Graph neural pre-training based drug-target affinity prediction.

Computational drug-target affinity prediction has the potential to accelerate drug discovery. Currently, pre-training models have achieved significant success in various fields due to their ability to train the model using vast amounts of unlabeled data. However, given the scarcity of drug-target interaction data, pre-training models can only be trained separately on drug and target data, resulting in features that are insufficient for drug-target affinity prediction. To address this issue, in this paper, we design a graph neural pre-training-based drug-target affinity prediction method (GNPDTA). This approach comprises three stages. In the first stage, two pre-training models are utilized to extract low-level features from drug atom graphs and target residue graphs, leveraging a large number of unlabeled training samples. In the second stage, two 2D convolutional neural networks are employed to combine the extracted drug atom features and target residue features into high-level representations of drugs and targets. Finally, in the third stage, a predictor is used to predict the drug-target affinity. This approach fully utilizes both unlabeled and labeled training samples, enhancing the effectiveness of pre-training models for drug-target affinity prediction. In our experiments, GNPDTA outperforms other deep learning methods, validating the efficacy of our approach.

Modeling of joint extraction of entity relationships in clinical electronic medical records.

The advancement of medical informatization necessitates extracting entities and their relationships from electronic medical records. Presently, research on electronic medical records predominantly concentrates on single-entity relationship extraction. However, clinical electronic medical records frequently exhibit overlapping complex entity relationships, thereby heightening the challenge of information extraction. To rectify the absence of a clinical medical relationship extraction dataset, this study utilizes electronic medical records from 584 patients in a hospital to create a compact clinical medical relationship extraction dataset. To address the pipelined relationship extraction model's limitation in overlooking the one-to-many correlation problem between entities and relationships, this paper introduces a cascading relationship extraction model. This model integrates the MacBERT pre-training model, gated recurrent network, and multi-head self-attention mechanism to enhance the extraction of text features. Simultaneously, adversarial learning is incorporated to bolster the model's robustness. In scenarios involving one-to-many relationships between entities, a two-phase task is employed. Initially, the main entity is predicted, followed by predicting the associated object and their correspondences. Employing this cascade-structured approach enables the model to flexibly manage intricate entity relationships, thereby enhancing extraction accuracy. Experimental results demonstrate the model's efficiency, yielding F1-scores of 82.8%, 76.8%, and 88.2% for fulfilling relational extraction requirements and tasks on DuIE, CHIP-CDEE, and private datasets, respectively. These scores represent improvements over the benchmark model. The findings indicate the model's applicability in practical domains, particularly in tasks such as biomedical information extraction.

Innovation practices in agricultural transformation in East China: Exploring the impact and implications of the new professional farmer training model.

This study aims to invigorate China's Rural Revitalization Strategy by focusing on the training of China's new generation of professional farmers, particularly enhancing the skills of modern agricultural practitioners. Utilizing a mixed-method research approach, which includes the analysis of 343 questionnaire surveys and personal interviews, significant shortcomings in the previous training of farmers were revealed, such as limited reach, high conformity in training content, and outdated methods. These findings highlight the challenges traditional training methods face in the digital technology era. In response to these challenges, this study proposes a novel training model designed for the digital era. This model encompasses seven aspects: training goals, subjects, content, means, management, methods, and effect assessment, with the aspiration to reform traditional farmer training methods. This research expands the existing literature by constructing a comprehensive framework for training the new generation of farmers, bridging the gap between traditional practices and the demands of modern agricultural training influenced by digital advancements. The study's innovation lies in its proposition of a modernized training model for the new generation of farmers, leveraging the advancements of the digital technology era. By modernizing agricultural training and enhancing the capabilities of new professional farmers, it significantly contributes to the progression of China's Rural Revitalization Strategy.

A Novel Training Model for Superficial Temporal Artery- Middle Cerebral Artery Anastomosis Using Microsurgical Techniques.

OBJECTIVE: To create a reusable and inexpensive training model with technological tools that simulates cerebral bypass surgery and a sensor system that provides tactile feedback to the surgeon. Furthermore, we aimed to evaluate the anastomotic stability and contribution to the surgeon's learning curve. METHODS: We created a superficial temporal artery-middle cerebral artery bypass simulation model using chicken and turkey brachial arteries. A cranium model was printed with a three-dimensional printer for craniotomy and cerebral parenchyma was created by pouring silicone into the cranial mold. A blood flow simulation system was also prepared. Pressure-sensitive sensors were placed on parenchyma and tactile conditioning was performed via audible warning from the sensors. Twenty-four anastomosis were performed with different sutures and hand tools. Anastomosis completion times and durability and the number of touches and pressures applied to the parenchyma were recorded. The stability of the anastomoses was evaluated by increasing the pressure in the blood flow simulation system, so usefulness of the training model was evaluated. RESULTS: The time required for anastomosis completion decreased as the number of practices increased (P < 0.05). As the number of practices increased, the number of parenchymal touches decreased (P < 0.05). CONCLUSIONS: With practice, the time required for anastomosis completion and number of parenchymal touches decreased. Thus, the model is useful, inexpensive, reusable, easily accessible, and contributes to the surgeon's learning curve. Our model with pressure-sensitive sensors can be used for microsurgery practice, enabling the surgeons to gain tactile conditioning and evaluate anastomotic stability and leakage.

Various handcrafted artificial vessels: Evaluation of practicality and feasibility for supermicrosurgery training.

BACKGROUND: Supermicrosurgery demands more refined skills compared to traditional microsurgery, necessitating comprehensive training prior to clinical implementation. Despite the existence of various training models, they often fall short in terms of cost, ethical considerations, and infection risk. Our objective was to develop and evaluate novel training models for supermicrosurgery that are cost-effective, ethical, and risk-free. METHODS: We fabricated tubes using polyvinyl alcohol (PVA) liquid glue, polyvinyl acetate resin (PAR) wood glue, and hydrocolloid dressing (HCD), aiming to identify suitable, low-cost candidates for a supermicrosurgery training model. These tubes were anastomosed under a microscope using 10-0 or 11-0 nylon sutures. We assessed the time and cost involved in tube fabrication, their diameters, and the overall feasibility of the models. RESULTS: The average time and cost to fabricate a 15-mm-long luminal tube were 33.5 min and 0.02 USD for the PVA group, 23 min and 0.02 USD for the PAR group, and 63 s and 0.40 USD for the HCD group, respectively. The average diameter of the tubes was 0.49, 0.58, and 1.55 mm in the PVA, PAR, and HCD groups, respectively. The PVA and PAR tubes, with their transparent and thin walls, facilitated easier evaluation of anastomosis patency compared to the HCD tubes. CONCLUSION: We successfully used non-living materials to develop new supermicrosurgery training models, characterized by their low cost, absence of ethical concerns, and elimination of infection risk. The PAR and PVA tubes, in particular, are suitable for resident training in supermicrosurgery.

Depression recognition using voice-based pre-training model.

The early screening of depression is highly beneficial for patients to obtain better diagnosis and treatment. While the effectiveness of utilizing voice data for depression detection has been demonstrated, the issue of insufficient dataset size remains unresolved. Therefore, we propose an artificial intelligence method to effectively identify depression. The wav2vec 2.0 voice-based pre-training model was used as a feature extractor to automatically extract high-quality voice features from raw audio. Additionally, a small fine-tuning network was used as a classification model to output depression classification results. Subsequently, the proposed model was fine-tuned on the DAIC-WOZ dataset and achieved excellent classification results. Notably, the model demonstrated outstanding performance in binary classification, attaining an accuracy of 0.9649 and an RMSE of 0.1875 on the test set. Similarly, impressive results were obtained in multi-classification, with an accuracy of 0.9481 and an RMSE of 0.3810. The wav2vec 2.0 model was first used for depression recognition and showed strong generalization ability. The method is simple, practical, and applicable, which can assist doctors in the early screening of depression.

Development of a porcine training model for microvascular fasciocutaneous free flap reconstruction.

BACKGROUND: In reconstructive surgery, improvements are needed in the effective teaching of free flap surgery. There is a need for easily accessible and widely available training without high financial costs or ethical concerns while still providing a realistic experience. Our aim was to develop an appropriate training model for microvascular flaps. METHODS: We identified pig head halves as most appropriate regarding availability, cost, and realism. These accrue largely by the food industry, so no animals need to be sacrificed, making it more ethical from an animal welfare perspective. We evaluated the suitability as flap donor site and analyzed the vascular anatomy of 51 specimens. RESULTS: Anatomical evaluation revealed a reliable and constant vascular anatomy, allowing the design of a flap model that can effectively illustrate the entire process of microvascular flap surgery. The process was divided into 6 key steps. The flap can be harvested after marking the vascular pedicle 5.3 cm from the lateral corner of the mouth. Skin island design and subsequent tissue dissection follow until a fasciocutaneous flap is raised, similar to a radial flap. Upon completion of flap harvesting, it can be freely transferred for defect reconstruction. Microvascular anastomosis can be performed on recipient vessels in the cervical region, and the difficulty can be individually adjusted. CONCLUSIONS: The developed training model is a reasonable compromise in terms of surgical realism, availability, didactic value, and cost/time effectiveness. We believe it is a powerful and effective tool with high potential for improving surgical education and training.

Novel Cost-Effective Model for Training Post-intubation Endotracheal Tube Placement Confirmation.

Intubation in emergency settings demands rapid confirmation of endotracheal tube (ETT) placement for establishing a definitive airway. Current methods, including capnography and auscultation, have limitations. This study introduces a cost-effective and easily accessible training model for confirming ETT placement using ultrasound, aiming to improve training and patient outcomes. We developed a gelatin and psyllium-based model that simulates adult ETT intubation, offering an alternative to costly cadaveric models. The model's construction is described, with materials costing approximately $7.34 per unit. Preliminary results show promise in simulating tracheal and esophageal intubation scenarios. This novel model provides an ethical and economical solution for training healthcare professionals in the ultrasound confirmation of ETT placement, paving the way for further validation and adoption in medical education.

FusPB-ESM2: Fusion model of ProtBERT and ESM-2 for cell-penetrating peptide prediction.

Cell-penetrating peptides have attracted much attention for their ability to break through cell membrane barriers, which can improve drug bioavailability, reduce side effects, and promote the development of gene therapy. Traditional wet-lab prediction methods are time-consuming and costly, and computational methods provide a short-time and low-cost alternative. Still, the accuracy and reliability need to be further improved. To solve this problem, this study proposes a feature fusion-based prediction model, where the protein pre-trained language models ProtBERT and ESM-2 are used as feature extractors, and the extracted features from both are fused to obtain a more comprehensive and effective feature representation, which is then predicted by linear mapping. Validated by many experiments on public datasets, the method has an AUC value as high as 0.983 and shows high accuracy and reliability in cell-penetrating peptide prediction.

Descriptive Anatomy of the Porcine Ventral Abdominal Wall as a Basis for Training Ventral Hernia Repair Techniques.

Background: In recent times there has been a surge in innovative techniques concerning complex abdominal wall surgery. The availability of simulation models for comprehensive training and skill set development remains limited. Methods: Cadaveric dissections of the porcine abdominal wall were conducted to assess the suitability of anesthetized porcine models for training in both minimally invasive and open surgical procedures. Results: The panniculus carnosus, a typical muscular layer in mammals, is the outermost layer covering the anterolateral abdominal wall. Beneath it, there are four main pairs of abdominal wall muscles, mirroring the human anatomy. The rectus abdominis muscle runs straight along the linea alba and is surrounded by the rectus sheath, which is formed by the fusion of the lateral abdominal wall muscles and differs along the different regions of abdominal wall. The orientation of the muscle fibers in the lateral abdominal wall muscles, i.e., musculus obliquus externus, internus and transversus, is comparable to human anatomy. Although the transition lines between their muscular and aponeurotic part differ to some extent. Relevant for the adoption of surgical techniques, the transversus abdominis muscle is well-developed and resembles a bell curve shape as it transitions from its muscular to aponeurotic part. Conclusion: Despite minor differences in abdominal wall anatomy between pigs and humans, the porcine model provides a high level of fidelity in terms of both anatomical features and the development of skill sets relevant to hernia surgery.

Development and validation of metrics for a new RAPN training model.

Robot-assisted partial nephrectomy (RAPN) is a complex and index procedure that urologists need to learn how to perform safely. No validated performance metrics specifically developed for a RAPN training model (TM) exist. A Core Metrics Group specifically adapted human RAPN metrics to be used in a newly developed RAPN TM, explicitly defining phases, steps, errors, and critical errors. A modified Delphi meeting concurred on the face and content validation of the new metrics. One hundred percent consensus was achieved by the Delphi panel on 8 Phases, 32 Steps, 136 Errors and 64 Critical Errors. Two trained assessors evaluated recorded video performances of novice and expert RAPN surgeons executing an emulated RAPN in the newly developed TM. There were no differences in procedure Steps completed by the two groups. Experienced RAPN surgeons made 34% fewer Total Errors than the Novice group. Performance score for both groups was divided at the median score using Total Error scores, into HiError and LoError subgroups. The LowErrs Expert RAPN surgeons group made 118% fewer Total Errors than the Novice HiErrs group. Furthermore, the LowErrs Expert RAPN surgeons made 77% fewer Total Errors than the HiErrs Expert RAPN surgeons. These results established construct and discriminative validity of the metrics. The authors described a novel RAPN TM and its associated performance metrics with evidence supporting their face, content, construct, and discriminative validation. This report and evidence support the implementation of a simulation-based proficiency-based progression (PBP) training program for RAPN.

Enhancing the Microneurosurgical Training: Development of the Folded Placenta Model for Simulation of the Deep Operative Field Challenges of Cranial Procedures.

BACKGROUND AND OBJECTIVE: Neurosurgery relies heavily on advanced manual skills, necessitating effective training models for skill development. While various models have been utilized, the human placenta has emerged as a promising candidate for microneurosurgical training due to its anatomical similarities with cerebral vasculature. However, existing placenta models have primarily focused on simulating superficial procedures, often neglecting the complexities encountered in deep operative fields during cranial surgeries. METHODS: This study obtained ethical approval and implemented a modified placenta model to address the limitations of existing training models. The key modification involved folding the placenta and placing it within a rigid container, closely mimicking the structural challenges of cranial procedures. The placenta preparation followed a standardized protocol, including the use of specialized equipment for documentation. RESULTS: The primary feature of the modified model is the folded placenta within the rigid container, which replicates cranial anatomy. This innovative approach enables trainees to engage in a comprehensive range of microsurgical exercises, encompassing vessel dissection, aneurysm clipping, tumor resection, and more. The model successfully mirrors the complexities of real cranial procedures, providing a realistic training experience. CONCLUSIONS: The presented modified placenta model serves as an effective tool for simulating the conditions encountered in deep cranial surgeries. By accurately replicating the challenges of deep operative fields, the model significantly enhances the training of neurosurgical residents. It successfully prepares trainees to navigate the intricacies and difficulties inherent in real cranial surgeries, thus contributing to improved surgical skills and readiness for neurosurgical practice.

RM-GPT: Enhance the comprehensive generative ability of molecular GPT model via LocalRNN and RealFormer.

Due to the surging of cost, artificial intelligence-assisted de novo drug design has supplanted conventional methods and become an emerging option for drug discovery. Although there have arisen many successful examples of applying generative models to the molecular field, these methods struggle to deal with conditional generation that meet chemists' practical requirements which ask for a controllable process to generate new molecules or optimize basic molecules with appointed conditions. To address this problem, a Recurrent Molecular-Generative Pretrained Transformer model is proposed, supplemented by LocalRNN and Residual Attention Layer Transformer, referred to as RM-GPT. RM-GPT rebuilds GPT model's architecture by incorporating LocalRNN and Residual Attention Layer Transformer so that it is able to extract local information and build connectivity between attention blocks. The incorporation of Transformer in these two modules enables leveraging the parallel computing advantages of multi-head attention mechanisms while extracting local structural information effectively. Through exploring and learning in a large chemical space, RM-GPT absorbs the ability to generate drug-like molecules with conditions in demand, such as desired properties and scaffolds, precisely and stably. RM-GPT achieved better results than SOTA methods on conditional generation.

Promoting the process of determining brain death through standardized training.

Objective: This study aims to explore the training mode for brain death determination to ensure the quality of subsequent brain death determination. Methods: A four-skill and four-step (FFT) training model was adopted, which included a clinical neurological examination, an electroencephalogram (EEG) examination, a short-latency somatosensory evoked potential (SLSEP) examination, and a transcranial Doppler (TCD) examination. Each skill is divided into four steps: multimedia theory teaching, bedside demonstration, one-on-one real or dummy simulation training, and assessment. The authors analyzed the training results of 1,577 professional and technical personnel who participated in the FFT training model from 2013 to 2020 (25 sessions), including error rate analysis of the written examination, knowledge gap analysis, and influencing factors analysis. Results: The total error rates for all four written examination topics were < 5%, at 4.13% for SLSEP, 4.11% for EEG, 3.71% for TCD, and 3.65% for clinical evaluation. The knowledge gap analysis of the four-skill test papers suggested that the trainees had different knowledge gaps. Based on the univariate analysis and the multiple linear regression analysis, among the six factors, specialty categories, professional and technical titles, and hospital level were the independent influencing factors of answer errors (p < 0.01). Conclusion: The FFT model is suitable for brain death (BD) determination training in China; however, the authors should pay attention to the professional characteristics of participants, strengthen the knowledge gap training, and strive to narrow the difference in training quality.

[A research on depression recognition based on voice pre-training model].

For the increasing number of patients with depression, this paper proposes an artificial intelligence method to effectively identify depression through voice signals, with the aim of improving the efficiency of diagnosis and treatment. Firstly, a pre-training model called wav2vec 2.0 is fine-tuned to encode and contextualize the speech, thereby obtaining high-quality voice features. This model is applied to the publicly available dataset - the distress analysis interview corpus-wizard of OZ (DAIC-WOZ). The results demonstrate a precision rate of 93.96%, a recall rate of 94.87%, and an F1 score of 94.41% for the binary classification task of depression recognition, resulting in an overall classification accuracy of 96.48%. For the four-class classification task evaluating the severity of depression, the precision rates are all above 92.59%, the recall rates are all above 92.89%, the F1 scores are all above 93.12%, and the overall classification accuracy is 94.80%. The research findings indicate that the proposed method effectively enhances classification accuracy in scenarios with limited data, exhibiting strong performance in depression identification and severity evaluation. In the future, this method has the potential to serve as a valuable supportive tool for depression diagnosis.

Mechanical properties evaluation of two resin-based three-dimensional printing materials and dentine for endodontic training model fabrication: In vitro study.

The objective of the study was to compare 4 mechanical properties of two 3-D printing resins with dentine. Four mechanical tests were performed on dentine specimens, VeroClear resin and Splint resin with 15 samples each. Vickers hardness test, samples were subjected to a 300-g load for 10 s. Flexural strength test, 8-millimetre, beam-shaped specimens were tested using the three-point bending method. Drilling force was measured on flat-surface specimens. Intra-canal torque was measured on specimens imitating root canals. The results were analysed using the Kruskal-Wallis and Mann-Whitney tests with adjusted Bonferroni's p-value (p < 0.05). Dentine exhibited the highest mechanical properties in all the tests. VeroClear's flexural strength and drilling force were higher, while the surface hardness was lower than that of Splint resin. The intra-canal torque of dentine and VeroClear resin were not significant and higher than that of Splint resin. VeroClear's mechanical properties were closer to dentine than Splint resin.

Towards detection of cancer biomarkers in human exhaled air by transfer-learning-powered analysis of odor-evoked calcium activity in rat olfactory bulb.

Detection of volatile organic compounds in exhaled air is a promising approach to non-invasive and scalable gastric cancer screening. This work proposes a new approach for the detection of volatile organic compounds by analyzing odor-evoked calcium responses in the rat olfactory bulb. We estimate the feasibility of gastric cancer biomarker detection added to the exhaled air of healthy participants. Our detector consists of a convolutional encoder and a similarity-based classifier over encoder outputs. To minimize overfitting on a small available training set, we involve a pre-training where the encoder is trained on synthetic data representing spatiotemporal patterns similar to real calcium responses in the olfactory bulb. We estimate the classification accuracy of exhaled air samples by matching their encodings with encodings of calibration samples of two classes: 1) exhaled air and 2) a mixture of exhaled air with the cancer biomarker. On our data, the accuracy increased from 0.68 on real data up to 0.74 if pre-training on synthetic data is involved. Our work is focused on proving the feasibility of proposed new approach rather than on comparing its efficiency with existing methods. Such detection is often performed with an electronic nose, but its output becomes unstable over time due to a sensor drift. In contrast to the electronic nose, rats can robustly detect low concentrations of biomarkers over lifetime. The feasibility of gastric cancer biomarker detection in exhaled air by bio-hybrid system is shown. Pre-training of neural models for images analysis increases the accuracy of detection.

Validation of a 3D-printed robot-assisted partial nephrectomy training model.

Objectives: Most renal tumours can be treated with a partial nephrectomy, with robot-assisted partial nephrectomy becoming the new gold standard. This procedure is challenging to learn in a live setting, especially the enucleation and renorraphy phases. In this study, we attempted to evaluate face, content, and preliminary construct validity of a 3D-printed silicone renal tumour model in robotic training for robot-assisted partial nephrectomy. Materials and Methods: We compared the operative results of three groups of surgeons with different experience levels (>20 partial nephrectomies, 1-20 partial nephrectomies and no experience at all) performing a robotic tumour excision of a newly developed silicone model with four embedded 3D-printed renal tumours. We evaluated the participants' performance using surgical margins, excision time, total preserved parenchyma, tumour injury and GEARS score (as assessed by two blinded experts) for construct validity. Postoperatively, the participants were asked to complete a survey to evaluate the usefulness, realism and difficulty of the model as a training and/or evaluation model. NASA-TLX scores were used to evaluate the operative workload. Results: Thirty-six participants were recruited, each group consisting of 10-14 participants. The operative performance was significantly better in the expert group as compared to the beginner group. NASA-TLX scores proved the model to be of an acceptable difficulty level.Expert group survey results showed an average score of 6.3/10 on realism of the model, 8.2/10 on the usefulness as training model and 6.9/10 score on the usefulness as an evaluation tool. GEARS scores showed a non-significant tendency to improve between trials, emphasizing its potential as a training model. Conclusion: Face and content validity of our 3D renal tumour model were demonstrated. The vast majority of participants found the model realistic and useful for training and for evaluation. To evaluate construct and predictive validity, we require further research, aiming to compare the results of 3D-model trained surgeons with those of untrained surgeons in real-life surgery.

iSUMO-RsFPN: A predictor for identifying lysine SUMOylation sites based on multi-features and feature pyramid networks.

SUMOylation is a protein post-translational modification that plays an essential role in cellular functions. For predicting SUMO sites, numerous researchers have proposed advanced methods based on ordinary machine learning algorithms. These reported methods have shown excellent predictive performance, but there is room for improvement. In this study, we constructed a novel deep neural network Residual Pyramid Network (RsFPN), and developed an ensemble deep learning predictor called iSUMO-RsFPN. Initially, three feature extraction methods were employed to extract features from samples. Following this, weak classifiers were trained based on RsFPN for each feature type. Ultimately, the weak classifiers were integrated to construct the final classifier. Moreover, the predictor underwent systematically testing on an independent test dataset, where the results demonstrated a significant improvement over the existing state-of-the-art predictors. The code of iSUMO-RsFPN is free and available at https://github.com/454170054/iSUMO-RsFPN.

Participatory training evaluation: Steps from the Center for Native American Health Native-CHART symposium.

This paper presents a case example of the Native-CHART Training Evaluation and describes the process of planning and administering a paper evaluation during the Native-CHART symposium in November 2019 led by the Center for Native American Health (CNAH) and an external evaluator. Training evaluation methodologies and the data collection instrument were grounded in the Health Belief Model (HBM) where health-related chronic disease and risk factor knowledge translates to perceived susceptibility, benefits, barriers, and self-efficacy. Kirkpatrick's Four-level Training Evaluation Model explored learning, reaction, behaviors, and results. The evaluation aims centered around the following questions: 1)Who attended the symposium, and why did they attend? 2)What knowledge did participants gain at the symposium? 3)Will attendees change their behaviors as a result of attending the symposium? 4) What parts of the symposium were most valuable? And 5) How can the symposium be improved? Data collected at the symposium answered these questions. After the Native-CHART symposium, CNAH staff and the external evaluator met to reflect on the steps necessary to plan and implement a participatory training evaluation. From these discussions, eight steps emerged. This paper presents these steps along with recommendations for future work. Participatory and collaborative approaches in training evaluation and the steps included in this case example may be useful to evaluators, communities, and programs working on designing and evaluating various trainings with Tribal populations.