Background: Depression is a serious personal and public mental health problem. Self-reporting is the main method used to diagnose depression and to determine the severity of depression. However, it is not easy to discover patients with depression owing to feelings of shame in disclosing or discussing their mental health conditions with others. Moreover, self-reporting is time-consuming, and usually leads to missing a certain number of cases. Therefore, automatic discovery of patients with depression from other sources such as social media has been attracting increasing attention. Social media, as one of the most important daily communication systems, connects large quantities of people, including individuals with depression, and provides a channel to discover patients with depression. In this study, we investigated deep-learning methods for depression risk prediction using data from Chinese microblogs, which have potential to discover more patients with depression and to trace their mental health conditions. Objective: The aim of this study was to explore the potential of state-of-the-art deep-learning methods on depression risk prediction from Chinese microblogs. Methods: Deep-learning methods with pretrained language representation models, including bidirectional encoder representations from transformers (BERT), robustly optimized BERT pretraining approach (RoBERTa), and generalized autoregressive pretraining for language understanding (XLNET), were investigated for depression risk prediction, and were compared with previous methods on a manually annotated benchmark dataset. Depression risk was assessed at four levels from 0 to 3, where 0, 1, 2, and 3 denote no inclination, and mild, moderate, and severe depression risk, respectively. The dataset was collected from the Chinese microblog Weibo. We also compared different deep-learning methods with pretrained language representation models in two settings: (1) publicly released pretrained language representation models, and (2) language representation models further pretrained on a large-scale unlabeled dataset collected from Weibo. Precision, recall, and F1 scores were used as performance evaluation measures. Results: Among the three deep-learning methods, BERT achieved the best performance with a microaveraged F1 score of 0.856. RoBERTa achieved the best performance with a macroaveraged F1 score of 0.424 on depression risk at levels 1, 2, and 3, which represents a new benchmark result on the dataset. The further pretrained language representation models demonstrated improvement over publicly released prediction models. Conclusions: We applied deep-learning methods with pretrained language representation models to automatically predict depression risk using data from Chinese microblogs. The experimental results showed that the deep-learning methods performed better than previous methods, and have greater potential to discover patients with depression and to trace their mental health conditions.

Fiber Bragg grating (FBG) based accelerometer technology receives much attention from various research communities. In this paper, we investigated a FBG accelerometer with high resonant frequency for mechanical equipment. A special steel tube-mass block elastic structure with two FBGs embedded has been designed for axial vibration sensing. The steel tube was fixed on the mass block and threaded pipe in the sensor by sintering and pre-stretched to serve as a spring element. Then, a sensor prototype has been manufactured and fully tested. Experimental results show that the sensor provides a high resonant frequency of 3806 Hz with a wide flat response region up to 1200 Hz, which is consistent with the theoretical value of 3993 Hz. The sensitivity of the sensor is 4.01 pm/g with a fitting linear correlation coefficient of 0.9997. Temperature change test and excitation test along the working direction and transverse direction have been carried out, which confirmed excellent anti-interference ability and temperature compensation ability of the sensor.

A novel technique, named auxiliary equation method, is applied in this research work for obtaining new traveling wave solutions for two interesting proposed systems: the Kaup-Boussinesq system and generalized Hirota-Satsuma coupled KdV system with beta time fractional derivative. Our solutions were obtained using MAPLE software. This technique shows a great potential to be applied in solving various nonlinear fractional differential equations arising from mathematical physics and ocean engineering. Since a standard equation has not been used as an auxiliary equation for this technique, different and novel solutions are obtained via this technique.

Internet of Things (IoT) devices have revolutionized real-time monitoring and distant patient care in smart cities’ healthcare systems. However, this advancement has come with several issues, such as data security, scalability, operational efficiency, and fault tolerance. Previous approaches are not well suited to the real-time processing of IoT data in healthcare, given the low latency, high throughput, and effective anomaly detection needed for such a task. Given these challenges, this paper proposes a hybrid Artificial Intelligence (AI)- and blockchain-based IoT governance framework for Internet hospitals using Proof-of-Authority (PoA) in smart cities. It encompasses application of the enhanced RSA for secure data transmission, real-time anomaly detection through the Isolation Forest algorithm, and a private blockchain architecture designed for high scalability. It effectively detects tampering and replay attacks to minimize illegitimate and unauthorized access or manipulation of patients’ data. The proposed framework achieves relatively significant improvements over a state-of-the-art baseline model. It has cut the transaction response time by 50%, doubled the Throughput Per Second (TPS), and attained a 100% detection performance in anomalies. Comparative analysis reveals its linear scalability with increasing workload, ensuring consistent performance under varying transaction volumes. This study’s findings highlight the proposed framework’s potential to mitigate key issues in IoT-enabled Internet hospitals in smart cities.

In the present paper, we consider a nonlinear fractional snap model with respect to a $\mathbb{G}$-Caputo derivative and subject to non-periodic boundary conditions. Some qualitative analysis of the solution, such as existence and uniqueness, are investigated in view of fixed-point theorems. Moreover, the stabilities of Ulam–Hyers and Ulam–Hyers–Rassias criterions are considered and investigated. Some numerical simulations were performed using MATLAB for understanding the theoretical results. All results in this work play an important role in understanding ocean engineering phenomena due to the huge applicability of jerk and snap in seakeeping, ride comfort, and shock response spectrum.

The increasing adoption of Internet hospital systems—enabled by the real-time data streaming capabilities of the Internet of Medical Things (IoMT)—has intensified the need for secure, scalable, and low-latency data management infrastructures. Existing blockchain-based solutions often fail to meet these requirements, particularly under high-frequency workloads and stringent privacy demands. To address these limitations, this study proposes a simulation-based post-quantum-inspired alliance blockchain architecture tailored for Internet hospital systems. The framework incorporates four key innovations: (1) a Kyber-inspired hybrid encryption simulation, reducing encryption and decryption times by 72.3% and 74.4%, respectively, compared to RSA-2048; (2) a lightweight patient-centric access control mechanism based on authorization proofs achieving an average verification latency of ∼0.002 ms; (3) a Raft-based scalable consensus protocol, tested under a synchronous constant-delay network assumption, reducing consensus latency by 92.3% while supporting up to 1000 nodes with sub-150 ms finality; and (4) a fault-tolerant IoMT data ingestion layer using 3-of-5 median filtering, sustaining 90–96.2% sensor correction accuracy under varying fault injection rates. The system is prototyped in Google Colab Pro using synthetic data from 1000 virtual patients. Comparative benchmarks against PBFT and RSA-based systems show a fivefold increase in throughput, ∼9.4–12.3% energy savings per transaction, and ∼14% lower memory consumption during encryption. With a modest daily storage footprint (∼15 MB/day), the proposed solution is both resource-efficient and deployment-ready in simulation environments. These results confirm the potential of this architecture to enable trustworthy, energy-aware, and real-time blockchain infrastructures for next-generation digital healthcare ecosystems.