Hasil untuk "Computer applications to medicine. Medical informatics"

Sudais Imtiaz, Cheng Yu, Xuewei Chen et al.

Background Telehealth applications and mobile services have been growing in popularity. As their reach expands service across international boundaries, it remains unclear to what extent Chinese immigrants residing in the United States are using China-based medical applications and the factors impacting their uptake. Although transnational telehealth apps are beneficial in bridging cultural and linguistic gaps, they come with distinct risks and challenges that need to be further explored. Objectives The study had three aims: (1) estimate the prevalence of China-based telehealth app usage by Chinese migrants in the US, (2) identify factors associated with China-based telehealth apps utilization among Chinese migrants in the US, and (3) describe how Chinese migrants in the US are using and can use China-based telehealth apps remotely from the US. Methods Four focus groups ( n  = 17) and a cross-sectional survey ( n  = 227) were conducted among recent Chinese immigrants to the US (arrived in the past 10 years). Results Overall, 15% indicated usage of China-based telehealth apps while living within the US. Use of China-based telehealth apps while living in the US was associated with: higher perceived frequency of experiencing healthcare discrimination in the US (odds ratio (OR): 1.43, 95% CI: 1.14–1.80), younger age (OR: 7.86, 95% CI: 1.32–47.01), female sex (OR: 4.29, 95% CI: 1.50–12.23), living in a community with a large Chinese community (OR: 9.53, 95% CI: 1.90–47.79), and lack of medical insurance (OR: 51.59, 95% CI: 3.88–685.70). Some Chinese migrants living in the US are using China-based telehealth apps to consult with medical providers in China as their first line of medical consultation. Conclusion Findings suggest uptake of China-based telehealth are partially driven by negative experiences within the US healthcare system. These results are indicative of possible shortcomings in existing healthcare services that diminish the capacity to appropriately address the needs of immigrant communities and groups.

Olga Maiatska, Torsten Dünnebacke, Martin Kreuels et al.

Organic scintillators using a solid polyurethane (PU) matrix have been introduced to combine the robustness of a construction material with scintillating properties that allow gamma rays and fast neutrons to be detected efficiently and at low cost. This work compares two corresponding materials, the older M600 and the more recent M700, with EJ-276D and EJ-200 representing common plastic scintillators with and without pulse-shape discrimination (PSD) capabilities, respectively. Characterization measurements were performed with small samples of 26 mm diameter and 10 mm height, which were coupled to a photomultiplier tube (PMT) and simultaneously exposed to 252Cf fission neutrons and 137Cs gamma rays. M700 turned out to provide the best PSD performance and about the same light yield as EJ-276D, while its light pulses exhibit a shorter pulse decay. An accelerated ageing process applied in between two test campaigns was too short to trigger distinct performance degradation in any of the materials, though optical degradation was visible in EJ-276D and in EJ-200 but not in the PU-based materials. Nevertheless, the extremely robust polyurethane matrix promises advantages in medical and industrial applications where resilience and long-term stability are of crucial importance.

Kristin Rogers, Alanna Miller, Ashley Girgis et al.

Abstract BackgroundHealth Evidence provides access to quality appraisals for >10,000 evidence syntheses on the effectiveness and cost-effectiveness of public health and health promotion interventions. Maintaining Health Evidence has become increasingly resource-intensive due to the exponential growth of published literature. Innovative screening methods using artificial intelligence (AI) can potentially improve efficiency. ObjectiveThe objectives of this project are to: (1) assess the ability of AI-assisted screening to correctly predict nonrelevant references at the title and abstract level and investigate the consistency of this performance over time, and (2) evaluate the impact of AI-assisted screening on the overall monthly manual screening set. MethodsTraining and testing were conducted using the DistillerSR AI Preview & Rank feature. A set of manually screened references (n=43,273) was uploaded and used to train the AI feature and assign probability scores to each reference to predict relevance. A minimum threshold was established where the AI feature correctly identified all manually screened relevant references. The AI feature was tested on a separate set of references (n=72,686) from the May 2019 to April 2020 monthly searches. The testing set was used to determine an optimal threshold that ensured >99% of relevant references would continue to be added to Health Evidence. The performance of AI-assisted screening at the title and abstract screening level was evaluated using recall, specificity, precision, negative predictive value, and the number of references removed by AI. The number and percentage of references removed by AI-assisted screening and the change in monthly manual screening time were estimated using an implementation reference set (n=272,253) from November 2020 to 2023. ResultsThe minimum threshold in the training set of references was 0.068, which correctly removed 37% (n=16,122) of nonrelevant references. Analysis of the testing set identified an optimal threshold of 0.17, which removed 51,706 (71.14%) references using AI-assisted screening. A slight decrease in recall between the 0.068 minimum threshold (99.68%) and the 0.17 optimal threshold (94.84%) was noted, resulting in four missed references included via manual screening at the full-text level. This was accompanied by an increase in specificity from 35.95% to 71.70%, doubling the proportion of references AI-assisted screening correctly predicted as not relevant. Over 3 years of implementation, the number of references requiring manual screening was reduced by 70%, reducing the time spent manually screening by an estimated 382 hours. ConclusionsGiven the magnitude of newly published peer-reviewed evidence, the curation of evidence supports decision makers in making informed decisions. AI-assisted screening can be an important tool to supplement manual screening and reduce the number of references that require manual screening, ensuring that the continued availability of curated high-quality synthesis evidence in public health is possible.

Kristy A Carpenter, Anna T Nguyen, Delaney A Smith et al.

Social media can provide real-time insight into trends in substance use, addiction, and recovery. Prior studies have used platforms such as Reddit and X (formerly Twitter), but evolving policies around data access have threatened these platforms' usability in research. We evaluate the potential of a broad set of platforms to detect emerging trends in the opioid use disorder and overdose epidemic. From these, we identified 11 high-potential platforms, for which we documented policies regulating drug-related discussion, data accessibility, geolocatability, and prior use in opioid-related studies. We quantified their volume of opioid discussion, including in informal language by including slang generated using a large language model. Beyond the most commonly used Reddit and X/Twitter, the platforms with high potential for use in opioid-related surveillance are TikTok, YouTube, and Facebook. Leveraging a variety of social platforms, instead of merely one, yields broader subpopulation representation and safeguards against reduced data access in any single platform.

Yufeng Li, Fangyuan Chang, Wen Zhang et al.

BackgroundAdolescent idiopathic scoliosis is a common spinal deformity typically treated with exercise therapy. Despite the increasing use of digital technologies in interventions, there remains a gap in understanding how to effectively integrate behavior change techniques (BCTs) and behavior theories within these digital solutions. ObjectiveThis review aims to identify the digital characteristics of interventions and the BCTs used, and to analyze potential theoretical mechanisms with the Theoretical Domains Framework and the capability, opportunity, motivation, and behavior model. MethodsWe conducted a scoping review according to the PRISMA-ScR (Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews) guidelines. A total of 5 databases, including PubMed, Web of Science, Embase, Cochrane Library, and CINAHL, were selected for screening eligible studies up to April 4, 2024. We included studies of any design type that involved patients with adolescent idiopathic scoliosis using digital interventions for exercise rehabilitation, including qualitative, quantitative, or mixed methods studies, and study protocols with detailed descriptions of digital interventions. Two researchers independently screened studies and extracted data into tables for descriptive analysis. The Mixed Methods Appraisal Tool was used to assess the quality of studies. ResultsOut of the 3267 identified papers, 21 (0.64%) studies were included. The most frequently used technologies were videoconferencing (n=7) and instructional videos (n=5). The three most common BCT clusters were “Shaping Knowledge” (n=19), “Social Support” (n=16), and “Antecedents” (n=16). “Knowledge” was the most used mechanism of action (n=21), followed by “Skills” (n=16), “Environmental Context and Resources” (n=16), and “Social Influences” (n=16). The studies primarily addressed “Capability” and “Opportunity,” with less emphasis on “Motivation,” particularly “Automatic Motivation.” ConclusionsThis review identified common digital technologies and their characteristics, analyzed potential mechanisms of behavior change in interventions, and provided recommendations for technology utilization. Future research should further evaluate the effectiveness of digital technologies while enhancing patient motivation and user experience. Trial RegistrationPROSPERO CRD42024530851; https://www.crd.york.ac.uk/PROSPERO/view/CRD42024530851

Shu Liu

Objective This study aimed to assess the effects of older Chinese adults’ online behaviors (including recent access to the Internet and duration of Internet use) on their current health habits and health status. Health status comprised self-rated physical and mental health. Methods Interviewees aged over 60 years who participated consecutively in the 2010, 2014, 2018, and 2020 China Family Panel Studies were selected. Ordinary least squares and ordered logit regression analyses were used to analyze the associations between older adults’ online behaviors and their health habits and status. Propensity score matching analysis was employed to mitigate selection bias. Structural equation modeling was conducted to test the robustness of the findings and to explore whether the associations between recent and continued Internet use and their health were mediated by older adults’ prioritization of online entertainment, learning, communication, and lifestyle-related activities. Results The results showed that Internet access among older adults was positively associated with health habits, increased self-rated health levels, and improved physical and mental health. Continuous Internet use among older adults may be a predictor of better mental health. While prioritization of online life style could improve physical health, prioritization of online communication might be prejudicial to acquiring health habits. Conclusion Internet access may support healthy aging by encouraging older adults to adopt healthier lifestyles. However, the varied effects of different types of online behavior underscore the importance of designing targeted digital interventions rather than blanket promotion of technology adoption.

Yun Zhu, Dong Zhang, Yi Lin et al.

Medical image segmentation demands the aggregation of global and local feature representations, posing a challenge for current methodologies in handling both long-range and short-range feature interactions. Recently, vision mamba (ViM) models have emerged as promising solutions for addressing model complexities by excelling in long-range feature iterations with linear complexity. However, existing ViM approaches overlook the importance of preserving short-range local dependencies by directly flattening spatial tokens and are constrained by fixed scanning patterns that limit the capture of dynamic spatial context information. To address these challenges, we introduce a simple yet effective method named context clustering ViM (CCViM), which incorporates a context clustering module within the existing ViM models to segment image tokens into distinct windows for adaptable local clustering. Our method effectively combines long-range and short-range feature interactions, thereby enhancing spatial contextual representations for medical image segmentation tasks. Extensive experimental evaluations on diverse public datasets, i.e., Kumar, CPM17, ISIC17, ISIC18, and Synapse demonstrate the superior performance of our method compared to current state-of-the-art methods. Our code can be found at https://github.com/zymissy/CCViM.

Pedro J. Martinez-Ferrer, Albert-Jan Yzelman, Vicenç Beltran

The tensor-vector contraction (TVC) is the most memory-bound operation of its class and a core component of the higher-order power method (HOPM). This paper brings distributed-memory parallelization to a native TVC algorithm for dense tensors that overall remains oblivious to contraction mode, tensor splitting and tensor order. Similarly, we propose a novel distributed HOPM, namely dHOPM3, that can save up to one order of magnitude of streamed memory and is about twice as costly in terms of data movement as a distributed TVC operation (dTVC) when using task-based parallelization. The numerical experiments carried out in this work on three different architectures featuring multi-core and accelerators confirm that the performances of dTVC and dHOPM3 remain relatively close to the peak system memory bandwidth (50%-80%, depending on the architecture) and on par with STREAM benchmark figures. On strong scalability scenarios, our native multi-core implementations of these two algorithms can achieve similar and sometimes even greater performance figures than those based upon state-of-the-art CUDA batched kernels. Finally, we demonstrate that both computation and communication can benefit from mixed precision arithmetic also in cases where the hardware does not support low precision data types natively.

Zhaoyi Sun, Wen-Wai Yim, Ozlem Uzuner et al.

Objective: This review aims to explore the potential and challenges of using Natural Language Processing (NLP) to detect, correct, and mitigate medically inaccurate information, including errors, misinformation, and hallucination. By unifying these concepts, the review emphasizes their shared methodological foundations and their distinct implications for healthcare. Our goal is to advance patient safety, improve public health communication, and support the development of more reliable and transparent NLP applications in healthcare. Methods: A scoping review was conducted following PRISMA guidelines, analyzing studies from 2020 to 2024 across five databases. Studies were selected based on their use of NLP to address medically inaccurate information and were categorized by topic, tasks, document types, datasets, models, and evaluation metrics. Results: NLP has shown potential in addressing medically inaccurate information on the following tasks: (1) error detection (2) error correction (3) misinformation detection (4) misinformation correction (5) hallucination detection (6) hallucination mitigation. However, challenges remain with data privacy, context dependency, and evaluation standards. Conclusion: This review highlights the advancements in applying NLP to tackle medically inaccurate information while underscoring the need to address persistent challenges. Future efforts should focus on developing real-world datasets, refining contextual methods, and improving hallucination management to ensure reliable and transparent healthcare applications.

Marco Lehmann, Lucy Jones, Felix Schirmann

BackgroundEarly weight loss is an established predictor for treatment outcomes in weight management interventions for people with obesity. However, there is a paucity of additional, reliable, and clinically actionable early predictors in weight management interventions. Novel blended-care weight management interventions combine coach and app support and afford new means of structured, continuous data collection, informing research on treatment adherence and outcome prediction. ObjectiveAgainst this backdrop, this study analyzes app engagement as a predictor for weight loss in large-scale, real-world, blended-care interventions. We hypothesize that patients who engage more frequently in app usage in blended-care treatment (eg, higher logging activity) lose more weight than patients who engage comparably less frequently at 3 and 6 months of intervention. MethodsReal-world data from 19,211 patients in obesity treatment were analyzed retrospectively. Patients were treated with 3 different blended-care weight management interventions, offered in Switzerland, the United Kingdom, and Germany by a digital behavior change provider. The principal component analysis identified an overarching metric for app engagement based on app usage. A median split informed a distinction in higher and lower engagers among the patients. Both groups were matched through optimal propensity score matching for relevant characteristics (eg, gender, age, and start weight). A linear regression model, combining patient characteristics and app-derived data, was applied to identify predictors for weight loss outcomes. ResultsFor the entire sample (N=19,211), mean weight loss was –3.24% (SD 4.58%) at 3 months and –5.22% (SD 6.29%) at 6 months. Across countries, higher app engagement yielded more weight loss than lower engagement after 3 but not after 6 months of intervention (P3 months<.001 and P6 months=.59). Early app engagement within the first 3 months predicted percentage weight loss in Switzerland and Germany, but not in the United Kingdom (PSwitzerland<.001, PUnited Kingdom=.12, and PGermany=.005). Higher age was associated with stronger weight loss in the 3-month period (PSwitzerland=.001, PUnited Kingdom=.002, and PGermany<.001) and, for Germany, also in the 6-month period (PSwitzerland=.09, PUnited Kingdom=.46, and PGermany=.03). In Switzerland, higher numbers of patients’ messages to coaches were associated with higher weight loss (P3 months<.001 and P6 months<.001). Messages from coaches were not significantly associated with weight loss (all P>.05). ConclusionsEarly app engagement is a predictor of weight loss, with higher engagement yielding more weight loss than lower engagement in this analysis. This new predictor lends itself to automated monitoring and as a digital indicator for needed or adapted clinical action. Further research needs to establish the reliability of early app engagement as a predictor for treatment adherence and outcomes. In general, the obtained results testify to the potential of app-derived data to inform clinical monitoring practices and intervention design.

Jinqiang Wang, Huansheng Ning, Yi Peng et al.

Large Language Models (LLMs) have demonstrated surprising performance across various natural language processing tasks. Recently, medical LLMs enhanced with domain-specific knowledge have exhibited excellent capabilities in medical consultation and diagnosis. These models can smoothly simulate doctor-patient dialogues and provide professional medical advice. Most medical LLMs are developed through continued training of open-source general LLMs, which require significantly fewer computational resources than training LLMs from scratch. Additionally, this approach offers better patient privacy protection than API-based solutions. Given the above advantages, this survey systematically summarizes how to train medical LLMs based on open-source general LLMs from a more fine-grained perspective. It covers (a) how to acquire training corpus and construct customized medical training sets, (b) how to choose an appropriate training paradigm, (c) how to choose a suitable evaluation benchmark, and (d) existing challenges and promising research directions are discussed. This survey can provide guidance for the development of LLMs focused on various medical applications, such as medical education, diagnostic planning, and clinical assistants. Related resources and supplemental information can be found on the GitHub repository.

Sibo Wei, Xueping Peng, Yi-Fei Wang et al.

The surge of large language models (LLMs) has driven significant progress in medical applications, including traditional Chinese medicine (TCM). However, current medical LLMs struggle with TCM diagnosis and syndrome differentiation due to substantial differences between TCM and modern medical theory, and the scarcity of specialized, high-quality corpora. To this end, in this paper we propose BianCang, a TCM-specific LLM, using a two-stage training process that first injects domain-specific knowledge and then aligns it through targeted stimulation to enhance diagnostic and differentiation capabilities. Specifically, we constructed pre-training corpora, instruction-aligned datasets based on real hospital records, and the ChP-TCM dataset derived from the Pharmacopoeia of the People's Republic of China. We compiled extensive TCM and medical corpora for continual pre-training and supervised fine-tuning, building a comprehensive dataset to refine the model's understanding of TCM. Evaluations across 11 test sets involving 31 models and 4 tasks demonstrate the effectiveness of BianCang, offering valuable insights for future research. Code, datasets, and models are available on https://github.com/QLU-NLP/BianCang.

Andrea D'Angelo, Tiziana Catarci, Antinisca Di Marco et al.

This study aims to enrich and leverage data from the Informatics Europe Higher Education (IEHE) data portal to extract and analyze trends in female participation in Informatics across Europe. The research examines the proportion of female students, first-year enrollments, and degrees awarded to women in the field. The issue of low female participation in Informatics has long been recognized as a persistent challenge and remains a critical area of scholarly inquiry. Furthermore, existing literature indicates that socio-economic factors can unpredictably influence female participation, complicating efforts to address the gender gap. The analysis focuses on participation data from research universities at various academic levels, including Bachelors, Masters, and PhD programs, and seeks to uncover potential correlations between female participation and geographical or economic zones. The dataset was first enriched by integrating additional information, such as each country's GDP and relevant geographical data, sourced from various online repositories. Subsequently, the data was cleaned to ensure consistency and eliminate incomplete time series. A final set of complete time series was selected for further analysis. We then used the data collected from the internet to assign countries to different clusters. Specifically, we employed Economic Zone, Geographical Area, and GDP quartile to cluster countries and compare their temporal trends both within and between clusters. We analyze the results for each classification and derive conclusions based on the available data.

Stefano Woerner, Arthur Jaques, Christian F. Baumgartner

While the field of medical image analysis has undergone a transformative shift with the integration of machine learning techniques, the main challenge of these techniques is often the scarcity of large, diverse, and well-annotated datasets. Medical images vary in format, size, and other parameters and therefore require extensive preprocessing and standardization, for usage in machine learning. Addressing these challenges, we introduce the Medical Imaging Meta-Dataset (MedIMeta), a novel multi-domain, multi-task meta-dataset. MedIMeta contains 19 medical imaging datasets spanning 10 different domains and encompassing 54 distinct medical tasks, all of which are standardized to the same format and readily usable in PyTorch or other ML frameworks. We perform a technical validation of MedIMeta, demonstrating its utility through fully supervised and cross-domain few-shot learning baselines.

Adriana Vieira, Inês Sousa, Sónia Dória-Nóbrega

When dealing with several years of daily data, such as the number of daily admissions to a hospital’s emergency department (ED), how complex does it get to forecast into the future? With that in mind, this study has two main goals: to explore the differences between several methodologies, considering both single and multiple-seasonal patterns; and to select the most suitable model for the administration of a Portuguese hospital to use while managing their ED. To that end, we first considered the data as a time series with a single weekly seasonal pattern. We then modelled the data using time series regression, linear regression with autoregressive integrated moving average (ARIMA) errors, seasonal ARIMA and exponential smoothing techniques. Second, the data was set to be a time series with weekly and annual seasonal patterns. Then, using Fourier terms, we applied time series regression, linear regression with ARIMA errors and trigonometric exponential smoothing state space models with Box–Cox transformation, ARMA errors, Trend and Seasonal components (TBATS) for the analysis. After selecting the best-fitting models using the Akaike Information Criteria (AIC) values, we forecasted into the future and compared the results using both training and test datasets’ root mean square error (RMSE), mean absolute error (MAE) and mean absolute percentage error (MAPE) values. The time series regression model based on seasonal variables and a weekly seasonal pattern gives the best results. However, we decided to use linear regression with ARIMA errors, seasonal variables, and both weekly and annual seasonal patterns. This produces similar results but allows for the annual seasonality to be considered, which is useful when more data is added.

Emile B Gordon, Peter Wingrove, Barton F Branstetter Iv et al.

After their rapid adoption at the onset of the coronavirus pandemic, remote case reviews (remote readouts) between diagnostic radiology residents and their attendings have persisted in an increasingly remote workforce, despite relaxing social distancing guidelines. Our objective was to evaluate the impact of the transition to remote readouts on resident case volumes after the recovery of institutional volumes. We tabulated radiology reports co-authored by first-to-third-year radiology residents (R1-R3) between July 1 and December 31 of the first pandemic year, 2020, and compared to the prior two pre-pandemic years. Half-years were analyzed because institutional volumes recovered by July 2020. Resident volumes were normalized to rotations, which were in divisions categorized by the location of the supervising faculty during the pandemic period; in 'remote' divisions, all faculty worked off-site, whereas 'hybrid' divisions had a mix of attendings working on-site and remotely. All residents worked on-site. Data analysis was performed with Student's t test and multivariate linear regression. The largest drops in total case volume occurred in the two remote divisions (38% [6,086 to 3,788], and 26% [11,046 to 8,149]). None of the hybrid divisions with both in-person and remote supervision decreased by more than 5%. With multivariate regression, a resident assigned to a standardized remote rotation in 2020 would complete 32% (253 to 172) fewer studies than in identical pre-pandemic rotations (coefficent of -81.6, p = .005) but would be similar for hybrid rotations. R1 residents would be expected to interpret 40% fewer (180 to 108) cases on remote rotations during the pandemic (coefficient of -72.3, p = .007). No significant effect was seen for R2 or R3 residents (p = .099 and p = .29, respectively). Radiology residents interpreted fewer studies during remote rotations than on hybrid rotations that included in-person readouts. As resident case volume is correlated with clinical performance and board pass rate, monitoring the readout model for downstream educational effects is essential. Until evidence shows that educational outcomes remain unchanged, radiology residencies may wish to preserve in-person resident readouts, particularly for junior residents.

Aysam Guerler, Dannon Baker, Marius van den Beek et al.

Abstract Background Protein–protein interactions play a crucial role in almost all cellular processes. Identifying interacting proteins reveals insight into living organisms and yields novel drug targets for disease treatment. Here, we present a publicly available, automated pipeline to predict genome-wide protein–protein interactions and produce high-quality multimeric structural models. Results Application of our method to the Human and Yeast genomes yield protein–protein interaction networks similar in quality to common experimental methods. We identified and modeled Human proteins likely to interact with the papain-like protease of SARS-CoV2’s non-structural protein 3. We also produced models of SARS-CoV2’s spike protein (S) interacting with myelin-oligodendrocyte glycoprotein receptor and dipeptidyl peptidase-4. Conclusions The presented method is capable of confidently identifying interactions while providing high-quality multimeric structural models for experimental validation. The interactome modeling pipeline is available at usegalaxy.org and usegalaxy.eu.

Neil K R Sehgal, Benjamin Rader, Autumn Gertz et al.

COVID-19 vaccination rates among children have stalled, while new coronavirus strains continue to emerge. To improve child vaccination rates, policymakers must better understand parental preferences and reasons for COVID-19 vaccination among their children. Cross-sectional surveys were administered online to 30,174 US parents with at least one child of COVID-19 vaccine eligible age (5-17 years) between January 1 and May 9, 2022. Participants self-reported willingness to vaccinate their child and reasons for refusal, and answered additional questions about demographics, pandemic related behavior, and vaccination status. Willingness to vaccinate a child for COVID-19 was strongly associated with parental vaccination status (multivariate odds ratio 97.9, 95% confidence interval 86.9-111.0). The majority of fully vaccinated (86%) and unvaccinated (84%) parents reported concordant vaccination preferences for their eligible child. Age and education had differing relationships by vaccination status, with higher age and education positively associated with willingness among vaccinated parents. Among all parents unwilling to vaccinate their children, the two most frequently reported reasons were possible side effects (47%) and that vaccines are too new (44%). Unvaccinated parents were much more likely to list a lack of trust in government (41% to 21%, p < .001) and a lack of trust in scientists (34% to 19%, p < .001) as reasons for refusal. Cluster analysis identified three groups of unwilling parents based on their reasons for refusal to vaccinate, with distinct concerns that may be obscured when analyzed in aggregate. Factors associated with willingness to vaccinate children and reasons for refusal may inform targeted approaches to increase vaccination.

Yang Liu, Shi Gu

The registration of pathological images plays an important role in medical applications. Despite its significance, most researchers in this field primarily focus on the registration of normal tissue into normal tissue. The negative impact of focal tissue, such as the loss of spatial correspondence information and the abnormal distortion of tissue, are rarely considered. In this paper, we propose GIRNet, a novel unsupervised approach for pathological image registration by incorporating segmentation and inpainting through the principles of Generation, Inpainting, and Registration (GIR). The registration, segmentation, and inpainting modules are trained simultaneously in a co-learning manner so that the segmentation of the focal area and the registration of inpainted pairs can improve collaboratively. Overall, the registration of pathological images is achieved in a completely unsupervised learning framework. Experimental results on multiple datasets, including Magnetic Resonance Imaging (MRI) of T1 sequences, demonstrate the efficacy of our proposed method. Our results show that our method can accurately achieve the registration of pathological images and identify lesions even in challenging imaging modalities. Our unsupervised approach offers a promising solution for the efficient and cost-effective registration of pathological images. Our code is available at https://github.com/brain-intelligence-lab/GIRNet.

Milin Zhang, Mohammad Abdi, Jonathan Ashdown et al.

Distributed deep neural networks (DNNs) have been shown to reduce the computational burden of mobile devices and decrease the end-to-end inference latency in edge computing scenarios. While distributed DNNs have been studied, to the best of our knowledge, the resilience of distributed DNNs to adversarial action remains an open problem. In this paper, we fill the existing research gap by rigorously analyzing the robustness of distributed DNNs against adversarial action. We cast this problem in the context of information theory and rigorously proved that (i) the compressed latent dimension improves the robustness but also affect task-oriented performance; and (ii) the deeper splitting point enhances the robustness but also increases the computational burden. These two trade-offs provide a novel perspective to design robust distributed DNN. To test our theoretical findings, we perform extensive experimental analysis by considering 6 different DNN architectures, 6 different approaches for distributed DNN and 10 different adversarial attacks using the ImageNet-1K dataset.