Open Access Publications - Department of Computer and Information Sciences

Permanent URI for this collection

https://udspace.udel.edu/handle/19716/29841

Open access publications by faculty, postdocs, and graduate students in the Department of Computer and Information Sciences.

Browse

Now showing 1 - 20 of 69

Hybrid Deep Learning Model to Estimate Cognitive Effort from fNIRS Signals
(Companion Proceedings of the 27th International Conference on Multimodal Interaction, 2025-10-12) Sharmin, Shayla; Barmaki, Roghayeh Leila
This study estimates cognitive effort based on functional near-infrared spectroscopy data and performance scores using a hybrid DeepNet model. The estimation of cognitive effort enables educators to modify material to enhance learning effectiveness and student engagement. In this study, we collected oxygenated hemoglobin using functional near-infrared spectroscopy during an educational quiz game. Participants (n=16) responded to 16 questions in a Unity-based educational game, each within a 30-second response time limit. We used DeepNet models to predict the performance score from the oxygenated hemoglobin, and compared traditional machine learning and DeepNet models to determine which approach provides better accuracy in predicting performance scores. The result shows that the proposed CNN-GRU gives better performance with 73% than other models. After the prediction, we used the predicted score and the oxygenated hemoglobin to observe cognitive effort by calculating relative neural efficiency and involvement in our test cases. Our result shows that even with moderate accuracy, the predicted cognitive effort closely follow the actual trends. This findings can be helpful in designing and improving learning environments and provide valuable insights into learning materials.
MENA: A Multimodal Framework for Analyzing Caregiver Emotions and Competencies in AR Geriatric Simulations
(Proceedings of the 27th International Conference on Multimodal Interaction, 2025-10-13) Kiafar, Behdokht; Ravva, Pavan Uttej; Daher, Salam; Ahmmed Joy, Asif; Barmaki, Roghayeh Leila
Improving the quality of geriatric care is a challenge that requires insights from stakeholders. While simulated trainings can boost competencies, extracting meaningful insights from these practices to enhance simulation effectiveness remains a challenge. In this study, we introduce Multimodal Epistemic Network Analysis (MENA), a novel framework for analyzing caregiver attitudes and emotional responses in an Augmented Reality simulation. By integrating a multimodal Emotional State Classifier, MENA extends traditional epistemic network analysis to reveal complex relationships between caregiving competencies and positive emotions. Applied in a pilot study (N = 20) comparing caregiver interactions with an unaware versus an aware virtual geriatric patient (VGP), MENA visualizations demonstrated how awareness in the VGP fostered more supportive and person-centered caregiving behaviors. These findings suggest that MENA not only enhances the analysis of mul-timodal interactions but also provides a powerful tool for designing emotionally intelligent training systems that prepare caregivers for the nuanced demands of real-world practice. The code and setup to reproduce the experiments are publicly available here, and data is available upon request.
Constructivity conditions on immune sets
(Archive for Mathematical Logic, 2025-01-29) Case, John
Definitionally: strongly effectively immune sets are infinite and their c.e. subsets have maximums effectively bounded in their c.e. indices; whereas, for effectively immune sets, their c.e. subsets’ cardinalities are what’re effectively bounded. This definitional difference between these two kinds of sets is very nicely paralleled by the following difference between their complements. McLaughlin: strongly effectively immune sets cannot have immune complements; whereas, the main theorem herein: effectively immune sets cannot have hyperimmune complements. Ullian: effectively immune sets can have effectively immune complements. The main theorem improves Arslanov’s, effectively hyperimmune sets cannot have effectively hyperimmune complements: the improvement omits the second ‘effectively’. Two natural examples of strongly effectively immune sets are presented with newcases of the first proved herein. The first is the set of minimal-Blum-size programs for the partial computable functions; the second, the set of Kolmogorov-random strings. A proved, natural example is presented of an effectively dense simple, not strongly effectively simple set; its complement is a set of maximal run-times. Further motivations for this study are presented. Kleene recursion theorem proofs herein emphasize how to conceptualize them. Finally, is suggested, future, related work—illustrated by a first, natural, strongly effectively 0 2 -immune set—included: solution of an open problem from Rogers’ book.
Structural analysis of bijels stabilized by magnetically responsive ellipsoidal particles
(Physics of Fluids, 2025-08-01) Karthikeyan, Nikhil; Schiller, Ulf D.
Bicontinuous interfacially jammed emulsion gels (bijels) offer a versatile platform for emulsion templating of functional porous materials including membranes, electrodes, and tissue-mimetic biomaterials. In many applications of such materials, the microstructure determines the properties and performance of devices. Characterization of the morphological structure of emulsion templates is thus an important step in developing fabrication methods for porous materials with tunable microstructure. We present a structural analysis of bijels stabilized by magnetic ellipsoidal particles. Using data from hybrid Lattice Boltzmann-Molecular Dynamics simulations of a binary liquid with suspended magnetic ellipsoidal particles, we analyze the bond orientational order within the interfacial particle layer, the mean and Gaussian curvature of the interfaces, and the topological properties of the emulsion morphology. The results suggests that the particle packing at the interface is influenced by the local topology as characterized by the Gaussian curvature, and the global topological properties can be linked to domain coarsening mechanisms such as coalescence of domains and pinch-off of channels. By analyzing independent simulation runs with different initial conditions, we probe the statistical variations of different properties, including the channel size distribution and the average channel size. Our analysis provides a more detailed picture of the structural properties of bijels stabilized by magnetically responsive ellipsoids and can guide the optimization of interfacial particle packing and domain structure of particle-stabilized emulsion systems.
Prognostic assessment of osteolytic lesions and mechanical properties of bones bearing breast cancer using neural network and finite element analysis
(Mechanobiology in Medicine, 2025-04-10) Wang, Shubo; Chu, Tiankuo; Wasi, Murtaza; Guerra, Rosa M.; Yuan, Xu; Wang, Liyun
The management of skeletal-related events (SREs), particularly the prevention of pathological fractures, is crucial for cancer patients. Current clinical assessment of fracture risk is mostly based on medical images, but incorporating sequential images in the assessment remains challenging. This study addressed this issue by leveraging a comprehensive dataset consisting of 260 longitudinal micro-computed tomography (μCT) scans acquired in normal and breast cancer bearing mice. A machine learning (ML) model based on a spatial–temporal neural network was built to forecast bone structures from previous μCT scans, which were found to have an overall similarity coefficient (Dice) of 0.814 with ground truths. Despite the predicted lesion volumes (18.5 % ± 15.3 %) being underestimated by ∼21 % than the ground truths’ (22.1 % ± 14.8 %), the time course of the lesion growth was better represented in the predicted images than the preceding scans (10.8 % ± 6.5 %). Under virtual biomechanical testing using finite element analysis (FEA), the predicted bone structures recapitulated the loading carrying behaviors of the ground truth structures with a positive correlation (y = 0.863x) and a high coefficient of determination (R2 = 0.955). Interestingly, the compliances of the predicted and ground truth structures demonstrated nearly identical linear relationships with the lesion volumes. In summary, we have demonstrated that bone deterioration could be proficiently predicted using machine learning in our preclinical dataset, suggesting the importance of large longitudinal clinical imaging datasets in fracture risk assessment for cancer bone metastasis. Graphical abstract available at: https://doi.org/10.1016/j.mbm.2025.100130 Highlights • Fracture risk assessment is critical in managing bone cancer metastasis. • Utilized 260 longitudinal μCT scans of both normal mice and cancer-bearing mice. • Bone lesion progression predicted from μCT scans using machine learning (ML). • Finite element analysis (FEA) revealed the rigidity of the predicted bone structures. • Predictive modeling of bone deterioration is a valuable tool to assess fracture risk.
A Machine Learning Model for Post-Concussion Musculoskeletal Injury Risk in Collegiate Athletes
(Sports Medicine, 2025-03-27) Claros, Claudio C.; Anderson, Melissa N.; Qian, Wei; Brockmeier, Austin J.; Buckley, Thomas A.
Background Emerging evidence indicates an elevated risk of post-concussion musculoskeletal injuries in collegiate athletes; however, identifying athletes at highest risk remains to be elucidated. Objective The purpose of this study was to model post-concussion musculoskeletal injury risk in collegiate athletes by integrating a comprehensive set of variables by machine learning. Methods A risk model was developed and tested on a dataset of 194 athletes (155 in the training set and 39 in the test set) with 135 variables entered into the analysis, which included participant’s heath and athletic history, concussion injury and recovery-specific criteria, and outcomes from a diverse array of concussion assessments. The machine learning approach involved transforming variables by the weight of evidence method, variable selection using L1-penalized logistic regression, model selection via the Akaike Information Criterion, and a final L2-regularized logistic regression fit. Results A model with 48 predictive variables yielded significant predictive performance of subsequent musculoskeletal injury with an area under the curve of 0.82. Top predictors included cognitive, balance, and reaction at baseline and acute timepoints. At a specified false-positive rate of 6.67%, the model achieves a true-positive rate (sensitivity) of 79% and a precision (positive predictive value) of 95% for identifying at-risk athletes via a well-calibrated composite risk score. Conclusions These results support the development of a sensitive and specific injury risk model using standard data combined with a novel methodological approach that may allow clinicians to target high injury risk student athletes. The development and refinement of predictive models, incorporating machine learning and utilizing comprehensive datasets, could lead to improved identification of high-risk athletes and allow for the implementation of targeted injury risk reduction strategies by identifying student athletes most at risk for post-concussion musculoskeletal injury. Key Points - There is a well-established elevated risk of post-concussion subsequent musculoskeletal injury; however, prior efforts have failed to identify risk factors. - This study developed a composite risk score model with an area under the curve of 0.82 from common concussion clinical measures and participant demographics. - By identifying athletes at elevated risk, clinicians may be able to reduce injury risk through targeted injury risk reduction programs.
"I spent 14 hours debugging just one assignment": Toward Computer-Mediated Personal Informatics for Computer Science Student Mental Health
(Proceedings of the 2025 CHI Conference on Human Factors in Computing Systems, 2025-04-25) Chandrasekaran, Aishwarya; Bielicke, London; Shah, Diya; Janakiraman, Harisha; Mauriello, Matthew Louis
Anxiety and depression rates in Computer Science (CS) students are double those of other undergraduates and 5-10 times higher than the general population. However, factors contributing to the elevated mental health issues in CS students remain unknown. To bridge this gap, we conducted need-finding interviews (N=20), which revealed that the complexity of debugging, along with imposter syndrome, are key contributors to stress and burnout. Participants expressed openness toward and feature preferences in a computer-based Personal Informatics (PI) tool to facilitate self-reflection. In response, we developed EmotionStream, an algorithm-assisted PI tool that provides both contextual and emotional insights based on individual behaviors. We found that participants rated their experience with the tool highly. Post-hoc analysis revealed that emotional states, augmented with contextual cues, show promise of predicting real-time stress. Based on our findings, we provide design implications for future PI tools to support CS student mental well-being.
Molecular-Scale Simulation of Wetting of Actin Filaments by Protein Droplets
(The Journal of Physical Chemistry B, 2025-01-12) Andrews, James; Weirich, Kimberly; Schiller, Ulf D.
Liquid phase-separating proteins can form condensates that play an important role in spatial and temporal organization of biological cells. The understanding of the mechanisms that lead to the formation of protein condensates and their interactions with other biomolecules may lead to processing routes for soft materials with tailored geometry and function. Fused in sarcoma (FUS) is an example of a nuclear protein that forms stable complexes, and recent studies have highlighted its ability to wet actin filaments and bundle them into networks. We perform coarse-grained molecular dynamics simulations to investigate the wetting and spreading of FUS droplets on actin filaments. We employ the Martini model and rescale the protein–protein and protein–actin interactions to tune the interfacial and wetting properties of FUS droplets. By measuring the molecular displacements in the three-phase region, we are able to relate contact angle, contact line velocity, and contact line friction in terms of a linear approximation of molecular kinetic theory. The results show that the rescaled Martini model can be used to study the molecular mechanisms of dynamic wetting at the nanoscale and to obtain quantitative predictions of the contact line friction and contact angles during dynamic wetting.
Advances in Set Function Learning: A Survey of Techniques and Applications
(ACM Computing Surveys, 2025-02-21) Xie, Jiahao; Tong, Guangmo
Set function learning has emerged as a crucial area in machine learning, addressing the challenge of modeling functions that take sets as inputs. Unlike traditional machine learning that involves fixed-size input vectors where the order of features matters, set function learning demands methods that are invariant to permutations of the input set, presenting a unique and complex problem. This survey provides a comprehensive overview of the current development in set function learning, covering foundational theories, key methodologies, and diverse applications. We categorize and discuss existing approaches, focusing on deep learning approaches, such as DeepSets and Set Transformer-based methods, as well as other notable alternative methods beyond deep learning, offering a complete view of current models. We also introduce various applications and relevant datasets, such as point cloud processing and multi-label classification, highlighting the significant progress achieved by set function learning methods in these domains. Finally, we conclude by summarizing the current state of set function learning approaches and identifying promising future research directions, aiming to guide and inspire further advancements in this promising field.
Constructivity conditions on immune sets
(Archive for Mathematical Logic, 2025-01-29) Case, John
Definitionally: strongly effectively immune sets are infinite and their c.e. subsets have maximums effectively bounded in their c.e. indices; whereas, for effectively immune sets, their c.e. subsets’ cardinalities are what’re effectively bounded. This definitional difference between these two kinds of sets is very nicely paralleled by the following difference between their complements. McLaughlin: strongly effectively immune sets cannot have immune complements; whereas, the main theorem herein: effectively immune sets cannot have hyperimmune complements. Ullian: effectively immune sets can have effectively immune complements. The main theorem improves Arslanov’s, effectively hyperimmune sets cannot have effectively hyperimmune complements: the improvement omits the second ‘effectively’. Two natural examples of strongly effectively immune sets are presented with new cases of the first proved herein. The first is the set of minimal-Blum-size programs for the partial computable functions; the second, the set of Kolmogorov-random strings. A proved, natural example is presented of an effectively dense simple, not strongly effectively simple set; its complement is a set of maximal run-times. Further motivations for this study are presented. Kleene recursion theorem proofs herein emphasize how to conceptualize them. Finally, is suggested, future, related work—illustrated by a first, natural, strongly effectively -immune set—included: solution of an open problem from Rogers’ book.
Understanding the liver under heat stress with statistical learning: an integrated metabolomics and transcriptomics computational approach
(BMC Genomics, 2019-06-17) Hubbard, Allen H.; Zhang, Xiaoke; Jastrebski, Sara; Singh, Abhyudai; Schmidt, Carl J.
Background We present results from a computational analysis developed to integrate transcriptome and metabolomic data in order to explore the heat stress response in the liver of the modern broiler chicken. Heat stress is a significant cause of productivity loss in the poultry industry, both in terms of increased livestock morbidity and its negative influence on average feed efficiency. This study focuses on the liver because it is an important regulator of metabolism, controlling many of the physiological processes impacted by prolonged heat stress. Using statistical learning methods, we identify genes and metabolites that may regulate the heat stress response in the liver and adaptations required to acclimate to prolonged heat stress. Results We describe how disparate systems such as sugar, lipid and amino acid metabolism, are coordinated during the heat stress response. Conclusions Our findings provide more detailed context for genomic studies and generates hypotheses about dietary interventions that can mitigate the negative influence of heat stress on the poultry industry.
Transcriptome response to heat stress in a chicken hepatocellular carcinoma cell line
(Cell Stress and Chaperones, 2024-01-05) Sun, Liang; Lamont, Susan J.; Cooksey, Amanda M.; McCarthy, Fiona; Tudo, Catalina O.; Vijay-Shanker, K.; DeRita, Rachael M.; Rothschild, Max; Ashwell, Chris; Persia, Michael E.; Schmidt, Carl J.
Heat stress triggers an evolutionarily conserved set of responses in cells. The transcriptome responds to hyperthermia by altering expression of genes to adapt the cell or organism to survive the heat challenge. RNA-seq technology allows rapid identification of environmentally responsive genes on a large scale. In this study, we have used RNA-seq to identify heat stress responsive genes in the chicken male white leghorn hepatocellular (LMH) cell line. The transcripts of 812 genes were responsive to heat stress (p < 0.01) with 235 genes upregulated and 577 downregulated following 2.5 h of heat stress. Among the upregulated were genes whose products function as chaperones, along with genes affecting collagen synthesis and deposition, transcription factors, chromatin remodelers, and genes modulating the WNT and TGF-beta pathways. Predominant among the downregulated genes were ones that affect DNA replication and repair along with chromosomal segregation. Many of the genes identified in this study have not been previously implicated in the heat stress response. These data extend our understanding of the transcriptome response to heat stress with many of the identified biological processes and pathways likely to function in adapting cells and organisms to hyperthermic stress. Furthermore, this study should provide important insight to future efforts attempting to improve species abilities to withstand heat stress through genome-wide association studies and breeding.
MRI-based whole-brain elastography and volumetric measurements to predict brain age
(Biology Methods and Protocols, 2024-11-20) Claros-Olivares, Claudio Cesar; Clements, Rebecca G.; McIlvain, Grace; Johnson, Curtis L.; Brockmeier, Austin J.
Brain age, as a correlate of an individual’s chronological age obtained from structural and functional neuroimaging data, enables assessing developmental or neurodegenerative pathology relative to the overall population. Accurately inferring brain age from brain magnetic resonance imaging (MRI) data requires imaging methods sensitive to tissue health and sophisticated statistical models to identify the underlying age-related brain changes. Magnetic resonance elastography (MRE) is a specialized MRI technique which has emerged as a reliable, non-invasive method to measure the brain’s mechanical properties, such as the viscoelastic shear stiffness and damping ratio. These mechanical properties have been shown to change across the life span, reflect neurodegenerative diseases, and are associated with individual differences in cognitive function. Here, we aim to develop a machine learning framework to accurately predict a healthy individual’s chronological age from maps of brain mechanical properties. This framework can later be applied to understand neurostructural deviations from normal in individuals with neurodevelopmental or neurodegenerative conditions. Using 3D convolutional networks as deep learning models and more traditional statistical models, we relate chronological age as a function of multiple modalities of whole-brain measurements: stiffness, damping ratio, and volume. Evaluations on held-out subjects show that combining stiffness and volume in a multimodal approach achieves the most accurate predictions. Interpretation of the different models highlights important regions that are distinct between the modalities. The results demonstrate the complementary value of MRE measurements in brain age models, which, in future studies, could improve model sensitivity to brain integrity differences in individuals with neuropathology.
Opportunities and Pitfalls with Large Language Models for Biomedical Annotation
(Biocomputing 2025, 2024-12) Arighi, Cecilia; Kim, Jin-Dong; Lu, Zhiyong; Rinaldi, Fabio
Large language models (LLMs) and biomedical annotations have a symbiotic relationship. LLMs rely on high-quality annotations for training and/or fine-tuning for specific biomedical tasks. These annotations are traditionally generated through expensive and time-consuming human curation. Meanwhile LLMs can also be used to accelerate the process of curation, thus simplifying the process, and potentially creating a virtuous feedback loop. However, their use also introduces new limitations and risks, which are as important to consider as the opportunities they offer. In this workshop, we will review the process that has led to the current rise of LLMs in several fields, and in particular in biomedicine, and discuss specifically the opportunities and pitfalls when they are applied to biomedical annotation and curation.
Formation of bijels stabilized by magnetic ellipsoidal particles in external magnetic fields
(Soft Matter, 2024-10-08) Karthikeyan, Nikhil; Schiller, Ulf D.
Bicontinuous interfacially-jammed emulsion gels (bijels) are increasingly used as emulsion templates for the fabrication of functional porous materials including membranes, electrodes, and biomaterials. Control over the domain size and structure is highly desirable in these applications. For bijels stabilized by spherical particles, particle size and volume fraction are the main parameters that determine the emulsion structure. Here, we investigate the use of ellipsoidal magnetic particles and study the effect of external magnetic fields on the formation of bijels. Using hybrid Lattice Boltzmann-molecular dynamics simulations, we analyze the effect of the magnetic field on emulsion dynamics and the structural properties of the resulting bijel. We find that the formation of bijels remains robust in the presence of magnetic fields, and that the domain size and tortuosity become anisotropic when ellipsoidal particles are used. We show that the magnetic fields lead to orientational ordering of the particles which in turn leads to alignment of the interfaces. The orientational order facilitates enhanced packing of particles in the interface which leads to different jamming times in the directions parallel and perpendicular to the field. Our results highlight the potential of magnetic particles for fabrication and processing of emulsion systems with tunable properties.
Soybean Bradyrhizobium spp. Spontaneously Produce Abundant and Diverse Temperate Phages in Culture
(Viruses, 2024-11-07) Richards, Vanessa A.; Ferrell, Barbra D.; Polson, Shawn W.; Wommack, K. Eric; Fuhrmann, Jeffry J.
Soybean bradyrhizobia (Bradyrhizobium spp.) are symbiotic root-nodulating bacteria that fix atmospheric nitrogen for the host plant. The University of Delaware Bradyrhizobium Culture Collection (UDBCC; 353 accessions) was created to study the diversity and ecology of soybean bradyrhizobia. Some UDBCC accessions produce temperate (lysogenic) bacteriophages spontaneously under routine culture conditions without chemical or other apparent inducing agents. Spontaneous phage production may promote horizontal gene transfer and shape bacterial genomes and associated phenotypes. A diverse subset (n = 98) of the UDBCC was examined for spontaneously produced virus-like particles (VLPs) using epifluorescent microscopy, with a majority (69%) producing detectable VLPs (>1 × 107 mL−1) in laboratory culture. Phages from the higher-producing accessions (>2.0 × 108 VLP mL−1; n = 44) were examined using transmission electron microscopy. Diverse morphologies were observed, including various tail types and lengths, capsid sizes and shapes, and the presence of collars or baseplates. In many instances, putative extracellular vesicles of a size similar to virions were also observed. Three of the four species examined (B. japonicum, B. elkanii, and B. diazoefficiens) produced apparently tailless phages. All species except B. ottawaense also produced siphovirus-like phages, while all but B. diazoefficiens additionally produced podovirus-like phages. Myovirus-like phages were restricted to B. japonicum and B. elkanii. At least three strains were polylysogens, producing up to three distinct morphotypes. These observations suggest spontaneously produced phages may play a significant role in the ecology and evolution of soybean bradyrhizobia.
A long-term high-fat diet induces differential gene expression changes in spatially distinct adipose tissue of male mice
(Physiological Genomics, 2024-11-11) Alradi, Malak; Askari, Hassan; Shaw, Mark; Bhavsar, Jaysheel D.; Kingham, Brewster F.; Polson, Shawn W.; Fancher, Ibra S.
The accumulation of visceral adipose tissue (VAT) is strongly associated with cardiovascular disease and diabetes. In contrast, individuals with increased subcutaneous adipose tissue (SAT) without corresponding increases in VAT are associated with a metabolic healthy obese phenotype. These observations implicate dysfunctional VAT as a driver of disease processes, warranting investigation into obesity-induced alterations of distinct adipose depots. To determine the effects of obesity on adipose gene expression, male mice (n = 4) were fed a high-fat diet to induce obesity or a normal laboratory diet (lean controls) for 12–14 mo. Mesenteric VAT and inguinal SAT were isolated for bulk RNA sequencing. AT from lean controls served as a reference to obesity-induced changes. The long-term high-fat diet induced the expression of 169 and 814 unique genes in SAT and VAT, respectively. SAT from obese mice exhibited 308 differentially expressed genes (164 upregulated and 144 downregulated). VAT from obese mice exhibited 690 differentially expressed genes (262 genes upregulated and 428 downregulated). KEGG pathway and GO analyses revealed that metabolic pathways were upregulated in SAT versus downregulated in VAT while inflammatory signaling was upregulated in VAT. We next determined common genes that were differentially regulated between SAT and VAT in response to obesity and identified four genes that exhibited this profile: elovl6 and kcnj15 were upregulated in SAT/downregulated in VAT while trdn and hspb7 were downregulated in SAT/upregulated in VAT. We propose that these genes in particular should be further pursued to determine their roles in SAT versus VAT with respect to obesity. NEW & NOTEWORTHY A long-term high-fat diet induced the expression of more than 980 unique genes across subcutaneous adipose tissue (SAT) and visceral adipose tissue (VAT). The high-fat diet also induced the differential expression of nearly 1,000 AT genes. We identified four genes that were oppositely expressed in SAT versus VAT in response to the high-fat diet and propose that these genes in particular may serve as promising targets aimed at resolving VAT dysfunction in obesity.
Heterogeneous Task Oriented Data Scheduling in Vehicular Edge Computing via Deep Reinforcement Learning
(IEEE Transactions on Vehicular Technology, 2024-08-16) Luo, Quyuan; Luan, Tom H.; Shi, Weisong; Fan, Pingzhi
In vehicular edge computing environment, massive computation-intensive tasks would be produced from diverse vehicular applications. Data scheduling among vehicles and roadside units(RSUs) is a fundamental issue in timely processing those tasks. However, the task heterogeneity with different computation resource requirements and delay constraints, the distinct capacities of vehicles and RSUs, and the stochastic task arrival, pose significant challenges in realizing efficient data scheduling. The existing literature ignores the multi-core feature of both vehicles and RSUs in data scheduling, which may lead to an inefficient resource usage. To cope with these challenges, in this paper, we first construct a multi-queue multi-block model for heterogeneous task oriented data caching on both vehicle and RSU sides. By fully utilizing the multi-core features of both vehicles and RSUs, a fine-grained offloading model is then developed, involving the association between data blocks and computing cores, and the allocation of computation and communication resources. After that, a long-term loss minimization problem is formulated to facilitate data processing. We leverage the Markov decision process (MDP) to model the optimization problem, which is then solved by our proposed deep deterministic policy gradient (DDPG) based association mapping and resource allocation algorithm (D-AMRA). In D-AMRA, an action transformation method is proposed to map the outputs of DDPG to the form of optimization variables. Eventually, extensive simulations with comparative benchmarks are conducted to evaluate the effectiveness of our proposed D-AMRA.
Enhancing severe hypoglycemia prediction in type 2 diabetes mellitus through multi-view co-training machine learning model for imbalanced dataset
(Scientific Reports, 2024-09-30) Agraz, Melih; Deng, Yixiang; Karniadakis, George Em; Mantzoros, Christos Socrates
Patients with type 2 diabetes mellitus (T2DM) who have severe hypoglycemia (SH) poses a considerable risk of long-term death, especially among the elderly, demanding urgent medical attention. Accurate prediction of SH remains challenging due to its multifaced nature, contributed from factors such as medications, lifestyle choices, and metabolic measurements. In this study, we propose a systematic approach to improve the robustness and accuracy of SH predictions using machine learning models, guided by clinical feature selection. Our focus is on developing long-term SH prediction models using both semi-supervised learning and supervised learning algorithms. Using the action to control cardiovascular risk in diabetes trial, which includes electronic health records for over 10,000 individuals, we focus on studying adults with T2DM. Our results indicate that the application of a multi-view co-training method, incorporating the random forest algorithm, improves the specificity of SH prediction, while the same setup with Naive Bayes replacing random forest demonstrates better sensitivity. Our framework also provides interpretability of machine learning models by identifying key predictors for hypoglycemia, including fasting plasma glucose, hemoglobin A1c, general diabetes education, and NPH or L insulins. The integration of data routinely available in electronic health records significantly enhances our model’s capability to predict SH events, showcasing its potential to transform clinical practice by facilitating early interventions and optimizing patient management. By enhancing prediction accuracy and identifying crucial predictive features, our study contributes to advancing the understanding and management of hypoglycemia in this population.
A Simple Mobile Plausibly Deniable System Using Image Steganography and Secure Hardware
(Proceedings of the 2024 ACM Workshop on Secure and Trustworthy Cyber-Physical Systems, 2024-06-19) Xia, Lichen; Liao, Jinghui; Chen, Niusen; Chen, Bo; Shi, Weisong
Traditional encryption methods cannot defend against coercive attacks in which the adversary captures both the user and the possessed computing device, and forces the user to disclose the decryption keys. Plausibly deniable encryption (PDE) has been designed to defend against this strong coercive attacker. At its core, PDE allows the victim to plausibly deny the very existence of hidden sensitive data and the corresponding decryption keys upon being coerced. Designing an efficient PDE system for a mobile platform, however, is challenging due to various design constraints bound to the mobile systems. Leveraging image steganography and the built-in hardware security feature of mobile devices, namely TrustZone, we have designed a Simple Mobile Plausibly Deniable Encryption (SMPDE) system which can combat coercive adversaries and, meanwhile, is able to overcome unique design constraints. In our design, the encoding/decoding process of image steganography is bounded together with Arm TrustZone. In this manner, the coercive adversary will be given a decoy key, which can only activate a DUMMY trusted application that will instead sanitize the sensitive information stored hidden in the stego-image upon decoding. On the contrary, the actual user can be given the true key, which can activate the PDE trusted application that can really extract the sensitive information from the stego-image upon decoding. Security analysis and experimental evaluation justify both the security and the efficiency of our design.

Browse

Recent Submissions