Objective: Rheumatoid arthritis and Ankylosing spondylitis are two common autoimmune inflammatory rheumatic diseases that negatively affect activities of daily living and can lead to structural and functional disability, reduced quality of life. Here, this study utilized Fourier transform infrared (FTIR) spectroscopy on dried serum samples and achieved early diagnosis of rheumatoid arthritis and ankylosing spondylitis based on deep learning models. Method: A total of 243 dried serum samples were collected in this study, including 81 samples each from ankylosing spondylitis, rheumatoid arthritis, and healthy controls. Three multi-scale convolutional modules with different specifications were designed based on the multi-scale convolutional neural network (MSCNN) to effectively fuse the local features to enhance the generalization ability of the model. The FTIR was then combined with the MSCNN model to achieve a non-invasive, fast, and accurate diagnosis of ankylosing spondylitis, rheumatoid arthritis, and healthy controls. Results: Spectral analysis shows that the curves and waveforms of the three spectral graphs are similar. The main differences are distributed in the spectral regions of 3300–3250 cm⁻¹, 3000–2800 cm⁻¹, 1750–1500 cm⁻¹, and 1500–1300 cm⁻¹ , which represent: Amides , fatty acids, cholesterol, proteins with a carboxyl group, amide II, free amino acids, and polysaccharides. Four classification models, namely artificial neural network (ANN), convolutional neural network (CNN), improved AlexNet model, and multi-scale convolutional neural network (MSCNN) were established. Through comparison, it was found that the diagnostic AUC value of the MSCNN model was 0.99, and the accuracy rate was as high as 0.93, which was much higher than the other three models. Conclusion: The study demonstrated the superiority of MSCNN in distinguishing ankylosing spondylitis from rheumatoid arthritis and healthy controls. FTIR may become a rapid, sensitive, and non-invasive means of diagnosing rheumatism.

This study focuses on the key structural locations to the south of the 1679 M8.0 Sanhe–Pinggu earthquake. In conjunction with prior deep seismic reflection exploration in the area, we conducted four shallow seismic investigations to the south of Sanhe–Pinggu seismic area to delineate the exact structure of identified faults and to ascertain the precise location, characteristics, and activity levels of active faults within the region. By analyzing the burial depth of the fault’s breakpoint as revealed by high-precision shallow seismic profiles, we postulate that the fault has been active since the middle and late Pleistocene epochs. In addition, we conducted a high-density borehole investigation in tandem with composite drilling profile at the corresponding sites of shallow breakpoints. Using chronological data from neighboring boreholes and accounting for the ages of samples acquired from these boreholes and staggered strata, the fault manifests as a Holocene active fault within the composite borehole–geological section. This study contradicted the previous conception that to the south of 1679 Sanhe–Pinggu seismic area contained no active faults. This new discovery not only has significant application value for evaluating the risk of large earthquakes in the southern part of the capital circle and understanding the earthquake disaster risk in Beijing but also has scientific significance for studying the development and evolution of faults and their deep–shallow coupling characteristics in North China since the late Cenozoic.

The ground-motion simulation of regional-specific earthquake scenarios is crucial for the seismic design of key facilities. Herein, we considered parameter uncertainty in ground-motion simulations and the impact of multiple faults when determining the maximum credible ground-motion parameters of large earthquakes at a near-field dam. The source models of the Daju–Lijiang, Xiaozhongdian–Daju, and Longpan–Qiaohou faults were established based on geological and geophysical data. Although the method for identifying asperity is not yet mature and still faces many difficulties, it provides an opportunity to identify the non-uniform slip distribution on the rupture plane by earthquake scenarios. A multi-scheme stochastic finite-fault simulation method was then used to estimate the minimum; mean; maximum; and 50th-, 84th-, and 95th-percentile values of the peak ground acceleration and pseudo-spectral acceleration response spectra. The results showed that the Longpan–Qiaohou fault can generate the largest ground-motion parameters compared with the other two faults. Moreover, this result was supported by the statistical analysis of the results of twelve thousand simulations of these three faults. Thus, it can be concluded that the maximum credible ground-motion parameters are represented by the 84th-percentile pseudo-spectral acceleration response spectrum of the Longpan–Qiaohou fault. This finding will benefit the seismic safety design of the target dam. More importantly, this multi-scheme method can be applied to other key facilities to obtain reasonable ground-motion parameters.

The East Kunlun Fault Zone, as a highly seismically active fault, has witnessed five earthquakes with magnitudes exceeding M7.0 to the west of Animaqing Mountain since 1900. Conversely, the historical records for the Maqin–Maqu segment in the east of Animaqing Mountain show no M7.0 or above earthquakes, designating it as a distinctive seismic gap within this fault zone. We analyzed the tectonic background and structural features of the Maqin–Maqu segment within the East Kunlun Fault Zone to evaluate its potential seismic capacity. Utilizing a new established probability recurrence model, we calculated the seismic hazard for both segments over the next 100 years. The results indicate that the probability of M7.0 or above earthquake occurring in the Maqu segment in the next 100 years is 11.47%, classified as a moderate probability event. The joint probability of at least one M7.0 or above strong earthquake occurring in the entire Maqin–Maqu segment in the next 100 years is 16.14%, also classified as a moderate probability event, while the probability for the Maqin segment alone is 5.36%, classified as a low probability event. Considering the uncertainty of the probability model, a qualitative hazard classification for each segment was further conducted. The comprehensive evaluation suggests a low risk of a major earthquake occurring in the Maqin segment in the next 100 years, while the Maqu segment is assessed to have a higher risk.

For the contour extraction from the images, the traditional active contour models may trap in local minimum and strongly depend on initial contour. A contour extraction algorithm based on a robust neural network is proposed in this paper. A searching series of circles are used in obtained feature pixels with adaptive threshold for the final curve function approaching by neural network. Robust back propagation algorithm has been used to control the final curve shape. The simulations also show that the proposed algorithm has a great performance for different kinds of images.