Training Procedure Research Articles

Deep-learning based autonomous-exploration for UAV navigation

Unmanned Aerial Vehicles (UAVs), functioning as aerial robots, undertake multifarious missions within intricate and uncharted environments, necessitating autonomous exploration for efficient UAV navigation. This paper introduces a Deep Learning-based Autonomous UAV Exploration method (DLAE), which distinguishes itself from traditional autonomous ground robot exploration by employing cameras instead of radar sensors. This difference propels us toward proposing a hybrid action space that integrates both positional and yaw actions, aimed at circumventing the limitations imposed by the Field of View (FOV). Furthermore, to enhance learning efficiency in autonomous UAV exploration, we design an invalid action masking scheme. Significantly, our contribution includes the development of an autoregressive network model alongside a distinct training procedure specifically tailored for autonomous UAV exploration. Comprehensive experiments conducted across a variety of test maps reveal that our proposed method surpasses existing learning-based approaches, delivering superior exploration efficiency and reduced decision-making time in comparison to conventional strategies. Moreover, physical simulations have corroborated the effectiveness of our method in practical scenarios, showcasing its potential for real-world applications.

Fast emulation of two-point angular statistics for photometric galaxy surveys

ABSTRACT We develop a set of machine-learning-based cosmological emulators, to obtain fast model predictions for the C(ℓ) angular power spectrum coefficients, characterizing tomographic observations of galaxy clustering and weak gravitational lensing from multiband photometric surveys (and their cross-correlation). A set of neural networks are trained to map cosmological parameters into the coefficients, achieving, with respect to standard Boltzmann solvers, a speed-up of $\mathcal {O}(10^3)$ in computing the required statistics for a given set of cosmological parameters, with an accuracy better than 0.175 per cent (<0.1 per cent for the weak lensing case). This corresponds to $\lesssim 2~{{\ \rm per\ cent}}$ of the statistical error bars expected from a typical Stage IV photometric surveys. Such overall improvement in speed and accuracy is obtained through (i) a specific pre-processing optimization, ahead of the training phase, and (ii) an effective neural network architecture. Compared to previous implementations in the literature, we achieve an improvement of a factor of 5 in terms of accuracy, while training a considerably lower amount of neural networks. This results in a cheaper training procedure and a higher computational performance. Finally, we show that our emulators can recover unbiased posteriors when analysing synthetic Stage-IV galaxy survey data sets.

Simulated subacromial injection instruction improves accuracy and skill level: a model for musculoskeletal procedural training

BackgroundMusculoskeletal (MSK) complaints often present initially to primary care physicians; however, physicians may lack appropriate instruction in MSK procedures. Diagnostic and therapeutic injections are useful orthopedic tools, but inaccuracy leads to unnecessary costs and inadequate treatment. The authors hypothesized that trainees afforded the opportunity to practice on a cadaver versus those receiving visual-aided instruction on subacromial injections (SAI) will demonstrate differences in accuracy and technique.MethodsDuring Spring of the year 2022, 24 Internal Medicine and Family Medicine residents were randomly divided into control and intervention groups to participate in this interventional randomized cadaveric study. Each group received SAI instruction via lecture and video; the intervention group practiced on cadavers under mentored guidance. Subjects underwent a simulated patient encounter culminating in injection of latex dye into a cadaveric shoulder. Participants were evaluated based on a technique rubric, and accuracy of injections was assessed via cadaver dissection.ResultsTwenty-three of twenty-four participants had performed at least one MSK injection in practice, while only 2 (8.3%) of participants had performed more than 10 SAIs. There was no difference in technique between control 18.4 ± 3.65 and intervention 19.2 ± 2.33 (p = 0.54). Dissections revealed 3 (25.0%) of control versus 8 (66.7%) of intervention injections were within the subacromial space. Chi-Square Analysis revealed that the intervention affected the number of injections that were within the subacromial space, in the tissues bordering the subacromial space, and completely outside the subacromial space and bordering tissues (p = 0.03). The intervention group had higher self-confidence in their injection as opposed to controls (p = 0.04). Previous SAI experience did not affect accuracy (p = 0.76).ConclusionsAlthough primary care physicians and surgeons develop experience with MSK procedures in practice, this study demonstrates a role for early integrated instruction and simulation to improve accuracy and confidence. The goal of improving accuracy in MSK procedures amongst all primary care physicians may decrease costs and avoid unnecessary referrals, diagnostic tests, and earlier than desired surgical intervention.

Awareness of disaster preparedness between administrative staff and residents in the vicinity of the Genkai and Ikata nuclear power plants following the Fukushima Daiichi nuclear power plant disaster.

When considering disaster preparedness, one challenge is mitigating the health impacts of evacuations. Nuclear disaster preparedness has evolved based on past experiences from numerous disasters, including the Fukushima Daiichi Nuclear Power Plant (FDNPP) accident. However, there is a lack of comprehensive reporting on the awareness of administrative staff, medical personnel, and residents in the areas surrounding nuclear power plants (NPPs). This study reports on a survey aimed at gaining insights into the understanding and current state of disaster preparedness and elucidating the differences in perceptions of nuclear disaster preparedness among the relevant stakeholders surrounding NPPs. Interview surveys were conducted from 14 to 16 September 2022 in the area surrounding Kyushu Electric Power's Genkai NPP in Saga Prefecture and from 11 to 13 January 2023 in the area around Shikoku Electric Power's Ikata NPP. The surveys targeted administrative, medical, and nursing care facilities and residents. Responses from 57 participants indicated a lack of awareness of natural and nuclear disasters, challenges in evacuation planning, and a gap between nuclear disaster training and residents' understanding of evacuation protocols. This study highlights inadequacies in nuclear disaster preparedness and the need for a better understanding among residents regarding evacuation procedures. This study identified three key issues: (1) a lack of awareness about disasters, including nuclear disasters; (2) concerns about complex disasters and the difficulties in creating evacuation plans; and (3) a discrepancy between nuclear disaster training and residents' understanding of evacuation procedures. To bridge this gap, it is important to deepen residents' understanding of nuclear disasters, continuously convey the lessons learned from the FDNPP accident, and regularly reassess and update nuclear disaster preparedness strategies.

Study on the relationship between fixation characteristics and hit rate in psychological procedure training of free throw.

Free throw is an important means of scoring in basketball games. With the improvement of basketball competition level and the enhancement of confrontation degree, the number of free throws in the game gradually increases, so the score of free throw will have an important impact on the result of the game. The purpose of this study is to explore the relationship between visual attention characteristics and hit rate of basketball players in free throw psychological procedure training, so as to provide scientific basis for basketball teaching and training. Forty players with similar free throw abilities were randomly assigned to the experimental group (10 males, 10 females) and control group (10 males, 10 females). The experimental group was free throw psychological procedure training, while the control group was trained with routine training, Eye movement indices (number of fixations, fixation duration, and pupil dilation) and the free throw hit rate and analyzed before and after the experiment. Group differences were examined using t-tests, while paired sample t-tests were conducted to compare pre- and post-test results within each group. The training time and training times of the two groups were the same. There were significant differences in fixation duration, number of fixations, pupil diameter and free throw hit rate between pre-test and post-test in the experimental group (P < 0.05). Post-test, there were significant differences in number of fixations, fixation duration, pupil diameter and free throw hit rate between the two groups (P < 0.05). There was a significant positive correlation between number of fixations and free throw hit rate in top (P < 0.01), and there was a significant positive correlation between fixation duration and hit rate in front (P < 0.01). The psychological procedure training can improve the visual information search strategy and information processing ability of free throw, and significantly improve the free throw hit rate. There was a positive correlation between the front fixation time and the free throw hit rate, and there was a positive correlation between the top number of fixations and the free throw hit rate.

Feasibility and acceptability of measuring positivity resonance in nurse-patient telehealth videoconferencing visits: A mixed-methods observational study.

To investigate the feasibility and acceptability of the training process, procedures, measures and recruitment strategies necessary for a future investigation to test the reliability and validity of using positivity resonance measures in health care encounters. Although the measurement of positivity resonance is promising, and non-participant observation is considered effective, their approaches to studying nurse-patient relationships have not been fully explored. A mixed-methods observational study. Video recordings of 30 nurse-patient dyads completing telehealth video visit encounters were edited and coded using behavioural indicators of positivity resonance. A post-visit survey gathered data on the participants' perceptions of positivity resonance and the study procedures. The research team completed memos and procedural logs to provide narrative data on the study's training, coding, recruitment and operational procedures. The study included 33 persons with cancer and 13 oncology nurses engaging in telehealth video visit encounters at an academic oncology ambulatory care center located in the southeastern United States. Study procedures were found to be feasible and acceptable to participants. An adequate sample of participants (N = 46) were enrolled and retained in the study. Interrater reliability, as evidenced by Cohen's weighted kappa, ranged from .575 to .752 and interclass correlation coefficients >.8 were attainable within a reasonable amount of time and with adequate training. Behavioural indicators of positivity resonance were observed in all telehealth visits and reported by the participants in the perceived positivity resonance survey. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) checklist guided reporting. Designing research around the concept of positivity resonance is an innovative and feasible approach to exploring how rapport is cultivated within nurse-patient relationships. Measuring positivity resonance may hold promise for exploring patient and nurse outcomes including trust, responsiveness, health-related behaviours, well-being, resilience and satisfaction. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) checklist guided the reporting of results to ensure that adequate details of the study were provided to ensure an accurate and complete report. Planning of the research design and study procedures was done in consultation with nurse clinicians with experience with telehealth and managers responsible within the practice setting where the study was conducted. This ensured the study procedures were ethical, safe, secure and did not create unnecessary burden to the study participants. The study included collecting data from nurse and patient participants about the acceptability of the study procedures.

Semi-supervised multi-modal medical image segmentation with unified translation

The two major challenges to deep-learning-based medical image segmentation are multi-modality and a lack of expert annotations. Existing semi-supervised segmentation models can mitigate the problem of insufficient annotations by utilizing a small amount of labeled data. However, most of these models are limited to single-modal data and cannot exploit the complementary information from multi-modal medical images. A few semi-supervised multi-modal models have been proposed recently, but they have rigid structures and require additional training steps for each modality. In this work, we propose a novel flexible method, semi-supervised multi-modal medical image segmentation with unified translation (SMSUT), and a unique semi-supervised procedure that can leverage multi-modal information to improve the semi-supervised segmentation performance. Our architecture capitalizes on unified translation to extract complementary information from multi-modal data which compels the network to focus on the disparities and salient features among each modality. Furthermore, we impose constraints on the model at both pixel and feature levels, to cope with the lack of annotation information and the diverse representations within semi-supervised multi-modal data. We introduce a novel training procedure tailored for semi-supervised multi-modal medical image analysis, by integrating the concept of conditional translation. Our method has a remarkable ability for seamless adaptation to varying numbers of distinct modalities in the training data. Experiments show that our model exceeds the semi-supervised segmentation counterparts in the public datasets which proves our network’s high-performance capabilities and the transferability of our proposed method. The code of our method will be openly available at https://github.com/Sue1347/SMSUT-MedicalImgSegmentation.

A multiscale and multicriteria generative adversarial network to synthesize 1-dimensional turbulent fields

This article introduces a new neural network stochastic model to generate a 1-dimensional stochastic field with turbulent velocity statistics. Both the model architecture and training procedure ground on the Kolmogorov and Obukhov statistical theories of fully developed turbulence, so guaranteeing descriptions of (1) energy distribution, (2) energy cascade and (3) intermittency across scales in agreement with experimental observations. The model is a generative adversarial network (GAN) with multiple multiscale optimization criteria. First, we use three physics-based criteria: the variance, skewness and flatness of the increments of the generated field, that retrieve respectively the turbulent energy distribution, energy cascade and intermittency across scales. Second, the GAN criterion, based on reproducing statistical distributions, is used on segments of different length of the generated field. Furthermore, to mimic multiscale decompositions frequently used in turbulence’s studies, the model architecture is fully convolutional with kernel sizes varying along the multiple layers of the model. To train our model, we use turbulent velocity signals from grid turbulence at Modane wind tunnel.

Coverage-enhanced fault diagnosis for Deep Learning programs: A learning-based approach with hybrid metrics

Context:Given the data-driven paradigm inherent to Deep Learning (DL), it is inevitable that DL software will exhibit incorrect behavior in real-world applications. DL programs have been identified as a primary source of DL faults. To tackle this, researchers have devised a unique framework that approaches fault diagnosis as a learning task, which leverages runtime data as metrics to construct predictive models, enabling effective fault diagnosis. Object:In this paper, we aim to propose new metrics, especially from the coverage view, to enhance the performance of fault diagnosis models. Method:We combine coverage criteria and statistical operators to propose 80 coverage metrics, which summarize the trend of coverage values in the model training procedure. We construct hybrid prediction models by combining our new coverage metrics and existing runtime metrics under four widely used classifiers. Results:To examine whether adding our new coverage metrics performs well in DL program fault diagnosis, we conduct our experiments on six widely used datasets under four indicators (i.e., accuracy, F1 score, AUC, and MCC). Through the experiments, we observe that (a) coverage metrics are not redundant with respect to the original runtime metrics, and (b) adding extra coverage metrics can significantly enhance the performance of fault diagnosis models. Conclusions:Our study shows that our proposed coverage metrics are helpful in constructing effective fault diagnosis models for DL programs.

Melody Generation using Deep Learning: Unleashing the Power of RNN and LSTM

This project aims to develop a novel approach for piano melody generation using Recurrent Neural Networks (RNN) and Long Short-Term Memory (LSTM) models in deep learning.The suggested models will be trained on a dataset of MIDI files with piano melodies to use sequential learning capabilities and capture the complex patterns and relationships present in musical compositions. [1] The project aims to gen- erate a variety of melodies that are both musically coherent and diverse by experimenting with various network designs, hyperparameters, and training procedures. The developed tunes will be evaluated primarily on their originality, conformity to stylistic elements, and general quality. The results of this study could lead to new developments in AI-driven music composition as well as opportunities for computational creativity in the music industry.

Evolving testing scenario generation and intelligence evaluation for automated vehicles

Interaction between the background vehicles (BVs) and automated vehicles (AVs) in scenario-based testing plays a critical role in evaluating the intelligence of the AVs. Current testing scenarios typically employ predefined or scripted BVs, which inadequately reflect the complexity of human-like social behaviors in real-world driving scenarios, and also lack a systematic metric for evaluating the comprehensive intelligence of AVs. Therefore, this paper proposes an evolving scenario generation method, employing deep reinforcement learning (DRL) to construct human-like BVs that interact with AVs, and this evolving scenario is designed to test and evaluate the intelligence of AVs. Firstly, a class of BV driver models with human-like competitive, mutual, and cooperative driving motivations is designed. Then, utilizing the Twin Delayed Deep Deterministic Policy Gradient (TD3) algorithm and an improved level-k training procedure, the three distinct driver models acquire game-based interactive driving policies. And these driver models are combined to generate evolving scenarios in which they can interact continuously and evolve diverse contents. Next, a framework including safety, driving efficiency, and interaction utility are presented to evaluate and quantify the intelligence performance of 3 systems under test (SUTs), indicating the effectiveness of the evolving scenario for intelligence testing. Finally, the complexity and fidelity of the proposed evolving testing scenario are validated. The results demonstrate that the proposed evolving scenario exhibits the highest level of complexity compared to other baseline scenarios and has more than 85% similarity to naturalistic driving data. This highlights the potential of the proposed method to facilitate the development and evaluation of high-level AVs in a realistic and challenging environment.

Accelerate microstructure evolution simulation using graph neural networks with adaptive spatiotemporal resolution

Surrogate models driven by sizeable datasets and scientific machine-learning methods have emerged as an attractive microstructure simulation tool with the potential to deliver predictive microstructure evolution dynamics with huge savings in computational costs. Taking 2D and 3D grain growth simulations as an example, we present a completely overhauled computational framework based on graph neural networks with not only excellent agreement to both the ground truth phase-field methods and theoretical predictions, but enhanced accuracy and efficiency compared to previous works based on convolutional neural networks. These improvements can be attributed to the graph representation, both improved predictive power and a more flexible data structure amenable to adaptive mesh refinement. As the simulated microstructures coarsen, our method can adaptively adopt remeshed grids and larger timesteps to achieve further speedup. The data-to-model pipeline with training procedures together with the source codes are provided.

Acute lymphocytic leukemia detection and subtype classification via extended wavelet pooling based-CNNs and statistical-texture features

Acute lymphoblastic leukemia (ALL) is considered the most fatal form of leukemia (also known as blood cancer). It propagates quickly among adults and children and could lead to their death. Early detection of ALL and ALL subtypes is the key factor in selecting effective treatment types and improving survival rates. However, routine diagnostic approaches have several drawbacks. Computer-assisted diagnosis (CAD) is the perfect solution to avoid these challenges and achieve a fast and accurate diagnosis. Current CAD models require enhancement/segmentation processing. Besides, they are either dependent on deep learning (DL) models or handcrafted features along with machine learning. Those CADs that employed DL approaches relied solely on spatial information during the training procedure. However, learning them with spectral temporal and temporal representations could improve performance. Furthermore, integrating deep features from DL models along with handcrafted features can increase the discrimination ability of attributes in medical image classification. This study aims to propose a novel CAD for ALL detection and subtype classification without pre-segmentation or enhancement steps. The proposed CAD extends the conventional DL models of convolutional neural networks by introducing an additional wavelet pooling, accompanied by a dense layer or a long-short-term memory (LSTM) layer, and then a SoftMax layer, acquiring spectral-temporal information along with temporal information. To further improve the framework's ability to discriminate, the introduced CAD then combines the wavelet-based deep features of every CNN with numerous handcrafted attributes. Afterward, a feature selection methodology is utilized to create a model with limited features and improved accuracy. The performance results show that the novel CAD is capable of achieving 100% ALL detection accuracy, as well as 100% ALL-subtype classification accuracy with just 88 and 146 features. Thus, this CAD can be employed to assist pathologists in the rapid and precise ALL identification and subcategories recognition.

Abstract PO4-25-12: Improving HER2 Low Scoring Consistency and Accuracy: Insights from the Australian HER2 Low Concordance Study for Invasive Breast Cancer

Abstract Introduction: The Destiny Breast-04 Trial (DB-04) demonstrated the survival benefits of Trastuzumab Deruxtecan (T-DXd) for women with metastatic HER2 Low breast cancer, characterised by 1+ or 2+ IHC staining without amplification. While the DB-04 study applied the standard 2018 ASCO CAP IHC scoring criteria, in clinical practice, distinguishing HER2 0 from 1+ cancers is challenging as i) HER2 Low is not a biologically distinct subset of breast cancer, ii) there are no reference standards for HER2 Low cancers, iii) second-tier test, like ISH, are not applicable, and iv) there are no known controls for cases that have 0 or 1+ HER2 scores. For two decades this distinction was clinically immaterial, but now differentiating between HER2 0 and 1+ has now become crucial for determining patient eligibility for T-DXd therapy. Concerns regarding the subjectivity, imprecision and poor concordance between pathologists in scoring IHC in HER2 Low cancers raise the potential for misalignments in patient treatment. Ensuring pathologists have access to focused training for interpreting IHC scores at the low end of the HER2 expression spectrum, quality assurance procedures and reference sets are essential steps to help pathologists assess HER2 Low breast cancers more accurately and consistently. Design: In this study, a group of 9 experienced breast pathologists compiled a deidentified set of 60 breast cancer core biopsies from 3 laboratories. The Ventana 4B5 HER2 assay had been used for evaluation and the local laboratories had scored the samples as HER2 0 or 1+. We teased out the ASCO CAP 2018 criteria and used our collective expertise of reporting HER2 IHC for many years to specify HER2 Low-focused scoring conventions, including some potential pitfalls. Subsequently, using these conventions, each pathologist reviewed digitized whole slide images of the IHC slides and scored HER2 expression for each case. At a consensus workshop, the cases were jointly reviewed to establish consensus scores and determine the percentage of HER2-expressing tumor cells in each case. We then evaluated the concordance between individual pathologists' HER2 scores and the consensus opinion and ascertained reasons for discordance. Results: Among the cases discussed during the consensus conference, 43 out of 60 (71.7%) were classified as HER2 Low, with 40 cases designated as 1+ and three as 2+ (known to be not amplified). The consensus score matched the majority opinion of the pathologists' independent scores in 93.3% (56 out of 60) of the cases. Utilizing the HER2 Low-focused IHC scoring conventions, 7 out of 17 (41.2%) cases locally reported as HER2 0 were reclassified as HER2 Low. Conversely, among the 32 cases with local scores of 1+, 7 (21.8%) were reclassified as ultralow or null. When compared to the consensus score, individual pathologists' scores demonstrated concordance levels ranging from 71.7% to 91.7%, with a mean concordance rate of 81.3%. Cases with less than 20% of tumor cells expressing HER2 had lower inter observer concordance. This reference set of cases with expert consensus HER2 scores obtained through our study will be invaluable for peer training and the development of external quality assurance programs for HER2 Low cancers, including the quality assurance program of the Royal College of Pathologists of Australasia. Conclusion: This study revealed that when breast pathologists were provided explicit instructions on scoring pitfalls and HER2 Low-focused scoring conventions, their HER2 scores were concordant with expert consensus scores in 71.7% to 91.7% of cases. Discordant cases primarily involved cases with less than 20% of tumor cells expressing HER2. Utilising such an approach, peer training and quality assurance procedures will improve the accuracy and consistency of HER2 IHC assessment for better patient care. Reassessing older cases using HER2 Low focused scoring conventions may result in revisions of HER2 scores from HER2 Low to zero, and vice versa. Table. Individual pathologists' concordance with the consensus HER2 IHC score Applying our HER2 Low-focused IHC scoring conventions in a set of 60 core biopsies of invasive breast cancer with low or 0 HER2 protein expression. Citation Format: Gelareh Farshid, Beena Kumar, Nirmala Pathmanathan, Hema Mahajan, Ben Dessauvagie, Jane Armes, Cameron Snell, Amardeep Gilhotra. Improving HER2 Low Scoring Consistency and Accuracy: Insights from the Australian HER2 Low Concordance Study for Invasive Breast Cancer [abstract]. In: Proceedings of the 2023 San Antonio Breast Cancer Symposium; 2023 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2024;84(9 Suppl):Abstract nr PO4-25-12.

Cutting-edge tool wear monitoring in AISI4140 steel hard turning using least square-support vector machine

ABSTRACT This research addresses the critical challenge of tool wear monitoring in AISI4140 steel hard turning through the innovative application of a Least Square-Support Vector Machine (LS-SVM) prediction model. While acknowledging the growing significance of hard turning over conventional grinding techniques, the background also recognizes the challenges posed by accelerated tool wear and decreased productivity. The main goal is to create a reliable tool wear prediction model that is tailored to harsh turning circumstances. Material selection, cutting force measurement, acceleration measurement, and accurate tool wear measurement are all included in the experimental procedures. Using time and frequency domain analyses, the research methodically examines how cutting force and acceleration affect tool wear. For feature selection, non-linear regression is used to determine which parameters have the greatest influence on wear. A comprehensive training and validation procedure, structural clarification, and a detailed mathematical foundation are used in the development of the LS-SVM-based tool wear prediction model. The presentation of the findings and related discussions explores the impact of acceleration and cutting force on wear. The LS-SVM model’s successful application for precise tool wear prediction in AISI4140 steel hard turning is highlighted in the conclusions, demonstrating the model’s potential to improve tool lifespan and manufacturing efficiency.

Applying the master adaptive learner framework to just‐in‐time training of procedures

AbstractBackgroundJust‐in‐time training (JITT) occurs in the clinical context when learners need immediate guidance for procedures due to a lack of proficiency or the need for knowledge refreshment. The master adaptive learner (MAL) framework presents a comprehensive model of transforming learners into adaptive experts, proficient not only in their current tasks but also in the ongoing development of lifelong skills. With the evolving landscape of procedural competence in emergency medicine (EM), trainees must develop the capacity to acquire and master new techniques consistently. This concept paper will discuss using JITT to support the development of MALs in the emergency department.MethodsIn May 2023, an expert panel from the Society for Academic Emergency Medicine (SAEM) Medical Educator's Boot Camp delivered a comprehensive half‐day preconference session entitled “Be the Best Teacher” at the society's annual meeting. A subgroup within this panel focused on applying the MAL framework to JITT. This subgroup collaboratively developed a practical guide that underwent iterative review and refinement.ResultsThe MAL‐JITT framework integrates the learner's past experiences with the educator's proficiency, allowing the educational experience to address the unique requirements of each case. We outline a structured five‐step process for applying JITT, utilizing the lumbar puncture procedure as an example of integrating the MAL stages of planning, learning, assessing, and adjusting. This innovative approach facilitates prompt procedural competence and cultivates a positive learning environment that fosters acquiring adaptable learning skills with enduring benefits throughout the learner's career trajectory.ConclusionsJITT for procedures holds the potential to cultivate a dynamic learning environment conducive to nurturing the development of MALs in EM.

Weight Decay With Tailored Adam on Scale-Invariant Weights for Better Generalization.

Weight decay (WD) is a fundamental and practical regularization technique in improving generalization of current deep learning models. However, it is observed that the WD does not work effectively for an adaptive optimization algorithm (such as Adam), as it works for SGD. Specifically, the solution found by Adam with the WD often generalizes unsatisfactorily. Though efforts have been made to mitigate this issue, the reason for such deficiency is still vague. In this article, we first show that when using the Adam optimizer, the weight norm increases very fast along with the training procedure, which is in contrast to SGD where the weight norm increases relatively slower and tends to converge. The fast increase of weight norm is adverse to WD; in consequence, the Adam optimizer will lose efficacy in finding solution that generalizes well. To resolve this problem, we propose to tailor Adam by introducing a regularization term on the adaptive learning rate, such that it is friendly to WD. Meanwhile, we introduce first moment on the WD to further enhance the regularization effect. We show that the proposed method is able to find solution with small norm and generalizes better than SGD. We test the proposed method on general image classification and fine-grained image classification tasks with different networks. Experimental results on all these cases substantiate the effectiveness of the proposed method in help improving the generalization. Specifically, the proposed method improves the test accuracy of Adam by a large margin and even improves the performance of SGD by 0.84% on CIFAR 10 and 1.03 % on CIFAR 100 with ResNet-50. The code of this article is public available at xxx.

Hand Motion Analysis Using Accelerometer-Based Sensors and Sheep's Head Model for Basic Training in Functional Endoscopic Sinus Surgery.

Motion analysis, the study of movement patterns to evaluate performance, plays a crucial role in surgical training. It provides objective data that canbe usedto assess and improve trainee's precision, efficiency, and overall surgical technique. The primary aim of this study is to employ accelerometer-based sensors placed on the wrist to analyze hand motions during endoscopic sinus surgery training using the sheep's head. By capturing detailed movement data, the study seeks to quantify the motion characteristics that distinguish different levels of surgical expertise. This approach seeks to quantify motion characteristics indicative of surgical expertise and enhance the objectivity and effectiveness of surgical training feedback mechanisms. Twenty-four participantswere dividedinto three groups based on their experience with endoscopic endonasal surgery. Each participantwas tasked with performingspecified procedures on an individual sheep's head, concentrating on exploring both nasal passages. A single Bluetooth Accelerometer WitMotion sensorwas mountedon the dorsal surface of each hand.Thisfacilitates the evaluation of efficiency parameters such as time, path length, and acceleration during the training procedures. Accelerometer data were collected and imported in CSV format (comma-separated values) for each group of surgeons-senior, specialist, and resident-mean values and standard deviationswere computed. The Shapiro-Wilk Test assessed the normality of the distribution. The Kruskal-Wallis test was employed to compare procedural time, acceleration, and path length differences across the three surgeon experience levels. For the procedural time, statistical significance appears in all surgical steps (p<0.001), with thebiggestdifference in the septoplasty group in favor of the senior group. A clear difference can be observed between the resulting acceleration of the dominant hands (instrument hand) and the non-dominant hand (endoscopic hand) and between the study groups. The difference between groups reaches statistical significance with a p-value <0.001. A statistically significant difference canbe seenbetween the paths covered by each hand of every participant (p<0.001). Also, senior doctors covered significantly less movement with both hands than the specialists and the resident doctors (p<0.001). The data showa clearlearning curve from resident to senior,with residents taking more time and using more hand movementsto complete the same tasks. Specialists are in the intermediate phase, showing signs of honing their technique towards efficiency. This comprehensive data set can help tailor training programs to focus onbothefficiency (quicker procedures) and economy of motion (reduced path length and acceleration), especially in more complex procedures where the difference in performance ismore pronounced.

FXAM: A unified and fast interpretable model for predictive analytics

Predictive analytics aims to build machine learning models to predict behavior patterns and use predictions to guide decision-making. Predictive analytics is human involved, thus the machine learning model is preferred to be interpretable. In literature, Generalized Additive Model (GAM) is a standard for interpretability. However, due to the one-to-many and many-to-one phenomena which appear commonly in real-world scenarios, existing GAMs have limitations to serve predictive analytics in terms of both accuracy and training efficiency. In this paper, we propose FXAM (Fast and eXplainable Additive Model), a unified and fast interpretable model for predictive analytics. FXAM extends GAM’s modeling capability with a unified additive model for numerical, categorical, and temporal features. FXAM conducts a novel training procedure called Three-Stage Iteration (TSI). TSI corresponds to learning over numerical, categorical, and temporal features respectively. Each stage learns a local optimum by fixing the parameters of other stages. We design joint learning over categorical features and partial learning over temporal features to achieve high accuracy and training efficiency. We prove that TSI is guaranteed to converge to the global optimum. We further propose a set of optimization techniques to speed up FXAM’s training algorithm to meet the needs of interactive analysis. Thorough evaluations conducted on diverse data sets verify that FXAM significantly outperforms existing GAMs in terms of training speed, and modeling categorical and temporal features. In terms of interpretability, we compare FXAM with the typical post-hoc approach XGBoost+SHAP on two real-world scenarios, which shows the superiority of FXAM’s inherent interpretability for predictive analytics.

Simulation training of intraoperative complication management in laparoscopic cholecystectomy for novices—A randomized controlled study

BackgroundIntraoperative complications occur more often in laparoscopic cholecystectomies (LCs) than in open cholecystectomies. Since LC is a frequent training procedure, surgical residents should be trained to deal with intraoperative complications in a safe and controlled environment. This study aimed to determine whether structured training in coping mechanisms in complication management in LC benefits novices compared to no training in complication management. MethodsThe study included sixty medical students without prior experience in minimally invasive surgery. They were randomly assigned to the intervention or control group and worked in teams of two. Both groups completed multimodal skill training, including one ex-vivo porcine LC. The intervention group (n=30) was then informed about the complications that can occur during an LC and how to manage them using the Coping Mechanisms for Intraoperative Complication Management in Laparoscopic Cholecystectomy (COMIC-LC) tool specially developed for this study. Afterward, they performed two LCs receiving structured training in coping mechanisms for intraoperative complication management. The control group (n=30) performed two LCs, receiving no information about managing complications. Both groups performed two postinterventional LCs. Complication management of the two groups was evaluated using the COMIC-LC assessment score. ResultsThe intervention group demonstrated better management of intraoperative complications in LCs compared to the control group (gallbladder perforation: 8.5 vs. 3.0 points, p<0.001; injury to the liver parenchyma: 7.0 vs. 2.0 points, p<0.001; cystic artery injury: 8.0 vs. 5.0 points, p<0.001; cystic duct injury: 6.0 vs. 1.0, p<0.001). ConclusionStructured training in coping mechanisms for managing intraoperative complications during LC can improve complication handling and potentially reduce adverse events. Translation to clinical practice should be evaluated.