UCSD Research Articles

The actigraphic evaluation of daytime sleep in central disorders of hypersomnolence: comparison with polysomnography.

The role of actigraphy in central disorders of hypersomnolence (CDH) is expanding but evidence of reliability with polysomnography (PSG) is scarce and provided only during nighttime. We explored the agreement between actigraphy and continuous 24-hour PSG at CDH diagnosis. Forty-four consecutive drug-naïve patients (28 narcolepsy, 16 idiopathic hypersomnia) underwent actigraphy during 24 hours of free-running PSG, during multiple sleep latency test (MSLT) and 13 of them also during maintenance of wakefulness test (MWT). Daytime and nighttime sleep features and MSLT and MWT mean sleep latencies (mSL) were estimated with the actigraphic algorithms by Cole-Kripke (CK) Sadeh, and University of California San Diego (UCSD). Agreement to corresponding PSG measures was assessed with Bland Altman plots. Nighttime-total sleep time (TST) in narcolepsy was significantly underestimated with CK (bias 27.8min, 95%CI 13.7-41.9) and Sadeh (bias 56.7min, 95%CI 38.8/74.5). Daytime-TST was overestimated in IH and narcolepsy with all algorithms (CK: bias -42.2, 95%CI -67/-17.4; Sadeh: bias -30.2min, 95%CI -52.7/-7.7; UCSD bias -86.9min, 95%CI -118.2/-55.6). 24-hour-TST was overestimated by CK and UCSD in IH (CK: bias -58.5min, 95%CI -105.5/-11.5; UCSD: bias -118.8min, 95% CI -172.5/-65), and by UCSD in narcolepsy (bias -68.8min, 95%CI -109.3/-38.2). In the entire cohort, actigraphy overestimated MSLT mSL but not MWT mSL. Conventional actigraphic algorithms overestimate 24-hour TST in IH and underestimate nighttime TST in narcolepsy. These discrepancies call for cautious application of actigraphy in the diagnostic process of CDH and the development of new quantitative signal analysis approaches.

Importance of Spin Channels from Radical-Radical Reactions in Hydrogen-Oxygen Combustion Mechanisms at High Temperatures.

Radical-radical reactions can generate two channels with high and low spins. In this work, ten radical-radical reactions with different spin channels and four radical-molecule reactions in hydrogen-oxygen combustion were systematically investigated from a theoretical perspective. The potential energy surface (PES) of radical-radical reactions reveals that the high- and low-spin states of the reactant are energetically degenerate and the two channels are energetically feasible. The difference in rate constants between the high- and low-spin channels gradually decreases as the temperature increases. Then, the kinetic parameters of the 14 bimolecular reactions in the hydrogen-oxygen mechanism of the University of California, San Diego (UCSD), were replaced to simulate the ignition delay time and laminar flame speed. The simulation results agree well with the available experimental findings, indicating the necessity of considering both high- and low-spin channels for kinetic simulation.

Abstract 6393: The predictive role of TNF-related genes in patients receiving immune checkpoint inhibitors

Abstract Introduction: Not all patients respond to immune checkpoint inhibitors (ICI) even when they have favorable FDA-approved biomarkers. The tumor necrosis factor superfamily (TNFSF) and tumor necrosis factor receptor superfamily (TNFRSR) play critical roles in various biological processes, including inflammation, immune regulation, and cell death. In this study, we sought to investigate the predictive role of TNF-related genes in patients receiving ICI. Methods: The RNA expression levels of 18 TNF-related genes were assessed in a pan-cancer cohort of 217 patients treated with immune checkpoint inhibitors (ICI) at the University of California San Diego (UCSD) Moores Cancer Center. Using a reference population of 735 tumors and 35 histologies, transcript abundance was adjusted to internal housekeeping gene profiles and rated (0-100th percentile). The patients were clustered into two groups based on the 18 TNF-related genes expression using Ward’s hierarchical clustering. Associations between overall survival and previously demonstrated independent ICI predictive variables were assessed. Bonferroni correction for multiple comparisons was used and considered statistically significant when p-value ≤ 0.05. Results: Four TNF-related genes (CD137, CD70, GITR, and TNFSF9) were associated with positive programmed death-ligand 1 (PD-L1) expression (PD-L1≥1% immunohistochemistry TPS), while TNFSF18 was associated with negative PD-L1 expression (corrected p-value<0.05). No gene was associated with microsatellite instability-high (MSI-H) or high tumor mutational burden (TMB-H) (TMB≥10 mutations/Mb) (corrected p-value>0.05). We clustered the patients based on their 18 TNF-related genes expression into two signature clusters, high (n=94) and low (n=123). Patients in the high expression signature cluster were more likely to have high PD-L1 compared to the low expression cluster (corrected p-value<0.05). The high expression signature cluster was associated with better survival compared with the low expression cluster (HR=1.49, 95%CI: 1.05-2.13 (p=0.026)). When adjusting for TMB, MSI, PD-L1, TP53 and KRAS status, the TNF high expression signature cluster was still associated with better overall survival (HR=1.71, 95%CI:1.11-2.66 (p=0.016)). Other than the TNF high expression signature cluster group, only the presence of KRAS mutation was a predictor of a longer survival (HR=2.09, 95%CI:1.28-3.42 (p=0.003)). Conclusion: A high TNF-related gene expression profile may serve as an independent predictor for longer overall survival for patients treated with ICI. Citation Format: Obada Ehab Ababneh, Shumei Kato, Daisuke Nishizaki, Hirotaka Miyashita, Suzanna Lee, Mary Nesline, Sarabjot Pabla, Jeffrey Conroy, Paul DePietro, Heidi Ko, Razelle Kurzrock. The predictive role of TNF-related genes in patients receiving immune checkpoint inhibitors [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 6393.

Anomaly detection using deep learning based model with feature attention

<span lang="EN-US">Anomaly detection is a difficult problem with numerous industrial applications, such as analyzing the quality of objects using images. Anomaly detection is the process of identifying outliers in a given dataset. Recently, machine learning approaches to computer vision problems have outperformed classical state-of-the-art approaches. Anomaly detection problems can be solved using supervised approaches. However, labelled datasets are hard to obtain. Thus, many researchers have taken an unsupervised approach to solving the problem of anomaly detection. In this study, we use an adversarial auto encoder model as a base model and create a custom model to detect anomalies in images and videos. The model was trained exclusively on normal data. The modified national institute of standards and technology database (MNIST) dataset achieved an area under curve (AUC) score of 0.872 for anomaly detection, while the University of California San Diego (UCSD) anomaly dataset (Video dataset) achieved an AUC score of 0.74 for Ped1 and 0.87 for Ped2. To calculate the anomaly score, the concept of attention weights is combined with the reconstruction loss, and the proposed method outperformed other similar methods designed for the same problem. However, the usefulness of the proposed model was demonstrated through the detection of anomalies, and the model is still being improved for use in real-world situations.</span>

A new computer-aided diagnosis tool based on deep learning methods for automatic detection of retinal disorders from OCT images.

Early detection of retinal disorders using optical coherence tomography (OCT) images can prevent vision loss. Since manual screening can be time-consuming, tedious, and fallible, we present a reliable computer-aided diagnosis (CAD) software based on deep learning. Also, we made efforts to increase the interpretability of the deep learning methods, overcome their vague and black box nature, and also understand their behavior in the diagnosis. We propose a novel method to improve the interpretability of the used deep neural network by embedding the rich semantic information of abnormal areas based on the ophthalmologists' interpretations and medical descriptions in the OCT images. Finally, we trained the classification network on a small subset of the online publicly available University of California San Diego (UCSD) dataset with an overall of 29,800 OCT images. The experimental results on the 1000 test OCT images show that the proposed method achieves the overall precision, accuracy, sensitivity, and f1-score of 97.6%, 97.6%, 97.6%, and 97.59%, respectively. Also, the heat map images provide a clear region of interest which indicates that the interpretability of the proposed method is increased dramatically. The proposed software can help ophthalmologists in providing a second opinion to make a decision, and primitive automated diagnoses of retinal diseases and even it can be used as a screening tool, in eye clinics. Also, the improvement of the interpretability of the proposed method causes to increase in the model generalization, and therefore, it will work properly on a wide range of other OCT datasets.

Efficient Road Traffic Video Congestion Classification Based on the Multi-Head Self-Attention Vision Transformer Model

Abstract Due to rapid population growth, traffic congestion has become one of the major issues in urban areas. The utilization of technology may help to address this issue. This paper proposes a new Multi-head Self-attention Vision Transformer (MSViT) based macroscopic approach, for road traffic congestion classification. To evaluate this approach, we use the UCSD (University of California San Diego) dataset that includes different weather conditions (clear, overcast and rainy) and different traffic scenarios (light, medium and heavy). The classification accuracy reached a high level of 99.76% with this dataset and 99.37% when night-mode frames are added to it. The proposed MSViT based method outperforms the state-of-the-art macroscopic and microscopic methods that have been evaluated using the same UCSD dataset, which makes it an efficient solution for traffic congestion prediction.

OP40 Transcriptomic signature of response to vedolizumab in patients with moderate-to-severe ulcerative colitis: Results from an international, multicentre, retrospective cohort study

Abstract Background Transcriptomic analysis has the potential to facilitate drug development and clinical decision making in inflammatory bowel disease. Current understanding of pharmacodynamic response to vedolizumab is limited by a lack of publicly available mucosal biopsy RNA samples/sequencing data and corresponding clinical information. We aimed to identify gene signatures associated with response to vedolizumab in ulcerative colitis (UC) patients by pooling data from 3 international sites. Methods Patients receiving vedolizumab for the treatment of moderate-to-severe UC at Hospital Clínic de Barcelona (IDIBAPS), Mount Sinai Hospital (MSH), and University of California, San Diego (UCSD) with baseline and post-treatment (week 14 [± 4 weeks]) mucosal biopsy RNA-sequencing data and Mayo Clinic endoscopic subscores (MCES) were eligible for inclusion. Endoscopic response was defined as an MCES <1. Demographic and clinical data were collected. Differential expressed genes (DEGs) between baseline and week 14 were identified using linear mixed models with a >1.5 or <-1.5-fold change (adjusted P-value <.05). Enrichment was performed using Reactome pathway analysis and a previously published UC gene signature (adjusted P-value <.05).1 Analyses were performed using R version 4.3.0. Results Twenty-three patients were included (IDIBAPS: n=12; MSH: n=8; UCSD: n=3). Site-specific differences in sex, surgical history, and corticosteroid use were observed (Table). Analysing data by site yielded a relatively small number of DEGs between baseline and week 14 (IDIBAPS=57 [27 up-regulated, 30 down-regulated]; MSH=0; UCSD=0) among endoscopic responders (IDIBAPS: n=5; MSH: n=7; UCSD: n=2). Combining data from the 3 sites increased the number of DEGs among endoscopic responders at week 14 (n=14) to 1162 (662 up-regulated and 500 down-regulated). The down-regulated DEGs were enriched for genes involved in 37 of 103 up-regulated Reactome pathways in UC, including neutrophil degranulation, interleukin-4 and interleukin-13 signalling, interleukin-10 signalling, and integrin cell surface interactions. The up-regulated DEGs were enriched for genes involved in 1 of 39 down-regulated Reactome pathways in UC (Drug ADME). Conclusion We defined a pharmacodynamic signature using mucosal biopsies from UC patients who had endoscopic response to vedolizumab at week 14. The signature was enriched for several, but not all, pathways involved in UC. In addition to providing insight into vedolizumab’s mechanism of action, we underscored the utility of multi-site collaboration to access data, increase statistical power, and enhance the generalizability of research findings. Reference: 1. Linggi et al. Sci Rep 2021;11:18243.

Views of a non-probability sample of corresponding authors with retracted publications in biomedical fields about the impact of different types of retractions on researchers' careers.

Echoing Arturo Casadevall and Ferric Fang in their Reforming Science: Methodological and Cultural Reforms, "great human enterprises must undergo periodic cycles of self-examination and renewal to maintain their vigor". Especially in the last decade, the research culture has undergone such cycles, partially driven by countercultural transformations that have been reshaping assumptions towards reward-deserving achievements. Addressing retractions is among the challenges in this culture. This work builds upon research carried out at the University of California, San Diego (UCSD), which explored the views of 224 reviewers serving on panels for the US National Science Foundation, the National Institutes of Health, among others. We show results of a survey that add to our previous data. It was sent to a population of 1,089 corresponding authors affiliated with institutions from the 20 most productive countries in biomedical fields. We explored how corresponding authors of at least one retracted publication issued between 2013 and 2015 in biomedical journals envisioned the impact of different types of retractions on the careers of the first and corresponding authors. As such impact (if any) is not always immediate, we selected this time frame to ensure that potential respondents would have tangible post-retraction experience.

Automated classification of age-related macular degeneration from optical coherence tomography images using deep learning approach

<p>Early detection of macular diseases can prevent vision loss. Manual screening can be unreliable due to the similarity in the pathological presentations of common retinal illnesses like age-related macular degeneration (AMD). Researchers are becoming more interested in the accurate automated computer-based detection of macular diseases. Using healthy optical coherence tomography (OCT) images, the drusens (early stage) and choroidal neovascularization (CNV) (late stage) of AMD are thus classified using a completely different approach in this paper. The new deep learning (DL) model is proposed for multiple OCT image segmentation of ophthalmological diseases using attention-based nested U-Net (ANU-Net). The flower pollination optimization algorithm (FPOA) is used to optimize the hyperparameters of the network. The SqueezeNet-based classification can be made in the pre-processed images. A dataset from the University of California San Diego (UCSD) is used to evaluate the proposed method. 98.7% accuracy, 99.8% specificity, and 99.7% sensitivity are achieved by the proposed method. The proposed method produces better identification results for automated preliminary diagnosis of macular diseases in hospitals and eye clinics due to the positive classification results.</p>

People with HIV at the end-of-life and their next-of-kin/loved ones are willing to participate in interventional HIV cure-related research.

The Last Gift study at the University of California San Diego (UCSD), United States enrolls terminally ill people with HIV (PWH) in HIV cure research. From 2017 to 2022, we conducted surveys with Last Gift participants and their next-of-kin/loved ones to evaluate willingness to participate in different types of HIV cure research at the end of life (EOL). We analyzed willingness data descriptively. We surveyed 17 Last Gift participants and 17 next-of-kin/loved ones. More than half of Last Gift participants ( n = 10; 58.8%) expressed willingness to participate in studies involving totally new treatments or approaches ('first-in-human' studies), a combination of different approaches, the use of unique antibodies, proteins or molecules, or therapeutic vaccines. Under one-quarter of Last Gift participants ( n = 4; 23.5%) expressed willingness to participate in research involving interventions that may shorten their life expectancy to benefit medical research. Most Last Gift participants and their next-of-kin/loved ones also expressed high acceptance for various types of donations and biopsies at the EOL (e.g. hair donations and skin, lymph node or gut biopsies). Knowing whether people would be willing to participate in different types of EOL HIV cure research can help inform the design of future innovative studies. As a research community, we have a duty to design studies with adequate safeguards to preserve the public trust in research and honor PWH's important gift to humanity.

Anomaly Detection in Pedestrian Walkways for Intelligent Transportation System Using Federated Learning and Harris Hawks Optimizer on Remote Sensing Images

Anomaly detection in pedestrian walkways is a vital research area that uses remote sensing, which helps to optimize pedestrian traffic and enhance flow to improve pedestrian safety in intelligent transportation systems (ITS). Engineers and researchers can formulate more potential techniques and tools with the power of computer vision (CV) and machine learning (ML) for mitigating potential safety hazards and identifying anomalies (i.e., vehicles) in pedestrian walkways. The real-world challenges of scenes and dynamics of environmental complexity cannot be handled by the conventional offline learning-based vehicle detection method and shallow approach. With recent advances in deep learning (DL) and ML areas, authors have found that the image detection issue ought to be devised as a two-class classification problem. Therefore, this study presents an Anomaly Detection in Pedestrian Walkways for Intelligent Transportation Systems using Federated Learning and Harris Hawks Optimizer (ADPW-FLHHO) algorithm on remote sensing images. The presented ADPW-FLHHO technique focuses on the identification and classification of anomalies, i.e., vehicles in the pedestrian walkways. To accomplish this, the ADPW-FLHHO technique uses the HybridNet model for feature vector generation. In addition, the HHO approach is implemented for the optimal hyperparameter tuning process. For anomaly detection, the ADPW-FLHHO technique uses a multi deep belief network (MDBN) model. The experimental results illustrated the promising performance of the ADPW-FLHHO technique over existing models with a maximum AUC score of 99.36%, 99.19%, and 98.90% on the University of California San Diego (UCSD) Ped1, UCSD Ped2, and avenue datasets, respectively. Therefore, the proposed model can be employed for accurate and automated anomaly detection in the ITS environment.

Surveillance detection of anomalous activities with optimized deep learning technique in crowded scenes

The performance of conventional surveillance systems is challenged by high error detection rates in busy scenes, which has significantly affected the accurate detection of the current surveillance system. Feature representation and object pattern extraction from different scenes have made deep learning (DL) promising methods in surveillance systems, compared to the approaches where features are created manually. To improve the detection accuracy, this paper presents an intelligent DL technique that combines convolutional neural network (CNN) and long short-term memory (LSTM). CNN extracts and learns the object features from a set of raw images, while the LSTM is then used by gated mechanisms to store important information from the extracted features. The proposed method was validated using datasets from the University of California San Diego (UCSD). The result shows that the model achieves 95% accuracy, which is superior compared to other conventional detection models.

The UC san diego health suspicion of cancer clinic.

e18528 Background: Physicians at UC San Diego Moores Cancer Center (MCC), like at many major cancer centers in the United States, are subspecialized by tumor type. Patients with symptoms concerning for cancer, but without a diagnosis, often face challenges in establishing care with an oncologist, which can potentially lead to delays in diagnosis and treatment. Methods: At our institution, we created a Suspicion of Cancer (SOC) clinic for referral of patients who have symptoms concerning for cancer, but without a definitive diagnosis of malignancy. The purpose of the clinic is to facilitate establishment of a diagnosis and access to an oncologist in a timely manner. We created an order for referral in our electronic medical record system that enables any provider to refer to patients to the SOC clinic. Once a referral is placed, a hematologist-oncologist or medical oncologist reviews the case and approves the patient for an appointment with the SOC clinic. Results: Between January 2022 and November 2022, we received 107 unique referrals to the SOC clinic. The average age of referred patients was 64 years, with 50 (46.7%) females and 57 (53.3%) males. Thirty-four (31.2%) referrals were from primary care doctors, 23 (21.5%) from hospital medicine, 22 (20.6%) from emergency medicine, 15 (14.0%) from medical subspecialties or neurology, 11 (10.3%) from surgery or surgical subspecialties, and 1 (0.93%) that was self-referred. Thirty (28.0%) patients already had definitive diagnosis of malignancy from either our institution or an outside institution, while 77 (72.0%) did not have a diagnosis at time of referral. Of the 107 referrals, 21 (19.6%) did not follow up with a medical oncologist or hematologist-oncologist within our health system. Reasons for not following up included insurance status, establishing care elsewhere, failed attempts to contact patients, or no shows to established clinic appointments. One patient was already established with a hematologist-oncologist within the MCC prior to referral. Eighty-six (80.4%) patients were seen in new consultation by a medical oncologist or hematologist-oncologist at our institution. After referral, the average time to consultation with a medical oncologist or a hematologist-oncologist was 15.3 days. In the 77 cases without diagnosis at time of referral, 39 (50.6%) underwent biopsy, while 38 (49.4%) were deemed to not require biopsy or declined biopsy. Twenty-five (32.5%) patients were diagnosed with cancer and 52 (67.5%) either did not have biopsy or had a negative biopsy. Ninety-two (86.0%) of the 107 referred patients were alive at time of analysis, while 15 (14.0%) were deceased. Conclusions: The MCC SOC clinic offers a referral pathway to expedite diagnosis and access to care for new oncology patients. In continuing to collect data related to the above outcomes, we hope to identify barriers to the delivery of oncologic care and improve patient outcomes.

Digital Histological Measurements Show a Range of Values Within Traditional Pathological Metrics of AD-related Tau Pathology (P3-6.002)

<h3>Objective:</h3> To examine spectrum of neocortical tau burden using digital histological methods in Alzheimer’s disease (AD) and Lewy Body Disease (dementia with Lewy bodies and Parkinson’s disease) with AD co-pathology (LBD+AD) with high Braak tau stages. <h3>Background:</h3> Neuropathological assessment remains gold-standard for diagnosing and staging neurodegenerative diseases. Digital histological assessment may capture a spectrum of pathological burden not easily demonstrated by traditional methods. <h3>Design/Methods:</h3> Cases from UC-San Diego (UCSD) and University of Pennsylvania (UPenn) with AD (UCSD:15, UPenn:33) and LBD+AD (UCSD:7, UPenn:10) with Braak tau stages V and VI (B3) had middle frontal cortex (MFC), superior temporal cortex (STC), and angular gyrus (ANG) immunohistochemically stained for pathological tau inclusions (AT8) and whole-slide images analyzed for tau %area occupied (tau%AO) using QuPath image analysis software. Natural-log tau%AO values were compared between ordinal severity scores (0–3), Braak stages, and between AD and LBD+AD in each region using ANOVA and t-tests. <h3>Results:</h3> In both cohorts, here were significant increases in tau%AO across blinded regional ordinal severity scores (UCSD F=28.4, p&lt;0.001, UPenn F=236, p&lt;0.001). All regions had higher tau%AO in Braak VI cases than Braak V in the UCSD cohort (p&lt;0.02). The STC harbored the highest tau%AO values in AD and LBD+AD. AD had higher tau%AO than LBD+AD in every region (MFC, STC, ANG, neocortical average: t=2.2–4.3, p&lt;0.03). <h3>Conclusions:</h3> Digital histological measurements recapitulate aspects of known patterns of AD-related tau pathology but reveal a range of pathological burden within current grading and staging systems. These methods may have relevance for ongoing biomarker development and modeling disease using human histology. <b>Disclosure:</b> The institution of Dr. Coughlin has received research support from American Academy of Neurology. The institution of Dr. Coughlin has received research support from NIA. The institution of Dr. Coughlin has received research support from NINDS. Mr. Kim has nothing to disclose. Ms. Arezoumandan has nothing to disclose. Ms. Peterson has nothing to disclose. Anne Hiniker has nothing to disclose. Larry Hansen has nothing to disclose. Dr. Rissman has nothing to disclose. Donald Pizzo has nothing to disclose. Edward Lee has nothing to disclose. Dr. Irwin has received personal compensation in the range of $500-$4,999 for serving on a Scientific Advisory or Data Safety Monitoring board for Denali Therapeutics. The institution of Dr. Irwin has received research support from NIH.

Hepatitis C Virus Micro-elimination Among People With HIV in San Diego: Are We on Track?

Rising incidence of hepatitis C virus (HCV) among people with HIV (PWH) in San Diego County (SDC) was reported. In 2018, the University of California San Diego (UCSD) launched a micro-elimination initiative among PWH, and in 2020 SDC launched an initiative to reduce HCV incidence by 80% across 2015-2030. We model the impact of observed treatment scale-up on HCV micro-elimination among PWH in SDC. A model of HCV transmission among people who inject drugs (PWID) and men who have sex with men (MSM) was calibrated to SDC. The model was additionally stratified by age, gender, and HIV status. The model was calibrated to HCV viremia prevalence among PWH in 2010, 2018, and 2021 (42.1%, 18.5%, and 8.5%, respectively), and HCV seroprevalence among PWID aged 18-39 years, MSM, and MSM with HIV in 2015. We simulate treatment among PWH, weighted by UCSD Owen Clinic (reaching 26% of HCV-infected PWH) and non-UCSD treatment, calibrated to achieve the observed HCV viremia prevalence. We simulated HCV incidence with observed and further treatment scale-up (+/- risk reductions) among PWH. Observed treatment scale-up from 2018 to 2021 will reduce HCV incidence among PWH in SDC from a mean of 429 infections/year in 2015 to 159 infections/year in 2030. County-wide scale-up to the maximum treatment rate achieved at UCSD Owen Clinic (in 2021) will reduce incidence by 69%, missing the 80% incidence reduction target by 2030 unless accompanied by behavioral risk reductions. As SDC progresses toward HCV micro-elimination among PWH, a comprehensive treatment and risk reduction approach is necessary to reach 2030 targets.

An Explainable Fully Dense Fusion Neural Network with Deep Support Vector Machine for Retinal Disease Determination

Retinal issues are crucial because they result in visual loss. Early diagnosis can aid physicians in initiating treatment and preventing visual loss. Optical coherence tomography (OCT), which portrays retinal morphology cross-sectionally and noninvasively, is used to identify retinal abnormalities. The process of analyzing OCT images, on the other hand, takes time. This study has proposed a hybrid approach based on a fully dense fusion neural network (FD-CNN) and dual preprocessing to identify retinal diseases, such as choroidal neovascularization, diabetic macular edema, drusen from OCT images. A dual preprocessing methodology, in other words, a hybrid speckle reduction filter was initially used to diminish speckle noise present in OCT images. Secondly, the FD-CNN architecture was trained, and the features obtained from this architecture were extracted. Then Deep Support Vector Machine (D-SVM) and Deep K-Nearest Neighbor (D-KNN) classifiers were proposed to reclassify those features and tested on University of California San Diego (UCSD) and Duke OCT datasets. D-SVM demonstrated the best performance in both datasets. D-SVM achieved 99.60% accuracy, 99.60% sensitivity, 99.87% specificity, 99.60% precision and 99.60% F1 score in the UCSD dataset. It achieved 97.50% accuracy, 97.64% sensitivity, 98.91% specificity, 96.61% precision, and 97.03% F1 score in Duke dataset. Additionally, the results were compared to state-of-the-art works on the both datasets. The D-SVM was demonstrated to be an efficient and productive strategy for improving the robustness of automatic retinal disease classification. Also, in this study, it is shown that the unboxing of how AI systems' black-box choices is made by generating heat maps using the local interpretable model-agnostic explanation method, which is an explainable artificial intelligence (XAI) technique. Heat maps, in particular, may contribute to the development of more stable deep learning-based systems, as well as enhancing the confidence in the diagnosis of retinal disease in the analysis of OCT image for ophthalmologists.

Experts Reflect on What New Law Might Mean for Cannabis Research

Experts Reflect on What New Law Might Mean for Cannabis Research

Large-scale shake table testing of shallow tunnel-ground system

Soil-structure interaction (SSI) is of critical importance in evaluating the seismic response of shallow tunnels. Experimental observations are crucial in assessing the influence of the surrounding and overburden soil near the ground surface. Therefore, this paper presents a series of shake table tests to investigate the seismic response of a scaled shallow tunnel with different backfill material properties and thickness of overburden soil (burial depth). A shallow tunnel was tested incorporating ground representative of realistic backfill conditions using the large-scale 1-g outdoor shake table at the University of California, San Diego (UCSD). The tunnel structure was designed at 1/9 scale based on the configuration of an actual recently constructed prototype structure in San Francisco, California. The test results were interpreted in full-scale, which demonstrated that the maximum tunnel deformation generally occurred at the instant of peak ground acceleration (PGA). Relatively large shear strains were developed (up to 6%) and significant reduction of the soil shear modulus were evaluated as the PGA increased during the shaking. The backfill shear strains were lower as the tunnel embedment depth increased because of the increased soil stiffness and strength due to the higher overburden stress. In addition, the tunnel racking deformation became closer to that of the ground as the tunnel embedment increased. At the largest observed soil shear strains, tunnel racking was significantly lower than that of ground deformation due to the decrease in soil. The FHWA simplified solution was used to estimate tunnel racking deformation and wall bending moments. The FHWA solution produced reasonable estimates of ground deformation. However, a discrepancy was noted in estimating the tunnel deformation in the given ground deformation known as racking ratio (Rr). This discrepancy contributed to conservatism in the FHWA racking prediction that was pronounced particularly with the decreased overburden soil stress.

Plasma enzymatic activity, proteomics and peptidomics in COVID-19-induced sepsis: A novel approach for the analysis of hemostasis.

Introduction: Infection by SARS-CoV-2 and subsequent COVID-19 can cause viral sepsis. We investigated plasma protease activity patterns in COVID-19-induced sepsis with bacterial superinfection, as well as plasma proteomics and peptidomics in order to assess the possible implications of enhanced proteolysis on major protein systems (e.g., coagulation). Methods: Patients (=4) admitted to the intensive care units (ICUs) at the University of California, San Diego (UCSD) Medical Center with confirmed positive test for COVID-19 by real-time reverse transcription polymerase chain reaction (RT-PCR) were enrolled in a study approved by the UCSD Institutional Review Board (IRB# 190699, Protocol #20-0006). Informed consent was obtained for the collection of blood samples and de-identified use of the data. Blood samples were collected at multiple time points and analyzed to quantify a) the circulating proteome and peptidome by mass spectrometry; b) the aminopeptidase activity in plasma; and c) the endopeptidase activity in plasma using fluorogenic substrates that are cleaved by trypsin-like endopeptidases, specific clotting factors and plasmin. The one patient who died was diagnosed with bacterial superinfection on day 7 after beginning of the study. Results: Spikes in protease activity (factor VII, trypsin-like activity), and corresponding increases in the intensity of peptides derived by hydrolysis of plasma proteins, especially of fibrinogen degradation products and downregulation of endogenous protease inhibitors were detected on day 7 for the patient who died. The activity of the analyzed proteases was stable in survivors. Discussion: The combination of multiomics and enzymatic activity quantification enabled to i) hypothesize that elevated proteolysis occurs in COVID-19-induced septic shock with bacterial superinfection, and ii) provide additional insight into malfunctioning protease-mediated systems, such as hemostasis.

UC San Diego Undergraduates and the Ocean Discovery Institute Collaborate to Form a Pilot Program in Culturally Responsive Mentoring

The geoscience workforce, particularly in ocean sciences, lacks diversity. Implementing culturally responsive teaching and mentoring practices throughout the K–university pipeline is one way to help increase diversity in the field. Here, we describe initial results and lessons learned from a pilot course conducted through a partnership between the University of California San Diego (UCSD) and the Ocean Discovery Institute (ODI), a local nonprofit that provides opportunities for low-income K–12 youths to learn about oceanography. The goals of the pilot course are to educate a racially diverse group of UCSD undergraduates in culturally sensitive mentoring practices and provide opportunities for them to apply their newly acquired skills while working with K–12 students at ODI. The undergraduates prepared and implemented lessons designed to demystify the college experience for their mentees. Through classroom experience and weekly journal reflections, undergraduates developed confidence and improved their ability to mentor youth.