Multimodal Training Research Articles

Adherence and potential factors of adherence to a resistance, coordination and endurance training in older retirement home residents over 6 months.

High exercise adherence is a key factor for effective exercise programmes. However, little is known about predictors of exercise adherence to a multimodal machine-based training in older retirement home residents. To assess exercise adherence and potential predictors of adherence. Furthermore, to evaluate user acceptance of the multimodal training and the change in exercise self-efficacy. In this sub-analysis of the bestform-F study, a total of 77 retirement home residents ≥65 years (mean age: 85.6 ± 6.6 years, 77.9% female) participated in a 6-month machine-based resistance, coordination and endurance training. Attendance to the training was documented for each training session. To identify potential predictors a multiple linear regression model was fitted to the data. Analyzed predictors included age, sex, body mass index (BMI), physical function, exercise self-efficacy, and physical activity history. Different domains of user acceptance (e.g. safety aspects, infrastructure) and exercise self-efficacy were assessed by a questionnaire and the exercise self-efficacy scale (ESES), respectively. Mean exercise adherence was 67.2% (median: 74.4%). The regression model (R2 = 0.225, p = 0.033) revealed that the 6-minute walk test (6-MWT) at baseline significantly predicted exercise adherence (β: 0.074, 95% confidence interval (CI): 0.006-0.142, p = 0.033). Different user domains were rated at least as good by 83.9%-96.9% of participants, reflecting high acceptance. No statistically significant change was found for exercise self-efficacy over 6 months (mean change: 0.47 ± 3.08 points, p = 0.156). Retirement home residents attended more than two thirds of offered training sessions and physical function at baseline was the key factor for predicting adherence. User acceptance of the training devices was highly rated. These findings indicate good potential for implementation of the exercise programme.

Incomplete Cross-Modal Retrieval with Deep Correlation Transfer

Most cross-modal retrieval methods assume the multi-modal training data is complete and has a one-to-one correspondence. However, in the real world, multi-modal data generally suffers from missing modality information due to the uncertainty of data collection and storage processes, which limits the practical application of existing cross-modal retrieval methods. Although some solutions have been proposed to generate the missing modality data using a single pseudo sample, this may lead to incomplete semantic restoration and sub-optimal retrieval results due to the limited semantic information it provides. To address this challenge, this article proposes an Incomplete Cross-Modal Retrieval with Deep Correlation Transfer (ICMR-DCT) method that can robustly model incomplete multi-modal data and dynamically capture the adjacency semantic correlation for cross-modal retrieval. Specifically, we construct intra-modal graph attention-based auto-encoder to learn modality-invariant representations by performing semantic reconstruction through intra-modality adjacency correlation mining. Then, we design dual cross-modal alignment constraints to project multi-modal representations into a common semantic space, thus bridging the heterogeneous modality gap and enhancing the discriminability of the common representation. We further introduce semantic preservation to enhance adjacency semantic information and achieve cross-modal semantic correlation. Moreover, we propose a nearest-neighbor weighting integration strategy with cross-modal correlation transfer to generate the missing modality data according to inter-modality mapping relations and adjacency correlations between each sample and its neighbors, which improves the robustness of our method against incomplete multi-modal training data. Extensive experiments on three widely tested benchmark datasets demonstrate the superior performance of our method in cross-modal retrieval tasks under both complete and incomplete retrieval scenarios. Our used datasets and source codes are available at https://github.com/shidan0122/DCT.git .

In-Theatre Simulation as a Training Tool for Laparoscopic Salpingectomy in Eldoret, Kenya

Background/Objective: Minimally invasive surgery (MIS) offers many advantages over open procedures including decreased patient safety risks and reduced burden on healthcare infrastructure. As low- and middle-income countries (LMICs) are disproportionately affected by these aspects of surgery, there is motivation to increase MIS. A multimodal training program in laparoscopic salpingectomies was piloted with a small cohort of OB-GYN registrars and consultants at Moi Teaching and Referral Hospital (MTRH) in Eldoret, Kenya. This project assesses the in-theatre simulation’s (1) effectiveness in improving laparoscopic knowledge and skill confidence, and (2) feasibility for long-term implementation at MTRH and in similar settings.
 Methods: Participants completed a half-day in-theatre simulation of a laparoscopic salpingectomy. The simulation required participants to demonstrate knowledge of laparoscopic setup, proper patient positioning, procedure completion, equipment troubleshooting, and peri-and intra-operative complication management. Participants completed a multiple-choice laparoscopic knowledge quiz and Likert scale skill confidence survey immediately prior to and following the simulation. Pre- and post-simulation responses were compared to assess knowledge and confidence acquisition overall and across content topics.
 Results: There was a significant increase in the average knowledge quiz score from pre- to posttest (p=0.028). A significant difference between pre- and posttest confidence was noted in four of the six skills assessed. By topic, equipment troubleshooting (p<0.001), and complication management (p<0.01) saw the most improvement. Barriers to long-term sustainability include unpredictable theatre and laparoscopic tower access and availability of supplies for uterine modeling. A modified model using nitrile gloves as fallopian tubes will be piloted in future simulations as a more accessible alternative for long-term implementation.
 Conclusion: Despite limitations, in-theatre simulation has the potential to be an effective and sustainable teaching tool within a long-term MIS training program at Moi Teaching and Referral Hospital. The low-cost model and methods outlined may also be replicable in similar low-resource settings.

Impact of exercise modalities on blood pressure control in hypertensive patients

Hypertension poses a significant health burden, necessitating effective interventions to manage blood pressure and mitigate associated cardiovascular risks. Exercise has emerged as a cornerstone non-pharmacological strategy in the management of hypertension. This comprehensive review synthesizes current evidence regarding the impact of various exercise modalities on blood pressure regulation in hypertensive individuals. The review commences by elucidating the physiological mechanisms underpinning exercise-induced blood pressure alterations, highlighting factors influencing blood pressure responses during different exercises. It delineates the differential impacts of aerobic and resistance training, examining their effectiveness in blood pressure reduction among diverse populations, considering age, fitness levels, and comorbidities. Moreover, the review explores the emerging prominence of high-intensity interval training (HIIT), showcasing its unique effects on blood pressure regulation and cardiorespiratory fitness improvement compared to traditional continuous aerobic exercises. Additionally, alternative mind-body exercises, including tai chi, qigong, and mindfulness-based interventions, are investigated for their potential to influence blood pressure levels among hypertensive populations. Furthermore, the review identifies novel exercise approaches such as deep breathing exercises and multi-modal isometric exercise training (MIET) as promising strategies for blood pressure control in hypertensive individuals, emphasizing their potential benefits and implications for tailored exercise prescriptions.

Nurse managers' perceptions of the prospective acceptability of an implementation leadership training programme: A qualitative descriptive study.

To explore the prospective acceptability of an implementation leadership training programme prototype for nurse managers in China to implement evidence-based practices, from the perspectives of potential programme participants and deliverers. A qualitative descriptive study was conducted in Spring 2022 at three tertiary hospitals in Hunan, China. We conducted individual semi-structured interviews with unit-level nurse managers (n = 14), including 12 potential participants, and two potential deliverers that have been involved in developing the programme prototype. Interview questions and thematic analysis were guided by the Theoretical Framework of Acceptability. After reviewing the programme content, potential participants and deliverers reported that unit nurse managers would benefit from engaging in the programme, acknowledging that the programme fit with professional nursing values for implementing research evidence. They expressed positive views about being involved in producing academic papers through the training process, and interactive multi-modal training activities such as group work, experience-sharing and coaching. Seven participants were not very confident about being fully engaged in the training, as they could not navigate the English research literature. Both participants and deliverers highlighted factors that would influence their participation, including time constraints, the impact of the COVID-19 pandemic, and support from senior organizational leadership. The training programme prototype was perceived to be useful and acceptable. The multimodal training activities were considered a strength and managers expressed an interest in writing academic papers about their implementation processes. Support from senior hospital leaders and programme deliverers was identified as critical to the training programme's success. The study helps understand nurse managers' perceptions and concerns of participating in an implementation leadership training programme and could inform the development and refinement of similar programmes in various nursing contexts globally.

A multimodal approach to data analysis in astronomy: sonouno applications in photometry and spectroscopy

ABSTRACT In the past decade, multimodal data analysis has gained importance, especially for including individuals with visual impairments in education and science dissemination. However, its application in scientific research is still limited due to a lack of conclusive evidence on its robustness and performance. Various sonification tools have been developed, including xsonify, starsound, strauss, and sonouno, which aim to enhance accessibility for both sighted and visually impaired users. This contribution presents sonouno (a data visualization and sonification tool) using data, and comparing to corresponding visuals displays, from established data bases like SDSS, ASAS-SN, and Project Clea for astronomical data. We show that sonouno is able to replicate the visual data displays and provide consistent auditory representations. Key features include marking absorption and emission lines (in both visual and sonification) and multicolumn sonification, which facilitates spectral comparisons through sound. This approach ensures consistency between visual and auditory data, making multimodal displays more viable for use in research, enabling greater inclusion in astronomical investigation. The study suggests that sonouno could be broadly adopted in scientific research and used to develop multimodal training courses and improve data analysis methods.

Enhanced Parkinson's gait, reduced fall risk, and improved cognitive function through multimodal rehabilitation combined with rivastigmine treatment.

This study aimed to examine the effects of combined rehabilitation and rivastigmine treatment on patients with Parkinson's disease (PD). Gait parameters were assessed using the Gibbon Gait Analyzer in fifteen patients. Baseline gait data and cognitive assessments were collected. Each patient underwent external counterpulsation therapy, transcranial magnetic stimulation therapy, and exercise therapy for one hour per day, five days a week for three weeks. Post-intervention, gait and cognitive data were re-evaluated. Alongside their standard PD medications, all participants were administered rivastigmine throughout the study period. The intervention significantly enhanced motor function in the single-task test, evidenced by marked improvements in gait metrics such as stride width and walking speed, and a substantial reduction in fall risk. Cognitive function, assessed by mini-mental state examination and Montreal cognitive assessment, showed an improvement trend after the three-week intervention. Improvements in dual-task walking function were observed, although these changes did not reach statistical significance. Multimodal exercise training combined with rivastigmine treatment significantly improves certain gait parameters in the single-task test, enhances balance, and reduces the risk of falling in patients with PD. Cognitive function also demonstrated improvement.

Multi-modal mindfulness training and well-being for healthcare students

IntroductionMindfulness is a mental state attained through focusing awareness on the present with calm acceptance of thoughts, feelings, and bodily sensations. This study evaluated impact of mindfulness activities on well-being of pharmacy and other healthcare students. MethodsResearch participants completed pre- and post-intervention questionnaires evaluating multi-modal mindfulness interventions. Due to the pandemic, sessions led by a certified mindfulness instructor were offered live online and recorded, supplemented by a well-being mobile app and reflective discussion component. Four composite scales were administered to participants. The Brief Resilient Coping Scale (BRCS), Five Facet Mindfulness Questionnaire (FFMQ-15), Warwick-Edinburgh Mental Wellbeing Scale (WEMWBS), and Depression Anxiety Stress Scales (DASS-21) inventories measured resilient coping skills, awareness, and psychological symptoms, including stress, depression, and anxiety. Exploratory factor analysis and Cronbach's alpha were used to determine scale reliability and validity. ResultsThirty-six participants completed both pre- and post-intervention questionnaires (n = 36). Significant improvement was found in well-being for BRCS, FFMQ-15, and WEMWBS measures (P < .05). Change in DASS-21 was not significant (P = .19). Mobile app use enhanced foundational mindfulness skills. Awareness, connection, and coping themes were identified from written comments. ConclusionsEvaluation of this multi-modal interprofessional intervention provides further evidence of benefits of mindfulness for pharmacy and other healthcare students. The mobile app and mindful movement with reflection improved all attributes measured by the composite scales. Further research may explore alternative multi-modal mindfulness interventions and incorporation into healthcare education curricula. Interprofessional collaboration is encouraged among faculty to enhance mindfulness while connecting healthcare professionals.

Youth Athletes' Self-Esteem: The Impact of Integrated Psychological Skills Training.

Study on the effect of psychological skills training on self-esteem (SE) in young athletes. 10 swimmers and 35 volleyball players, split into an intervention group (25) and a control group (18). The intervention entailed a 3-week psychological skills training program covering arousal management, breathing, relaxation, mental imagery, and self-talk. Multilevel growth curve analyses evaluated SE changes. The intervention group showed significant improvements in multiple SE dimensions-physical self-worth, fitness, athletic competence, strength, and body attractiveness but not general SE. This study provides initial evidence of a multimodal psychological skills training's effectiveness in enhancing young athletes' domain-specific SE. It highlights the role of domain-specific SE in young athletes' well-being. Future research should examine psychological and physiological correlations and assess the long-term SE development in adolescent athletes.

Bridging MRI Cross-Modality Synthesis and Multi-Contrast Super-Resolution by Fine-Grained Difference Learning.

In multi-modal magnetic resonance imaging (MRI), the tasks of imputing or reconstructing the target modality share a common obstacle: the accurate modeling of fine-grained inter-modal differences, which has been sparingly addressed in current literature. These differences stem from two sources: 1) spatial misalignment remaining after coarse registration and 2) structural distinction arising from modality-specific signal manifestations. This paper integrates the previously separate research trajectories of cross-modality synthesis (CMS) and multi-contrast super-resolution (MCSR) to address this pervasive challenge within a unified framework. Connected through generalized down-sampling ratios, this unification not only emphasizes their common goal in reducing structural differences, but also identifies the key task distinguishing MCSR from CMS: modeling the structural distinctions using the limited information from the misaligned target input. Specifically, we propose a composite network architecture with several key components: a label correction module to align the coordinates of multi-modal training pairs, a CMS module serving as the base model, an SR branch to handle target inputs, and a difference projection discriminator for structural distinction-centered adversarial training. When training the SR branch as the generator, the adversarial learning is enhanced with distinction-aware incremental modulation to ensure better-controlled generation. Moreover, the SR branch integrates deformable convolutions to address cross-modal spatial misalignment at the feature level. Experiments conducted on three public datasets demonstrate that our approach effectively balances structural accuracy and realism, exhibiting overall superiority in comprehensive evaluations for both tasks over current state-of-the-art approaches. The code is available at https://github.com/papshare/FGDL.

Serious game the Emotion Detectives helps to improve social–emotional skills of children with neurodevelopmental disorders

AbstractChildren with neurodevelopmental disorders often have social–emotional difficulties. Serious games can be used to support their social–emotional learning. This study investigated if 6–10‐year‐old children with autism spectrum condition, attention deficit hyperactivity disorder and developmental language disorder would improve their emotion discrimination skills by playing a serious game. Intervention (n = 30) and control groups (n = 20), comprising children with neurodevelopmental disorders, and a group of typically developing (TD) age peers (n = 106), completed six emotion discrimination tasks. ABA research design was applied in this study. Prior to the game intervention, children with neurodevelopmental disorders were significantly delayed in relation to their TD age peers in all tasks. After playing the game for two months, the children in the intervention group had significantly improved their emotion discrimination skills in four tasks, whereas the control group had significantly advanced in only one task without practice. Intervention gains were maintained at the one‐month follow‐up. Parents reported that the children in the intervention group somewhat improved their emotion recognition and behavioural skills in daily life in comparison to the control group. Implications for practice are discussed. Practitioner notesWhat is already known about this topic Children with neurodevelopmental disorders often exhibit various social–emotional symptoms. Within the field of social–emotional learning, many serious games have a narrow focus on specific skills, and they often only provide possibilities for observing target behaviours, not actual skill training. There are mixed results on the efficacy of game‐based interventions targeted at improving social–emotional skills. What this paper adds The Emotion Detectives game used in this study offers a well‐designed learning environment that provides adaptive learning and feedback systems. The participating children's social–emotional skills were assessed with versatile methods and via triangulation. Intervention with the Emotion Detectives game helped to decrease some emotion and behavioural symptoms of the players. Implications for practice This study provides information about the efficacy of a web‐based game that is freely available for supporting children's social–emotional learning. Improvements in emotion discrimination and recognition, as well as behavioural skills, can be gained by playing serious games. For optimal intervention outcomes, multimodal training should be utilized in serious games targeting social–emotional learning.

A Description of Exercise Programs Implemented During a Firefighter Training Academy.

The aim of the study is to thoroughly describe exercise programs that were performed during a firefighter training academy, including their effects on body composition and physical fitness. Fourteen recruits performed exercise programs during a 24-week training academy and several components of their body composition and physical fitness were measured. Moreover, their heart rate (HR) and session rating of perceived exertion (sRPE) were recorded during and after every training session, respectively. Fat-free mass, subcutaneous fat in the abdomen, cardiovascular fitness, flexibility, lower-body power, and muscular endurance significantly improved during the study ( P ≤ 0.05). The highest HR and sRPE responses were observed during aerobic training (AT) sessions. The average HR responses relative to maximum HR were similar during AT, high-intensity circuit training, multimodal high-intensity interval training, and high-intensity functional training; however, the sRPEs during high-intensity circuit training and multimodal high-intensity interval training were significantly lower than AT. Several effective exercise programs can be implemented during firefighter training academies.

AI‐based Digital Biomarkers for Mild Dementia Decision Support using Mobile Playful Exercises with High Adherence

AbstractBackgroundEarly detection of cognitive decline and monitoring of cognitive functioning in mild dementia are fundamental for timely adaptation of lifestyle and intervention strategies. The development of digital dementia biomarkers through playful exercises with high adherence rate was a key objective of the national project multimodAAL (no. FFG 875345). The results of a study on computer‐based cognitive and physical training (CCPT) in persons diagnosed with mild Alzheimer’s disease (PwAD) are presented.MethodTablet‐PC‐based intervention was applied within 6 months in Austria, engaging PwADs living at home by means of playful multimodal training and activation (n = 11; female N = 8, male N = 3; age M = 76.6 / SD = 9.2 years, MMSE score M = 21.50 / SD = 4.41). PwADs interacted with a prototypical version of the BRAINMEE app that included a suite of cognitive exercises (puzzle, pairs, text gap filling) based on audiovisual information. The playful training app was introduced and assisted by mobile care professionals with weekly visits, however, PwDs played alone between these visits.ResultPwADs applied training with high adherence, finalizing M = 72 (M = 32) digital exercises per day within the first (last) month of the study. Duration of using exercise type ‘outsider’(p = .028*) and ‘quiz’ (p = .001**), averaged about 2 week figures, each provided statistically significant correlations (Spearman) with MMSE test scores, as well as ‘spot‐the‐difference’ (p = .003**) with Trail Making Test A, ‘outsider’ (p = .005**) with Auditory Verbal Learning Test (AVLT), respectively. A neural network (Support Vector Machine, linear kernel, 11‐fold cross validation) using duration of use of ‘quiz’, ’outsider’ and ‘hearing’ (guessing animal sounds) as input data resulted in M = 2.16 absolute error in MMSE score estimation on test data.ConclusionThe work outlined within the Austrian study on digital biomarker development indicates successful steps towards daily use of cognitive assessment using highly adherent playful training. The playful training app is applied in the European project MARA (no. FFG 886427) to enable continuous estimates of MCI’s mental state over time. The app was very well accepted by both PwADs and persons with MCI. It offers with its pervasive mental assessment tool a large potential for future long‐term monitoring in dementia prevention, early detection as well as in numerous dementia care services.

Deep learning in computed tomography super resolution using multi-modality data training.

One of the limitations in leveraging the potential of artificial intelligence in X-ray imaging is the limited availability of annotated training data. As X-ray and CT shares similar imaging physics, one could achieve cross-domain data sharing, so to generate labeled synthetic X-ray images from annotated CT volumes as digitally reconstructed radiographs (DRRs). To account for the lower resolution of CT and the CT-generated DRRs as compared to the real X-ray images, we propose the use of super-resolution (SR) techniques to enhance the CT resolution before DRRgeneration. As spatial resolution can be defined by the modulation transfer function kernel in CT physics, we propose to train a SR network using paired low-resolution (LR) and high-resolution (HR) images by varying the kernel's shape and cutoff frequency. This is different to previous deep learning-based SR techniques on RGB and medical images which focused on refining the sampling grid. Instead of generating LR images by bicubic interpolation, we aim to create realistic multi-detector CT (MDCT) like LR images from HR cone-beam CT (CBCT)scans. We propose and evaluate the use of a SR U-Net for the mapping between LR and HR CBCT image slices. We reconstructed paired LR and HR training volumes from the same CT scans with small in-plane sampling grid size of . We used the residual U-Net architecture to train two models. SRUN : trained with kernel-based LR images, and SRUN : trained with bicubic downsampled data as baseline. Both models are trained on one CBCT dataset (n = 13 391). The performance of both models was then evaluated on unseen kernel-based and interpolation-based LR CBCT images (n = 10 950), and also on MDCT images (n = 1392). Five-fold cross validation and ablation study were performed to find the optimal hyperparameters. Both SRUN and SRUN models show significant improvements (p-value 0.05) in mean absolute error (MAE), peak signal-to-noise ratio (PSNR) and structural similarity index measures (SSIMs) on unseen CBCT images. Also, the improvement percentages in MAE, PSNR, and SSIM by SRUN is larger than SRUN . For SRUN , MAE is reduced by 14%, and PSNR and SSIMs increased by 6 and 8%, respectively. To conclude, SRUN outperforms SRUN , which the former generates sharper images when tested with kernel-based LR CBCT images as well as cross-modality LR MDCTdata. Our proposed method showed better performance than the baseline interpolation approach on unseen LR CBCT. We showed that the frequency behavior of the used data is important for learning the SR features. Additionally, we showed cross-modality resolution improvements to LR MDCT images. Our approach is, therefore, a first and essential step in enabling realistic high spatial resolution CT-generated DRRs for deep learningtraining.

Unlocking Potential: The Development and User-Friendly Evaluation of a Virtual Reality Intervention for Attention-Deficit/Hyperactivity Disorder

(1) Background: Attention-Deficit/Hyperactivity Disorder (ADHD) is typically first diagnosed in early childhood. Medication and cognitive behavioural therapy are considered effective in treating children with ADHD, whereas these treatments appear to have some side effects and restrictions. Virtual reality (VR), therefore, has been applied to exposure therapy for mental disorders. Previous studies have adopted VR in the cognitive behavioural treatment for children with ADHD; however, no research has used VR to develop social skills training for children with ADHD. This study aimed to develop a VR-based intervention (Social VR) to improve social skills in children with symptoms of ADHD. Prior to conducting the pilot trial to assess the effectiveness of Social VR, valuable user feedback was gathered regarding the mechanics of Social VR, satisfaction and motion sickness. This study presented the development and preliminary usability of Social VR to enhance social interaction skills among children with ADHD. (2) Methods: The development process of the Social VR intervention was demonstrated. The Social VR intervention consisted of three scenarios, namely MTR, Campus and Market and Restaurant. In the usability study, 25 children with ADHD were recruited to test the Social VR during the preliminary usability stage of a clinical trial at preinclusion. The participants completed a survey about their experience of playing Social VR, satisfaction, and motion sickness. (3) Results: The participants indicated the three conditions had easy-to-follow instructions, were easy to pick up, and that they understood when the situations changed. The control and beauty of the graphics of Market and Restaurant were relatively lower compared with those of MTR and Campus. The three scenarios are applicable to children diagnosed with any subtype of ADHD. (4) Conclusion: The participants were satisfied with Social VR. Social VR was generally considered realistic and immersive. Further trials to assess the feasibility and efficacy were discussed. If proven effective, VR-based intervention can be an adjunctive approach to current multimodal training for children with ADHD.

Abstract 11784: A Deep-Learning-Based Multi-Modal Phonocardiogram and Electrocardiogram Processing Framework for Screening Left Ventricular Systolic Dysfunction Using a Wearable Cardiac Patch

Introduction: Left ventricular systolic dysfunction (LVSD) is associated with poor health outcomes. Previous study has demonstrated the effectiveness of electromechanical activation time (EMAT) derived from phonocardiography (PCG) and electrocardiography (ECG) signals in identifying LVSD. However, high throughput and well-performed algorithm is needed in automatically detecting LVSD. Hypothesis: An artificial intelligence (AI) algorithm was developed and validated in detecting LVSD. Methods: Using ECG and PCG data collected by wearable patch from 1020 admitted patients in Ruijin hospital, we trained an AI algorithm to detect the LVSD, which was determined by ejection fraction &lt;50% from echocardiogram. A separate 590 patients were taken as independent test set. The AI algorithm followed a two-step detection structure, where the first step is to use LSTM learning to quantify heartbeat and heartbeat interval in ECG, and the second step is to apply convolution neural network (CNN) in estimating optimal EMAT cutoff points in detecting LVSD. To overcome the challenge of multi-modal signal training, we develop an alignment-based contrastive learning approach that generates shared feature space embeddings for ECG and PCG signals. Results: Among training and test sets, 18.2% and 26.1% were LVSD, respectively. The AI algorithm yields an AUROC of 0.92 in detecting LVSD with sensitivity of 85.63%, and specificity of 79.39%. Among the test set, model yielded an AUROC of 0.86, sensitivity of 81.29%, and specificity of 77.13%. Conclusions: A wearable device incorporated with a high performed AI algorithm can be a viable, convenient, cost-effective tool in screening LVSD.

Self-supervised multi-modal training from uncurated images and reports enables monitoring AI in radiology

The escalating demand for artificial intelligence (AI) systems that can monitor and supervise human errors and abnormalities in healthcare presents unique challenges. Recent advances in vision-language models reveal the challenges of monitoring AI by understanding both visual and textual concepts and their semantic correspondences. However, there has been limited success in the application of vision-language models in the medical domain. Current vision-language models and learning strategies for photographic images and captions call for a web-scale data corpus of image and text pairs which is not often feasible in the medical domain. To address this, we present a model named medical cross-attention vision-language model (Medical X-VL), which leverages key components to be tailored for the medical domain. The model is based on the following components: self-supervised unimodal models in medical domain and a fusion encoder to bridge them, momentum distillation, sentencewise contrastive learning for medical reports, and sentence similarity-adjusted hard negative mining. We experimentally demonstrated that our model enables various zero-shot tasks for monitoring AI, ranging from the zero-shot classification to zero-shot error correction. Our model outperformed current state-of-the-art models in two medical image datasets, suggesting a novel clinical application of our monitoring AI model to alleviate human errors. Our method demonstrates a more specialized capacity for fine-grained understanding, which presents a distinct advantage particularly applicable to the medical domain.

Use of Multimodal and Interactive Digital Teaching Materials in 21st Century Mathematics Learning

Multimodal interactive digital teaching materials are learning materials that contain learning materials that are in accordance with learning objectives designed by utilizing various resources including text, images, audio, video, music, and social media in the form of YouTube, tik tok, and appropriate internet links and are interactive because they are digital based. The aim of this activity is to improve teachers' understanding and skills in using multimodal digital teaching materials in 21st century learning. Participants in the community service activity were 25 participants consisting of teachers. This activity was carried out at MTs Al Gebra, Sorong City and MI Al Islah, Sorong City. The methods used in multimodal digital training are material presentation, discussion and training simulation. The results of this training have a positive impact on teachers in managing the learning process. This can be seen from the analysis results that 80% of participants have increased knowledge, 80% of participants have increased creativity and 80% of participants need continuous training in increasing knowledge to prepare themselves in facing the digital era 4.0.

Multimodal agility-based exercise training (MAT) versus strength and endurance training (SET) to improve multiple sclerosis-related fatigue and fatigability during inpatient rehabilitation: a randomized controlled pilot and feasibility study [ReFEx

BackgroundMultimodal agility-based exercise training (MAT) is a group-based exercise training framework for persons with multiple sclerosis (pwMS) with a potential to impact fatigue and fatigability. In a mixed-methods design, this study evaluated the feasibility of implementing MAT in an inpatient rehabilitation setting and the feasibility of a randomized controlled trial (RCT) study protocol with ‘traditional’ strength and endurance training (SET) as an active control condition. Secondarily, preliminary outcome data was acquired.MethodsPwMS with low to moderate disability and self-reported fatigue were randomly allocated to either MAT or SET when starting inpatient rehabilitation (4–6 weeks). The MAT-participants exercised in a group following a MAT-manual (sessions were gym- (5x/week) and pool-based (3x/week)). SET-participants exercised individually 5x/week on a cycle ergometer, and 3x/week on strength training machines. Feasibility assessments focused on processes, resources, management, time, and scientific domains. Assessed clinical outcomes at admission and discharge included perceived fatigue, motor and cognitive fatigability, cognitive performance, motor function, and balance confidence. Perceived fatigue was reassessed 1, 4, and 12 weeks after discharge. Feasibility was determined regarding predetermined progression criteria.ResultsTwenty-two participants were randomized. Both groups performed the minimum number of sessions (> 18), and retention was adequate (73–91%). SET-participants performed more sessions than MAT-participants (30.8 vs. 22.7) and stayed longer in the facility (34.2 vs. 31.6 days). Non-eligibility of admitted pwMS was high (74% non-eligible), mainly due to high EDSS and inability to attend pool-based sessions. Consequently, recruitment (1.8/month) was slower than the predetermined progression criterium. Baseline assessments took longer than required (only 50% completed within 3 days). Short-term fatigue reduction was similar for both groups. Motor fatigability also improved in both groups, whereas cognitive fatigability deteriorated. In MAT, average improvement in walking endurance (43.9 m) exceeded minimal important change values for individuals (> 26.9 m).ConclusionsProgressing to a definitive RCT necessitates adaptation of eligibility criteria. In the present design it will also be difficult to attain similar dosing of interventions. A multicenter RCT focused only on gym-based MAT might be another option to assess the effect of MAT. The primary outcome measure should be able to measure change in perceived fatigue more robustly.Trial registrationGerman Clinical Trials Register: DRKS00023943, date of registration: 23 September 2021.

MinJoT: Multimodal infusion Joint Training for noise learning in text and multimodal classification problems

Amidst the critical role that high-quality labeled data plays in advancing machine learning, the persistence of noise within widely-used datasets remains a challenge. While noise learning has gained traction within machine learning, particularly in computer vision, its exploration in text and multimodal classification domains has lagged. Furthermore, a comprehensive comparison of noise learning techniques in text and multimodal classification has been lacking, partly due to variations in experimental noise settings across prior studies. Addressing these gaps, this research introduces a pioneering Multimodal Infusion Joint Training (MinJoT) framework featuring a novel co-regularized loss function that seamlessly integrates multimodal information during joint training. This framework notably excels in maintaining model robustness amidst noisy data environments. Adapting established noise learning methods from computer vision to text classification, the study conducts extensive experiments across five English and Chinese textual and multimodal datasets, involving four methods, five noise modes, and seven noise rates. Critically, this work challenges the implicit assumption that widely-used datasets are devoid of noise, revealing that these datasets indeed encompass noise levels ranging from 0.61% to 15.77% which is defined as intrinsic noise in this study. For the first time, the study investigates the impact of intrinsic noise on model performance, categorizing it into distinct levels of ambiguity. To facilitate accurate method comparison, a new dataset, Golden-Chnsenticorp (G-Chnsenticorp), is introduced, carefully crafted to be free of intrinsic noise. This research establishes the inaugural noise learning benchmark for text classification, while simultaneously pioneering the first noise learning framework tailored for multimodal sentiment classification. Through these contributions, the study advances the understanding of noise learning in text and multimodal contexts, providing a novel framework, benchmarks, and insights to propel the field forward.