In swimming, the posture and technique of athletes are crucial for improving performance. However, traditional swimming coaches often struggle to capture and analyze athletes\' movements in real-time, which limits the effectiveness of coaching. Therefore, this paper proposes RL-CWtrans Net: a robot vision-driven multimodal swimming training system that provides precise and real-time guidance and feedback to swimmers. The system utilizes the Swin-Transformer as a computer vision model to effectively extract the motion and posture features of swimmers. Additionally, with the help of the CLIP model, the system can understand natural language instructions and descriptions related to swimming. By integrating visual and textual features, the system achieves a more comprehensive and accurate information representation. Finally, by employing reinforcement learning to train an intelligent agent, the system can provide personalized guidance and feedback based on multimodal inputs. Experimental results demonstrate significant advancements in accuracy and practicality for this multimodal robot swimming coaching system. The system is capable of capturing real-time movements and providing immediate feedback, thereby enhancing the effectiveness of swimming instruction. This technology holds promise.

The dried ripe fruit of Terminalia chebula Retz. is a common Chinese materia medica, and ellagic acid (EA), isolated from the plant, is an important bioactive component for medicinal purposes. This study aimed to delineate the optimal extraction parameters for extracting the EA content from Chebulae Fructus (CF), focusing on the variables of ethanol concentration, extraction temperature, liquid-solid ratio, and extraction time. Utilizing a combination of the response surface methodology (RSM) and an artificial neural network (ANN), we systematically investigated these parameters to maximize the EA extraction efficiency. The extraction yields for EA obtained under the predicted optimal conditions validated the efficacy of both the RSM and ANN models. Analysis using the ANN-predicted data showed a higher coefficient of determination (R2) value of 0.9970 and a relative error of 0.79, compared to the RSM\'s 2.85. The optimal conditions using the ANN are an ethanol concentration of 61.00%, an extraction temperature of 77 °C, a liquid-solid ratio of 26 mL g-1 and an extraction time of 103 min. These findings significantly enhance our understanding of the industrial-scale optimization process for EA extraction from CF.

Alzheimer\'s Disease (AD) is an irreversible, degenerative condition that, while incurable, can have its progression slowed or impeded. While there are numerous methods utilizing neural networks for AD detection, there is a scarcity of High-performance AD detection chips. Moreover, excessively complex neural networks are not conducive to on-chip implementation and clinical applications. This study addresses the challenges of high misdiagnosis rates and significant hardware costs inherent in traditional AD detection techniques. A novel and efficient AD detection framework based on a recurrent computational strategy is proposed. The framework harnesses an Artificial Neural Network (ANN) embedded within a System on Chip (SoC) to perform sophisticated Electroencephalogram (EEG) analysis. The approach began by employing a reduced IEEE754 single-precision encoding method to hardware-encode the preprocessed EEG data, thereby minimizing the memory storage area. Next, data remapping techniques were utilized to ensure the continuity of the input data read addresses and reduce the memory access pressure during neural network computations. Subsequently, hierarchical and Processing Element (PE) reuse technologies were leveraged to perform the multiply-accumulate operations of the ANN. Finally, a step function was chosen to establish binary classification circuits dedicated to AD detection. Experimental results indicate that the optimized SoC achieves a 70 % reduction in area and a 50 % reduction in power consumption compared to traditional designs. For various neural network models, the detection model proposed in this paper incurs less overhead, with a training speed 3 to 4 times faster than other traditional models, and a high accuracy rate of 98.53 %.

Biomimetic neuromorphic sensing systems, inspired by the structure and function of biological neural networks, represent a major advancement in the field of sensing technology and artificial intelligence. This review paper focuses on the development and application of electrolyte gated transistors (EGTs) as the core components (synapses and neuros) of these neuromorphic systems. EGTs offer unique advantages, including low operating voltage, high transconductance, and biocompatibility, making them ideal for integrating with sensors, interfacing with biological tissues, and mimicking neural processes. Major advances in the use of EGTs for neuromorphic sensory applications such as tactile sensors, visual neuromorphic systems, chemical neuromorphic systems, and multimode neuromorphic systems are carefully discussed. Furthermore, the challenges and future directions of the field are explored, highlighting the potential of EGT-based biomimetic systems to revolutionize neuromorphic prosthetics, robotics, and human-machine interfaces. Through a comprehensive analysis of the latest research, this review is intended to provide a detailed understanding of the current status and future prospects of biomimetic neuromorphic sensory systems via EGT sensing and integrated technologies.

The objectives of this study were to evaluate the energy partition patterns of growing pigs fed diets with different net energy (NE) levels based on machine learning methods, and to develop prediction models for the NE requirement of growing pigs. Twenty-four Duroc × Landrace × Yorkshire crossbred barrows with an initial body weight of 24.90 ± 0.46 kg were randomly assigned to 3 dietary treatments, including the low NE group (2,325 kcal/kg), the medium NE group (2,475 kcal/kg), and the high NE group (2,625 kcal/kg). The total feces and urine produced from each pig during each period were collected, to calculate the NE intake, NE retained as protein (NEp), and NE retained as lipid (NEl). A total of 240 sets of data on the energy partition patterns of each pig were collected, 75% of the data in the dataset was randomly selected as the training dataset, and the remaining 25% was set as the testing dataset. Prediction models for the NE requirement of growing pigs were developed using algorithms including multiple linear regression (MR), artificial neural networks (ANN), k-nearest neighbor (KNN), and random forest (RF), and the prediction performance of these models was compared on the testing dataset. The results showed pigs in the low NE group showed a lower average daily gain, lower average daily feed intake, lower NE intake, but greater feed conversion ratio compared to pigs in the high NE group in most growth stages. In addition, pigs in the 3 treatment groups did not show a significant difference in NEp in all growth stages, while pigs in the medium and high NE groups showed greater NEl compared to pig in the low NE group in growth stages from 25 to 55 kg (P < 0.05). Among the developed prediction models for NE intake, NEp, and NEl, the ANN models demonstrated the most optimal prediction performance with the smallest root mean square error (RMSE) and the largest R2, while the RF models had the worst prediction performance with the largest RMSE and the smallest R2. In conclusion, diets with varied NE concentrations within a certain range did not affect the NEp of growing pigs, and the models developed with the ANN algorithm could accurately achieve the NE requirement prediction in growing pigs.
Net energy (NE) can unify the energy value of the feed with the energy requirements of the pig more accurately and is the optimal system for accurately predicting the growth performance of pigs. The evaluation of the NE partition pattern is difficult and costly, thus, establishing a predicted model is a more efficient way. This study was conducted to evaluate the energy partition patterns of growing pigs fed diets with different NE levels based on machine learning methods. Diets with varied NE concentrations within a certain range did not affect the growth performance and NE requirement for lipid deposition in growing pigs. Among the 4 models developed to predict NE requirements, the artificial neural networks model had the highest accuracy, while the multiple linear regression model had the highest interpretability.

暂无摘要。

BACKGROUND: The escalating global prevalence of obesity has necessitated the exploration of novel diagnostic approaches. Recent scientific inquiries have indicated potential alterations in voice characteristics associated with obesity, suggesting the feasibility of using voice as a noninvasive biomarker for obesity detection.
OBJECTIVE: This study aims to use deep neural networks to predict obesity status through the analysis of short audio recordings, investigating the relationship between vocal characteristics and obesity.
METHODS: A pilot study was conducted with 696 participants, using self-reported BMI to classify individuals into obesity and nonobesity groups. Audio recordings of participants reading a short script were transformed into spectrograms and analyzed using an adapted YOLOv8 model (Ultralytics). The model performance was evaluated using accuracy, recall, precision, and F1-scores.
RESULTS: The adapted YOLOv8 model demonstrated a global accuracy of 0.70 and a macro F1-score of 0.65. It was more effective in identifying nonobesity (F1-score of 0.77) than obesity (F1-score of 0.53). This moderate level of accuracy highlights the potential and challenges in using vocal biomarkers for obesity detection.
CONCLUSIONS: While the study shows promise in the field of voice-based medical diagnostics for obesity, it faces limitations such as reliance on self-reported BMI data and a small, homogenous sample size. These factors, coupled with variability in recording quality, necessitate further research with more robust methodologies and diverse samples to enhance the validity of this novel approach. The findings lay a foundational step for future investigations in using voice as a noninvasive biomarker for obesity detection.

The application of artificial neural networks (ANNs) in the treatment of wastewater has achieved increasing attention, as it enhances the efficiency and sustainability of wastewater treatment plants (WWTPs). This paper explores the application of ANN-based models in WWTPs, focusing on the latest published research work, by presenting the effectiveness of ANNs in predicting, estimating, and treatment of diverse types of wastewater. Furthermore, this review comprehensively examines the applicability of the ANNs in various processes and methods used for wastewater treatment, including membrane and membrane bioreactors, coagulation/flocculation, UV-disinfection processes, and biological treatment systems. Additionally, it provides a detailed analysis of pollutants viz organic and inorganic substances, nutrients, pharmaceuticals, drugs, pesticides, dyes, etc., from wastewater, utilizing both ANN and ANN-based models. Moreover, it assesses the techno-economic value of ANNs, provides cost estimation and energy analysis, and outlines promising future research directions of ANNs in wastewater treatment. AI-based techniques are used to predict parameters such as chemical oxygen demand (COD) and biological oxygen demand (BOD) in WWTP influent. ANNs have been formed for the estimation of the removal efficiency of pollutants such as total nitrogen (TN), total phosphorus (TP), BOD, and total suspended solids (TSS) in the effluent of WWTPs. The literature also discloses the use of AI techniques in WWT is an economical and energy-effective method. AI enhances the efficiency of the pumping system, leading to energy conservation with an impressive average savings of approximately 10%. The system can achieve a maximum energy savings state of 25%, accompanied by a notable reduction in costs of up to 30%.

UNASSIGNED: Acute abdominal pain (AAP) is a common symptom presented in the emergency department (ED), and it is crucial to have objective and accurate triage. This study aims to develop a machine learning-based prediction model for AAP triage. The goal is to identify triage indicators for critically ill patients and ensure the prompt availability of diagnostic and treatment resources.
UNASSIGNED: In this study, we conducted a retrospective analysis of the medical records of patients admitted to the ED of Wuhan Puren Hospital with acute abdominal pain in 2019. To identify high-risk factors, univariate and multivariate logistic regression analyses were used with thirty-one predictor variables. Evaluation of eight machine learning triage prediction models was conducted using both test and validation cohorts to optimize the AAP triage prediction model.
UNASSIGNED: Eleven clinical indicators with statistical significance (p < 0.05) were identified, and they were found to be associated with the severity of acute abdominal pain. Among the eight machine learning models constructed from the training and test cohorts, the model based on the artificial neural network (ANN) demonstrated the best performance, achieving an accuracy of 0.9792 and an area under the curve (AUC) of 0.9972. Further optimization results indicate that the AUC value of the ANN model could reach 0.9832 by incorporating only seven variables: history of diabetes, history of stroke, pulse, blood pressure, pale appearance, bowel sounds, and location of the pain.
UNASSIGNED: The ANN model is the most effective in predicting the triage of AAP. Furthermore, when only seven variables are considered, including history of diabetes, etc., the model still shows good predictive performance. This is helpful for the rapid clinical triage of AAP patients and the allocation of medical resources.

Mechanical computing provides an information processing method to realize sensing-analyzing-actuation integrated mechanical intelligence and, when combined with neural networks, can be more efficient for data-rich cognitive tasks. The requirement of solving implicit and usually nonlinear equilibrium equations of motion in training mechanical neural networks makes computation challenging and costly. Here, an explicit mechanical neuron is developed of which the response can be directly determined without the need of solving equilibrium equations. A training method is proposed to ensure the robustness of the neuron, i.e., insensitivity to defects and perturbations. The explicitness and robustness of the neurons facilitate the assembly of various network structures. Two exemplified networks, a robust mechanical convolutional neural network and a mechanical recurrent neural network with long short-term memory capabilities for associative learning, are experimentally demonstrated. The introduction of the explicit and robust mechanical neuron streamlines the design of mechanical neural networks fulfilling robotic matter with a level of intelligence.