Showing papers by "San Jose State University published in 2022"

The lifetime risk of revision following total knee arthroplasty

Abstract: The success of total knee arthroplasty (TKA) is usually measured using functional outcome scores and revision-free survivorship. However, reporting the lifetime risk of revision may be more meaningful to patients when gauging risks, especially in younger patients. We aimed to assess the lifetime risk of revision for patients in different age categories at the time of undergoing primary TKA.The New Zealand Joint Registry database was used to obtain revision rates, mortality, and the indications for revision for all primary TKAs performed during an 18-year period between January 1999 and December 2016. Patients were stratified into age groups at the time of the initial TKA, and the lifetime risk of revision was calculated according to age, sex, and the American Society of Anesthesiologists (ASA) grade. The most common indications for revision were also analyzed for each age group.The overall ten-year survival rate was 95.6%. This was lowest in the youngest age group (between 46 and 50 years) and increased sequentially with increasing age. The lifetime risk of requiring revision was 22.4% in those aged between 46 and 50 years at the time of the initial surgery, and decreased linearly with increasing age to 1.15% in those aged between 90 and 95 years at the time of surgery. Higher ASA grades were associated with increased lifetime risk of revision in all age groups. The three commonest indications for revision were aseptic loosening, infection, and unexplained pain. Young males, aged between 46 and 50 years, had the highest lifetime risk of revision (25.2%).Lifetime risk of revision may be a more meaningful measure of outcome than implant survival at defined time periods when counselling patients prior to TKA. This study highlights the considerably higher lifetime risk of revision surgery for all indications, including infection, in younger male patients. Cite this article: Bone Joint J 2022;104-B(2):235-241.

A review of factors influencing the quality and sensory evaluation techniques applied to Greek yogurt

Abstract: Greek yogurt is one of the fastest growing products in the dairy industry. It is also known as strained yogurt, which is obtained after draining the whey. As a result of the draining process, Greek yogurt has higher total solids and lower lactose than regular yogurt. Since it is a concentrated yogurt, its sensory characteristics are different from regular yogurt. However, there is little information about factors influencing the quality of Greek yogurt and sensory evaluation techniques applied to Greek yogurt. This review aims to describe the effects of ingredients, starter cultures, processing techniques and other parameters on quality characteristics and sensory properties of Greek yogurt. In addition, advantages and limitations of novel sensory evaluation techniques applied to Greek yogurt products are discussed. In particular, we take a look at advanced techniques such as the electronic nose and electronic tongue and the benefits of these techniques with regard to Greek yogurt. This review should help the Greek yogurt industry to improve its current products and develop innovative products based on appropriate food evaluation techniques.

Convolutional Long-Short Term Memory Network with Multi-Head Attention Mechanism for Traffic Flow Prediction

Abstract: Accurate predictive modeling of traffic flow is critically important as it allows transportation users to make wise decisions to circumvent traffic congestion regions. The advanced development of sensing technology makes big data more affordable and accessible, meaning that data-driven methods have been increasingly adopted for traffic flow prediction. Although numerous data-driven methods have been introduced for traffic flow predictions, existing data-driven methods cannot consider the correlation of the extracted high-dimensional features and cannot use the most relevant part of the traffic flow data to make predictions. To address these issues, this work proposes a decoder convolutional LSTM network, where the convolutional operation is used to consider the correlation of the high-dimensional features, and the LSTM network is used to consider the temporal correlation of traffic flow data. Moreover, the multi-head attention mechanism is introduced to use the most relevant portion of the traffic data to make predictions so that the prediction performance can be improved. A traffic flow dataset collected from the Caltrans Performance Measurement System (PeMS) database is used to demonstrate the effectiveness of the proposed method.

Identification of Distinct Characteristics of Antibiofilm Peptides and Prospection of Diverse Sources for Efficacious Sequences

Abstract: A majority of microbial infections are associated with biofilms. Targeting biofilms is considered an effective strategy to limit microbial virulence while minimizing the development of antibiotic resistance. Toward this need, antibiofilm peptides are an attractive arsenal since they are bestowed with properties orthogonal to small molecule drugs. In this work, we developed machine learning models to identify the distinguishing characteristics of known antibiofilm peptides, and to mine peptide databases from diverse habitats to classify new peptides with potential antibiofilm activities. Additionally, we used the reported minimum inhibitory/eradication concentration (MBIC/MBEC) of the antibiofilm peptides to create a regression model on top of the classification model to predict the effectiveness of new antibiofilm peptides. We used a positive dataset containing 242 antibiofilm peptides, and a negative dataset which, unlike previous datasets, contains peptides that are likely to promote biofilm formation. Our model achieved a classification accuracy greater than 98% and harmonic mean of precision-recall (F1) and Matthews correlation coefficient (MCC) scores greater than 0.90; the regression model achieved an MCC score greater than 0.81. We utilized our classification-regression pipeline to evaluate 135,015 peptides from diverse sources for potential antibiofilm activity, and we identified 185 candidates that are likely to be effective against preformed biofilms at micromolar concentrations. Structural analysis of the top 37 hits revealed a larger distribution of helices and coils than sheets, and common functional motifs. Sequence alignment of these hits with known antibiofilm peptides revealed that, while some of the hits showed relatively high sequence similarity with known peptides, some others did not indicate the presence of antibiofilm activity in novel sources or sequences. Further, some of the hits had previously recognized therapeutic properties or host defense traits suggestive of drug repurposing applications. Taken together, this work demonstrates a new in silico approach to predicting antibiofilm efficacy, and identifies promising new candidates for biofilm eradication.

A multitask deep learning approach for pulmonary embolism detection and identification

Abstract: Pulmonary embolism (PE) is a blood clot traveling to the lungs and is associated with substantial morbidity and mortality. Therefore, rapid diagnoses and treatments are essential. Chest computed tomographic pulmonary angiogram (CTPA) is the gold standard for PE diagnoses. Deep learning can enhance the radiologists'workflow by identifying PE using CTPA, which helps to prioritize important cases and hasten the diagnoses for at-risk patients. In this study, we propose a two-phase multitask learning method that can recognize the presence of PE and its properties such as the position, whether acute or chronic, and the corresponding right-to-left ventricle diameter (RV/LV) ratio, thereby reducing false-negative diagnoses. Trained on the RSNA-STR Pulmonary Embolism CT Dataset, our model demonstrates promising PE detection performances on the hold-out test set with the window-level AUROC achieving 0.93 and the sensitivity being 0.86 with a specificity of 0.85, which is competitive with the radiologists'sensitivities ranging from 0.67 to 0.87 with specificities of 0.89-0.99. In addition, our model provides interpretability through attention weight heatmaps and gradient-weighted class activation mapping (Grad-CAM). Our proposed deep learning model could predict PE existence and other properties of existing cases, which could be applied to practical assistance for PE diagnosis.

Secure human action recognition by encrypted neural network inference

Abstract: Abstract Advanced computer vision technology can provide near real-time home monitoring to support “aging in place” by detecting falls and symptoms related to seizures and stroke. Affordable webcams, together with cloud computing services (to run machine learning algorithms), can potentially bring significant social benefits. However, it has not been deployed in practice because of privacy concerns. In this paper, we propose a strategy that uses homomorphic encryption to resolve this dilemma, which guarantees information confidentiality while retaining action detection. Our protocol for secure inference can distinguish falls from activities of daily living with 86.21% sensitivity and 99.14% specificity, with an average inference latency of 1.2 seconds and 2.4 seconds on real-world test datasets using small and large neural nets, respectively. We show that our method enables a 613x speedup over the latency-optimized LoLa and achieves an average of 3.1x throughput increase in secure inference compared to the throughput-optimized nGraph-HE2.

Free-Text Keystroke Dynamics for User Authentication

Abstract: In this research, we consider the problem of verifying user identity based on keystroke dynamics obtained from free-text. We employ a novel feature engineering method that generates image-like transition matrices. For this image-like feature, a convolution neural network (CNN) with cutout achieves the best results. A hybrid model consisting of a CNN and a recurrent neural network (RNN) is also shown to outperform previous research in this field.

How older adults use Ride-hailing booking technology in California

Abstract: Ride-hailing services like Lyft and Uber have the potential to improve mobility for many older adults, especially those who cannot or prefer not to drive. We used survey data from 2,917 Californians 55 years and older to investigate (1) how older adults who currently ride-hail booked their trips, and (2) what personal characteristics, including attitudes towards technology, were correlated with booking trips online versus by phone or with help. We specified four binary probit models in which the outcome variables are the manner in which a respondent accessed ride-hailing services: self-booked by phone, self-booked by app, booked by a family/friend/caregiver but rode alone, or booked by others and rode with them. We controlled for two attitudinal constructs (confident and cautious about technology), residential location, general travel behavior, physical health, and standard socio-economic factors. We found that respondents who were more confident using technology booked via apps, while those who were more cautious about technology were less likely to book using apps. This latter group was more likely to book by phone or rely on others for help. Other characteristics associated with higher likelihood of booking via apps were: living in the suburbs, not relying on others for rides, having physical health issues, being college educated, and being non-Hispanic. Our findings provide a basis to think about expanding ride-hailing to other older adults, particularly those who are not comfortable with technology, through convenient access to ride-hailing booking.

Machine Learning and Deep Learning for Fixed-Text Keystroke Dynamics

Abstract: Keystroke dynamics can be used to analyze the way that users type by measuring various aspects of keyboard input. Previous work has demonstrated the feasibility of user authentication and identification utilizing keystroke dynamics. In this research, we consider a wide variety of machine learning and deep learning techniques based on fixed-text keystroke-derived features, we optimize the resulting models, and we compare our results to those obtained in related research. We find that models based on extreme gradient boosting (XGBoost) and multi-layer perceptrons (MLP) perform well in our experiments. Our best models outperform previous comparable research.

Clickbait Detection for YouTube Videos

Abstract: YouTube videos often include captivating descriptions and intriguing thumbnails designed to increase the number of views, and thereby increase the revenue for the person who posted the video. This creates an incentive for people to post clickbait videos, in which the content might deviate significantly from the title, description, or thumbnail. In effect, users are tricked into clicking on clickbait videos. In this research, we consider the challenging problem of detecting clickbait YouTube videos. We experiment with multiple state-of-the-art machine learning techniques using a variety of textual features.

Addressing Cultural Competency and Primary Palliative Care Needs in Community Health Nursing Education

Abstract: To meet the primary palliative care needs of older adults, especially the underserved and those of color, a qualitative descriptive study was conducted among nursing students (n = 34) in community health nursing whose clinical rotations included independent living facilities for low-income populations. Two themes were identified that influence student nurse confidence and self-efficacy in initiating conversations with seriously ill clients and assessing their primary palliative care needs: (1) emotional assessment skills and cultural competency and (2) an assessment guide to primary palliative care needs that provides structure and guidance to conduct such conversations. Two major categories were identified that influence student nurse confidence and self-efficacy in initiating conversations with seriously ill clients: (1) educational needs and desired experience and (2) cultural impact. Two major categories were identified that facilitated the assessment of primary palliative care needs: (1) a guide that provided structure and guidance in initiating and conducting conversations about serious illness and (2) insight into the primary palliative care needs of clients in the community.

Relating Machine Learning to the Real-World: Analogies to Enhance Learning Comprehension

Abstract: Machine learning is an exciting field for many, but the rigor, math, and its rapid evolution are often found to be formidable, keeping them away from studying and pursuing a career in this area. Similarity has been substantially explored in machine learning algorithms such as in the K-nearest neighbors, Kernel methods, Support Vector Machines, but not so much in human learning, particularly when it comes to teaching machine learning. In the course of teaching the subject to undergraduate, graduate, and general pool of students, the author found that relating the concepts to real-world examples greatly enhances student comprehension and makes the topics much more approachable despite the math and the methods involved. This paper relates some of the concepts, artifacts, and algorithms in machine learning such as overfitting, regularization, and Generative Adversarial Networks to the real world using illustrative examples. Most of the analogies included in the paper were well appreciated by the students in the course of the author’s teaching and acknowledged as enhancing comprehension. It is hoped that the material presented in this paper will benefit larger audiences, drawing more learners to the field, resulting in enhanced contributions to the area. The paper concludes by suggesting deep learning for automatically generating similarities and analogies as a future direction.

Peripherality in networks: theory and applications

Abstract: We investigate several related measures of peripherality and centrality for vertices and edges in networks, including the Mostar index which was recently introduced as a measure of peripherality for both edges and networks. We refute a conjecture on the maximum possible Mostar index of bipartite graphs. We asymptotically answer another problem on the maximum difference between the Mostar index and the irregularity of trees. We also prove a number of extremal bounds and computational complexity results about the Mostar index, irregularity, and measures of peripherality and centrality. We discuss graphs where the Mostar index is not an accurate measure of peripherality. We construct a general family of graphs with the property that the Mostar index is strictly greater for edges that are closer to the center. We also investigate centrality and peripherality in two graphs which represent the SuperFast and MOZART-4 systems of atmospheric chemical reactions by computing various measures of peripherality and centrality for the vertices and edges in these graphs. For both of these graphs, we find that the Mostar index is closer to a measure of centrality than peripherality of the edges. We also introduce some new indices which perform well as measures of peripherality on the SuperFast and MOZART-4 graphs.

Crowdfunding Non-fungible Tokens on the Blockchain

Abstract: Abstract Non-fungible tokens (NFTs) have been used as a way of rewarding content creators. Artists publish their works on the blockchain as NFTs, which they can then sell. The buyer of an NFT then holds ownership of a unique digital asset, which can be resold in much the same way that real-world art collectors might trade paintings. However, while a deal of effort has been spent on selling works of art on the blockchain, very little attention has been paid to using the blockchain as a means of fundraising to help finance the artist’s work in the first place. Additionally, while blockchains like Ethereum are ideal for smaller works of art, additional support is needed when the artwork is larger than is feasible to store on the blockchain. In this paper, we propose a fundraising mechanism that will help artists to gain financial support for their initiatives, and where the backers can receive a share of the profits in exchange for their support. We discuss our prototype implementation using the SpartanGold framework. We then discuss how this system could be expanded to support large NFTs with the 0Chain blockchain, and describe how we could provide support for ongoing storage of these NFTs.

Nicotine Affects Murine Aortic Stiffness and Fatigue Response During Supraphysiological Cycling

Abstract: Nicotine exposure is a major risk factor for several cardiovascular diseases. Although the deleterious effects of nicotine on aortic remodeling processes have been studied to some extent, the biophysical consequences are not fully elucidated. In this investigation, we applied quasi-static and dynamic loading to quantify ways in which exposure to nicotine affects the mechanical behavior of murine arterial tissue. Segments of thoracic aortas from C57BL/6 mice exposed to 25 mg/kg/day of subcutaneous nicotine for 28 days were subjected to uniaxial tensile loading in an open-circumferential configuration. Comparing aorta segments from nicotine-treated mice relative to an equal number of control counterparts, stiffness in the circumferential direction was nearly twofold higher (377 kPa ± 165 kPa versus 191 kPa ± 65 kPa, n = 5, p = 0.03) at 50% strain. Using a degradative power-law fit to fatigue data at supraphysiological loading, we observed that nicotine-treated aortas exhibited significantly higher peak stress, greater loss of tension, and wider oscillation band than control aortas (p ≤ 0.01 for all three variables). Compared to simple stress relaxation tests, fatigue cycling is shown to be more sensitive and versatile in discerning nicotine-induced changes in mechanical behavior over many cycles. Supraphysiological fatigue cycling thus may have broader potential to reveal subtle changes in vascular mechanics caused by other exogenous toxins or pathological conditions.

Auxiliary-Classifier GAN for Malware Analysis

Abstract: Generative adversarial networks (GAN) are a class of powerful machine learning techniques, where both a generative and discriminative model are trained simultaneously. GANs have been used, for example, to successfully generate “deep fake” images. A recent trend in malware research consists of treating executables as images and employing image-based analysis techniques. In this research, we generate fake malware images using auxiliary classifier GANs (AC-GAN), and we consider the effectiveness of various techniques for classifying the resulting images. Our results indicate that the resulting multiclass classification problem is challenging, yet we can obtain strong results when restricting the problem to distinguishing between real and fake samples. While the AC-GAN generated images often appear to be very similar to real malware images, we conclude that from a deep learning perspective, the AC-GAN generated samples do not rise to the level of deep fake malware images.

The Transition from Supercooled Liquid Water to Ice Crystals in Mixed-phase Clouds based on Airborne In-situ Observations

Abstract: Abstract. The on-set of ice nucleation in mixed-phase clouds determines cloud lifetime and their microphysical properties. In this work, we develop a novel method that differentiates the early and later transition phases of mixed-phase clouds, i.e., ice crystals are initially surrounded by supercooled liquid water droplets, then as they grow, pure ice segments are formed. Using this method, we examine the relationship between the macrophysical and microphysical properties of mixed-phase clouds. The results show that evolution of cloud macrophysical properties, represented by the increasing spatial ratio of regions containing ice crystals relative to the total in-cloud region (defined as ice spatial ratio), is positively correlated with the evolution of microphysical properties, represented by the increasing ice water content and decreasing liquid water content. The mass partition transition from liquid to ice becomes more significant during the later transition phase (i.e., transition phase 3) when pure ice cloud regions (ICRs) start to appear. Occurrence frequencies of cloud thermodynamic phases show significant transition from liquid to ice at a similar temperature (i.e., -17.5 °C) among three types of definitions of mixed-phase clouds based on ice mass fraction, ice number fraction, or ice spatial ratio. Aerosol indirect effects are quantified for different transition phases using number concentrations of aerosols greater than 100 nm or 500 nm (N>100 and N>500, respectively). N>500 shows stronger positive correlations with ice spatial ratios compared with N>100. This result indicates that larger aerosols potentially contain ice nucleating particles, which facilitate the formation of ice crystals in mixed-phase clouds. The impact of N>500 is also more significant on the earlier transition phase when ice crystals just start to appear compared with the later transition phase. The thermodynamic and dynamic conditions are quantified for each transition phase. The results show in-cloud turbulence as a main mechanism for both the initiation of ice nucleation and the maintenance of supercooled liquid water, while updrafts are important for the latter but not the former. Overall, these results illustrate the varying effects of aerosols, thermodynamics, and dynamics throughout cloud evolution based on this new method that categorizes cloud transition phases.