficult to detect at early stages. Heart rate variability has been identified as a prognostic indicator for congestive heart failure. The traditional diagnosis methods based on analyzing the electrocardiogram is time-consuming and laborious, and the interpretation of the results is subjective. Inspired by the outstanding performance of U-shaped networks in medical image segmentation, in this paper, we propose a novel end-to-end classification model based on 2000 intervals between successive R-peaks of electrocardiogram signals. The proposed model integrates the outputs of encoders, decoders and intermediate units through a unified scale operation, which can not only preserve low-level details from the input signals but also extract the high-level pathology-related information. We further employ a variant of residual module with group convolution and Squeeze-and-Excitation block, enhancing the network’s expression capability. In addition, considering the challenge of collecting large and diverse samples by individual institutions, we decentralize the data across different clients and extend the proposed model with an federated version which is able to facilitate multi-institutional collaborations while maintaining data anonymity. A total of 29 CHF patients and 177 non-CHF subjects (i.e., 54 normal sinus rhythm subjects, 84 atrial fibrillation and 39 Apnea subjects) from PhysioBank are included in this paper. The experimental results show that the proposed model outperforms the state of the art both in centralized and decentralized learning, with an accuracy of 89.83% and 87.54% respectively. The diagnosis model trained in federated framework provides competitive performance to that in centralized learning, which demonstrates its potential of utilizing multi-site data to improve congestive heart failure detection performance without sharing patient privacy.

Effective and rapid assessment of pork freshness is significant for monitoring pork quality. However, a traditional sensory evaluation method is subjective and physicochemical analysis is time-consuming. In this study, the near-infrared spectroscopy (NIRS) technique, a fast and non-destructive analysis method, is employed to determine pork freshness. Considering that commonly used statistical modeling methods require preprocessing data for satisfactory performance, this paper presents a one-dimensional squeeze-and-excitation residual network (1D-SE-ResNet) to construct the complex relationship between pork freshness and NIRS. The developed model enhances the one-dimensional residual network (1D-ResNet) with squeeze-and-excitation (SE) blocks. As a deep learning model, the proposed method is capable of extracting features from the input spectra automatically and can be used as an end-to-end model to simplify the modeling process. A comparison between the proposed method and five popular classification models indicates that the 1D-SE-ResNet achieves the best performance, with a classification accuracy of 93.72%. The research demonstrates that the NIRS analysis technique based on deep learning provides a promising tool for pork freshness detection and therefore is helpful for ensuring food safety.

An intelligent analytical technique which is able to accurately identify maceral components is highly desired in the fields of mining and geology. However, currently available methods based on fixed-size window neglect the shape information, and thus do not work in identifying maceral composition from one entire photomicrograph. To address these concerns, we propose a novel Maceral Identification strategy based on image Segmentation and Classification (MISC). Considering the complex and heterogeneous nature of coal, a two-level coarse-to-fine clustering method based on K-means is employed to divide microscopic images into a sequence of regions with similar attributes (i.e., binder, vitrinite, liptinite and inertinite). Furthermore, comprehensive features along with random forest are utilized to automatically classify binder and seven types of maceral components, including vitrinite, fusinite, semifusinite, cutinite, sporinite, inertodetrinite and micrinite. Evaluations on 39 microscopic images show that the proposed method achieves the state-of-the-art accuracy of 90.44% and serves as the baseline for future research on maceral analysis. In addition, to support the decisions of petrologists during maceral analysis, we developed a standalone software, which is freely available at https:/github.com/GuyooGu/MISC-Master.

Congestive heart failure (CHF), a progressive and complex syndrome caused by ventricular dysfunction, is difficult to detect at an early stage. Heart rate variability (HRV) was proposed as a prognostic indicator for CHF. Inspired by the success of 2-D UNet++ in medical image segmentation, in this paper, we introduce an end-to-end encoder-decoder model to detect CHF using HRV signals. The developed model enhances the UNet++ model with Squeeze-and-Excitation (SE) residual blocks to extract deep features hierarchically and distinguish CHF patients from normal subjects. Two open-source databases are utilized for evaluating the proposed method, and three segment lengths of intervals between successive R-peaks are employed in comparison with state-of-the-art methods. The proposed method achieves an accuracy of 85.64%, 86.65% and 88.79% when 500, 1000 and 2000 RR intervals are utilized, respectively. It demonstrates that HRV evaluation based on deep learning can be an important tool for early detection of CHF, and may assist clinicians in achieving timely and accurate diagnoses.

Proximate analysis of coal indicates the moisture, ash, volatile content, and calorific value, which has been widely utilized as the basis for coal characterization. It involves heating the coal under various conditions until a constant weight is obtained. Although it is a relatively simple process that does not require expensive analytical equipment, determining these characteristics is time-consuming. An alternative way for proximate analysis is spectral analysis in combination with various machine learning methods. However, most previous works analyze individual characteristics and fail to explore the relationship among them. In this study, we propose a method for proximate analysis based on near-infrared spectroscopy and a multi-output attention Unet (MOA-Unet), which can predict multiple characteristics simultaneously. First, an attention-based Unet is designed as the shared feature extraction sub-network, including an encoder, a decoder, convolutional block attention modules and multi-scale feature fusion modules, which can improve the representation power of the U-shape network through aggregating features of shallower layers and concatenating features of deeper layers. Second, four individual sub-networks with fully connected layers, designed for four outputs, are utilized for regressing those four characteristics. We employ the gradient normalization algorithm to alleviate the gradient magnitude masking effect caused by training imbalance among different task. The proposed MOA-Unet is compared with classical chemometric methods on 670 coal samples from on-site test, including support vector regression, partial least squares regression, random forest, extreme gradient boosting and a few deep learning methods. The experimental results demonstrate that the proposed model achieves state-of-the-art performance with correlation coefficients of 0.9015, 0.9538, 0.8986, and 0.8884, corresponding to moisture, ash, volatile content, and calorific value, respectively.