University of Southern California

Objectives Identify existing research on impacts of transitions between electronic health record (EHR) systems on patients' healthcare experiences. Methods Scoping review. We searched MedLine, OVID, Embase, CINAHL, and PsycInfo databases for articles on patient experiences with EHR-to-EHR transitions. Results Three studies met inclusion criteria. All three used validated surveys to compare patient satisfaction with care pre- and post-transition. The surveys did not include specific questions about the EHR transition; one study focused on patient perceptions of provider computer use. Satisfaction levels initially decreased following EHR implementation, then returned to baseline between six and 15 months later in two of three studies. Factors associated with changes in observed satisfaction are unknown. Conclusions Patient experience has been given limited attention in studies of EHR-to-EHR transitions. Future research should look beyond satisfaction, and examine how an EHR-to-EHR transition can impact the quality of patients' care, including safety, effectiveness, timeliness, efficiency, and equity. Innovation To our knowledge, this is the first literature review on EHR transitions that specifically focused on patient experiences. In preparation for a transition from one EHR to another, healthcare system leaders should consider the multiple ways patients' experiences with care may be impacted and develop strategies to minimize disruptions in care.

Human activity recognition (HAR) with wearables is one of the serviceable technologies in ubiquitous and mobile computing applications. The sliding-window scheme is widely adopted while suffering from the multi-class windows problem. As a result, there is a growing focus on joint segmentation and recognition with deep-learning methods, aiming at simultaneously dealing with HAR and time-series segmentation issues. However, obtaining the full activity annotations of wearable data sequences is resource-intensive or time-consuming, while unsupervised methods yield poor performance. To address these challenges, we propose a novel method for joint activity segmentation and recognition with timestamp supervision, in which only a single annotated sample is needed in each activity segment. However, the limited information of sparse annotations exacerbates the gap between recognition and segmentation tasks, leading to sub-optimal model performance. Therefore, the prototypes are estimated by class-activation maps to form a sample-to-prototype contrast module for well-structured embeddings. Moreover, with the optimal transport theory, our approach generates the sample-level pseudo-labels that take advantage of unlabeled data between timestamp annotations for further performance improvement. Comprehensive experiments on four public HAR datasets demonstrate that our model trained with timestamp supervision is superior to the state-of-the-art weakly-supervised methods and achieves comparable performance to the fully-supervised approaches.

In Federated Learning (FL), clients may have weak devices that cannot train the full model or even hold it in their memory space. To implement large-scale FL applications, thus, it is crucial to develop a distributed learning method that enables the participation of such weak clients. We propose EmbracingFL , a general FL framework that allows all available clients to join the distributed training regardless of their system resource capacity. The framework is built upon a novel form of partial model training method in which each client trains as many consecutive output-side layers as its system resources allow. Our study demonstrates that EmbracingFL encourages each layer to have similar data representations across clients, improving FL efficiency. The proposed partial model training method guarantees convergence to a neighbor of stationary points for non-convex and smooth problems. We evaluate the efficacy of EmbracingFL under a variety of settings with a mixed number of strong, moderate ( $\sim 40\%$ memory), and weak ( $\sim 15\%$ memory) clients, datasets (CIFAR-10, FEMNIST, and IMDB), and models (ResNet20, CNN, and LSTM). Our empirical study shows that EmbracingFL consistently achieves high accuracy as like all clients are strong, outperforming the state-of-the-art width reduction methods (i.e. HeteroFL and FjORD).

Single flux quantum (SFQ) technology has garnered significant attention due to its low switching power and high operational speed. Researchers have been actively pursuing more advanced devices and technologies to further reduce the reliance on inductors, bias, and dynamic power. Recently, innovative magnetic Josephson junction devices have emerged, enhancing the field of superconductor electronics (SCE) logic. This paper introduces a novel cell library design that relies entirely on Josephson junctions (JJs), showing promising potential for eliminating the need for inductors in conventional SFQ cells. This results in a 55% reduction in cell size and an 80% decrease in both static and dynamic power consumption. The proposed library implements a half flux quantum (HFQ) logic, where each pulse duration is half that of a single flux quantum pulse. The paper presents the schematics of the basic cells, emphasizing critical circuit parameters and their margins. Additionally, it examines layout blueprints, showcasing the advantageous areasaving characteristics of the proposed design.

Temporal information plays a pivotal role in Bird’s-Eye-View (BEV) driving scene understanding, which can alleviate the visual information sparsity. However, the indiscriminate temporal fusion method will cause the barrier of feature redundancy when constructing vectorized High-Definition (HD) maps. In this paper, we revisit the temporal fusion of vectorized HD maps, focusing on temporal instance consistency and temporal map consistency learning. To improve the representation of instances in single-frame maps, we introduce a novel method, DTCLMapper. This approach uses a dual-stream temporal consistency learning module that combines instance embedding with geometry maps. In the instance embedding component, our approach integrates temporal Instance Consistency Learning (ICL), ensuring consistency from vector points and instance features aggregated from points. A vectorized points pre-selection module is employed to enhance the regression efficiency of vector points from each instance. Then aggregated instance features obtained from the vectorized points preselection module are grounded in contrastive learning to realize temporal consistency, where positive and negative samples are selected based on position and semantic information. The geometry mapping component introduces Map Consistency Learning (MCL) designed with self-supervised learning. The MCL enhances the generalization capability of our consistent learning approach by concentrating on the global location and distribution constraints of the instances. Extensive experiments on well-recognized benchmarks indicate that the proposed DTCLMapper achieves state-of-the-art performance in vectorized mapping tasks, reaching $61.9\%$ and $65.1\%$ mAP scores on the nuScenes and Argoverse datasets, respectively. The source code is available at https://github.com/lynn-yu/DTCLMapper.

The emergence of diverse machine learning (ML) models has led to groundbreaking revolutions in computer vision (CV). These ML models include convolutional neural networks (CNNs), graph neural networks (GNNs), and vision transformers (ViTs). However, existing hardware accelerators designed for CV lack the versatility to support various ML models, potentially limiting their applicability to real-world scenarios. To address this limitation, we introduce VisionAGILE, a domain-specific accelerator designed to be versatile and capable of accommodating a range of ML models, including CNNs, GNNs, and ViTs. VisionAGILE comprises a compiler, a runtime system, and a hardware accelerator. For the hardware accelerator, we develop a novel unified architecture with a flexible data path and memory organization to support the computation primitives in various ML models. Regarding the compiler design, we develop a unified compilation workflow that maps various ML models to the proposed hardware accelerator. The runtime system executes dynamic sparsity exploitation to reduce inference latency and dynamic task scheduling for workload balance. The compiler, the runtime system, and the hardware accelerator work synergistically to support a variety of ML models in CV, enabling low-latency inference. We deploy the hardware accelerator on a state-of-the-art data center FPGA (Xilinx Alveo U250). We evaluate VisionAGILE on diverse ML models for CV, including CNNs, GNNs, hybrid models (comprising both CNN and GNN), and ViTs. The experimental results indicate that, compared with state-of-the-art CPU (GPU) implementations, VisionAGILE achieves a speedup of 81.7× (4.8×) in terms of latency. Evaluated on standalone CNNs, GNNs, and ViTs, VisionAGILE demonstrates comparable or higher performance with state-of-the-art CNN accelerators, GNN accelerators, and ViT accelerators, respectively

Introduction Robotic Radical Prostatectomy using the Da-Vinci Single-Port (SP) robot can provide comparable functional and oncological outcomes with potential advantages pertaining to peri-operative morbidity, especially in patients with an extensive history of prior abdominal surgeries (¹1 Ferguson EL, Ramos-Carpinteyro R, Soputro N, Chavali JS, Geskin A, Kaouk JH. Single-Port Robotic Radical Prostatectomy Using Transvesical and Transperineal Access in Patients with a Hostile Abdomen. J Endourol. 2024;38:150-8. doi: 10.1089/end.2023.0128. https://doi.org/10.1089/end.2023.0128... , ²2 Pettenuzzo G, Ditonno F, Cannoletta D, Morgantini L, Sauer RC, Torres-Anguiano JR, et al. Single Port Radical Prostatectomy as a Viable Option for Highly Complex Patients: A Single Center Experience. Urology. 2024;189:55-63. doi: 10.1016/j.urology.2024.04.051. https://doi.org/10.1016/j.urology.2024.0... ). Materials and Methods Our case is a 74-year-old male with a history of diabetes, cardiac bypass, hypertension, and hyperlipidemia, presenting with a PSA of 7.2. His MRI showed a PIRADS-5 lesion in the left apex and mid-gland peripheral zone, and he was diagnosed with unfavorable intermediate-risk prostate cancer after MRI guided fusion biopsy. His BMI was 31, and past surgical history was pertinent for two exploratory laparotomies due to gunshot wounds and a colostomy creation followed by reversal. The standardized steps of robotic radical prostatectomy were carried out using SP robotic platform performed by author SH (³3 Menon M, Tewari A, Peabody JO, Shrivastava A, Kaul S, Bhandari A, et al. Vattikuti Institute prostatectomy, a technique of robotic radical prostatectomy for management of localized carcinoma of the prostate: experience of over 1100 cases. Urol Clin North Am. 2004;31:701-17. doi: 10.1016/j.ucl.2004.06.011. https://doi.org/10.1016/j.ucl.2004.06.01... , ⁴4 Soputro NA, Kaouk J. Single-port robot-assisted radical prostatectomy. World J Urol. 2024;42:245. doi: 10.1007/s00345-024-04914-5. https://doi.org/10.1007/s00345-024-04914... ). Results Total operative time and estimated blood loss were 210 minutes and 150mL respectively. The patient was discharged on postoperative day one and final pathology showed adenocarcinoma of the prostate Gleason score 4+3=7, pT2NxR0 and negative surgical margins. The patient was continent four weeks after surgery and the PSA continues to be undetectable after three months. Conclusion Transvesical Radical prostatectomy using the single port platform provides acceptable oncological and functional outcomes and quicker recovery given decreased risk of ileus and peritoneal irritation. Given that the abdominal cavity is not violated, the risk of bowel or vascular injury is mitigated, especially in patients with a hostile abdomen.

Diffusive memristors, characterized by their nonlinear dynamic conductance changes in response to stimulations, enable effective and efficient hardware implementation of various neuromorphic principles and computing algorithms. However, the existing diffusive memristor models either involve overly complicated evolutionary equations that are time-consuming to execute during simulation or miss some important features of the diffusive memristor dynamics that are crucial for many applications. This hindered the progress of algorithm and circuit co-design based on diffusive memristors. Here, we construct a faithful and compact diffusive memristor model capturing the dynamics of the filament length, residue length, channel cross section and stochasticity within the physical devices. Using this model, we successfully replicated the experimentally observed quasistatic behaviors and dynamic behaviors of diffusive memristors. Moreover, our model accurately emulates the high-order dynamic behaviors of leaky integrate-and-fire (LIF) neurons reported in previous literature. These results underscore the significance of our model in enabling precise and efficient algorithm and circuit co-design based on diffusive memristors, paving the way towards the large-scale intelligent systems involving diffusive memristors.

An interacting spin system is an excellent testbed for fundamental quantum physics and applications in quantum sensing and quantum simulation. For these investigations, detailed information on the interactions, e.g., the number of spins and their interaction strengths, is often required. In this study, we present the identification and characterization of a single nitrogen vacancy (NV) center coupled to two electron spins. In the experiment, we first identify a well-isolated single NV center and characterize its spin decoherence time. Then, we perform NV-detected electron paramagnetic resonance (EPR) spectroscopy to detect surrounding electron spins. From the analysis of the NV-EPR signal, we precisely determine the number of detected spins and their interaction strengths. Moreover, the spectral analysis indicates that the candidates of the detected spins are diamond surface spins. This study demonstrates a promising approach for the identification and characterization of an interacting spin system for realizing entangled sensing using electron spin as quantum reporters.

A non-uniform (NU) sub-sampling receiver (RX) with a NU discrete-time FIR (NU DT FIR) filter can create multiple tunable frequency notches both near and far from the passband via predesigned NU sampling clocks and filter coefficients. NU DT finite impulse response (FIR) acts as an anti-aliasing (AA) filter for a nonuniformly sampled signal which relaxes the subsequent ADC speed and dynamic range (DR). To save power and area and to improve linearity, the FIR filter is implemented in the current domain and shares the capacitive DAC with a subsequent asynchronous SAR ADC. A proof-of-concept NU DT FIR RX is implemented in 28 nm CMOS. It achieves up to 42 dB of blocker rejection with B $_{\text{1dB}}$ of 4 dBm. The receiver measures $-$ 27.4 dB EVM for a 64-QAM 100 MSymbol/s signal centered at 20 GHz in the presence of a 10-dBc blocker. The end-to-end RX consumes 24 mW from a 1-V supply and occupies an active area of 0.072 mm $^{\rm{2}}$ .

Despite substantial reductions in the cost of sequencing over the last decade, genetic panels remain relevant due to their cost‐effectiveness and flexibility across a variety of sample types. In particular, single nucleotide polymorphism (SNP) panels are increasingly favoured for conservation applications. SNP panels are often used because of their adaptability, effectiveness with low‐quality samples, and cost‐efficiency for population monitoring and forensics. However, the selection of diagnostic SNPs for population assignment and individual identification can be challenging. The consequences of poor SNP selection are under‐powered panels, inaccurate results, and monetary loss. Here, we develop a novel and user‐friendly SNP selection pipeline (mPCRselect) that can be used to select SNPs for population assignment and/or individual identification. mPCRselect allows any researcher, who has sufficient SNP‐level data, to design a successful and cost‐effective SNP panel for a diploid species of conservation concern.

Background Over 90% of individuals with mild cognitive impairment (MCI) may not receive a timely diagnosis. Understanding community-based practice patterns, where most individuals are seen, is critical to improving patient care. Objective To understand how patients with MCI and mild dementia due to Alzheimer's disease (AD) are diagnosed and managed in community-based settings, including the use of clinical and cognitive assessments, referrals to dementia-related specialties, and receipt of treatment. Methods This observational study recruited community-based primary care physicians (PCPs) (N = 177) and neurologists (N = 147) in August-September 2023, through a verified physician panel with broad geographic representation across the US. Physicians abstracted medical chart data from patients diagnosed with MCI or mild AD within the previous two years. Data collected included use of neurocognitive assessments, biomarker and structural imagine tests, referrals, and treatments. Descriptive statistics were used. Results Medical records for 817 MCI and 467 mild AD patients were abstracted. The mean age was 70.2 years, 56.4% were female, and 67.2% were White. Symptoms were commonly reported by a family member (67.2%). Nearly 1 in 6 patients did not receive any neurocognitive assessments (16.1%), and nearly 1 in 4 did not receive a structural imaging or AD-specific biomarker test (23.7%). AD-specific biomarker tests were more common among patients aged ≥65 (87.1% versus 75.3%; p < 0.05). Less than 1 in 4 patients were referred for cognitive/behavioral concerns. Conclusions As the diagnostic and treatment landscape changes, education on symptom recognition, and physician training on new technologies may facilitate timely diagnoses and improve patient outcomes.

In this chapter, we review the current state of the art for uveal melanoma. Diagnosis, prognosis, and disease management are reviewed with special attention to the latest developments in molecular genetics. We also review the many eye conserving options for treatment of this disease with a special focus on advanced episcleral plaque brachytherapy. Modern therapeutic advances have now reached a point that nearly all eyes can be managed in a conservative manner with significant visual preservation for most patients.

Conventional contrast-enhanced MRI is very limited in its ability to differentiate tumor from post-treatment radiation effect (PTRE) in brain tumor patients following radiation therapy. T2*-weighted dynamic susceptibility contrast (DSC) perfusion MRI is a well-established method of advanced functional imaging that can provide valuable additional information in the form of relative cerebral blood volume (rCBV) to differentiate these entities. It is common to think of new or enlarging contrast-enhancing lesions in the post radiotherapy setting as either all PTRE or all tumor, when they are often actually composed of an admixture of them. A more recent application of DSC-MRI called fractional tumor burden (FTB) is able to provide quantitative information of the amount of PTRE vs. tumor in any such lesion. While there is some evidence that rCBV may predict outcome in brain tumor patients, larger, well-designed clinical trials are needed to validate rCBV as a biomarker to predict survival. In addition, more standardization of technique as well as knowledge of the precision of DSC-MRI is needed for more widespread adoption of this useful imaging method beyond academic centers.

Epigenetic clocks provide powerful tools for estimating health and lifespan but their ability to predict brain degeneration and neuronal damage during the aging process is unknown. In this study, we use GrimAge, an epigenetic clock correlated to several blood plasma proteins, to longitudinally investigate brain cellular microstructure in axonal white matter from a cohort of healthy aging individuals. A specific focus was made on white matter hyperintensities, a visible neurological manifestation of small vessel disease, and the axonal pathways throughout each individual's brain affected by their unique white matter hyperintensity location and volume. 98 subjects over 55 years of age were scanned at baseline with 41 returning for a follow‐up scan 2 years later. Using diffusion MRI lesionometry, we reconstructed subject‐specific networks of affected axonal tracts and examined the diffusion cellular microstructure composition of these areas, both at baseline and longitudinally, for evidence of cellular degeneration. A chronological age‐adjusted version of GrimAge was significantly correlated with baseline WMH volume and markers of neuronal decline, indicated by increased extracellular free water, increased intracellular signal, and decreased axonal signal within WMH. By isolating subject‐specific axonal regions “lesioned” by crossing through a WMH, age‐adjusted GrimAge was also able to predict longitudinal development of similar patterns of neuronal decline throughout the brain. This study is the first to demonstrate WMH lesionometry as a subject‐specific precision imaging technique to study degeneration in aging and the first to establish a relationship between accelerated epigenetic GrimAge and brain cellular microstructure in humans.

In this paper we present new empirical findings on the determinants of well-being of 54–75 year old individuals in China, Japan, and Korea. Using the harmonized Health and Retirement Surveys (HRS) that are designed to be similar to the Rand HRS, namely CHARLS in China, JSTAR in Japan and KLoSA in Korea, we run country- and gender-specific panel regressions using all available waves measures of subjective well-being (SWB) are associated with various economic, social, and demographic characteristics. Consistent with previous findings that highlight the role of education, health, employment status and social interactions in single wave studies, we find that these common factors continue to be important across countries and over time. In addition, we find that older Korean individuals in more recent waves have lower SWB than those in the first wave (2006). Finally, we find that there remain important differences across countries about the role of factors such as housing wealth, relative income and sources of life satisfaction, suggesting further country-specific research into the determinants of well-being.

Purpose To develop a robust single breath‐hold approach for volumetric lung imaging at 0.55T. Method A balanced‐SSFP (bSSFP) pulse sequence with 3D stack‐of‐spiral (SoS) out‐in trajectory for volumetric lung imaging at 0.55T was implemented. With 2.7× undersampling, the pulse sequence enables imaging during a 17‐s breath‐hold. Image reconstruction is performed using 3D SPIRiT and 3D l1‐Wavelet regularizations. In two healthy volunteers, single breath‐hold SoS out‐in bSSFP was compared against stack‐of‐spiral UTE (spiral UTE) and half‐radial dual‐echo bSSFP (bSTAR), based on signal intensity (SI), blood‐lung parenchyma contrast, and image quality. In six patients with pathologies including lung nodules, fibrosis, emphysema, and air trapping, single breath‐hold SoS out‐in and bSTAR were compared against low‐dose computed tomography (LDCT). Results SoS out‐in bSSFP achieved 2‐mm isotropic resolution lung imaging with a single breath‐hold duration of 17 s. SoS out‐in (2‐mm isotropic) provided higher lung parenchyma and blood SI and blood‐lung parenchyma contrast compared to spiral UTE (2.4 × 2.4 × 2.5 mm³) and bSTAR (1.6‐mm isotropic). When comparing SI normalized by voxel size, SoS out‐in has lower lung parenchyma signal, higher blood signal, and a higher blood‐lung parenchyma contrast compared to bSTAR. In patients, SoS out‐in bSSFP was able to identify lung fibrosis and lung nodules of size 4 and 8 mm, and breath‐hold bSTAR was able to identify lung fibrosis and 8 mm nodules. Conclusion Single breath‐hold volumetric lung imaging at 0.55T with 2‐mm isotropic spatial resolution is feasible using SoS out‐in bSSFP. This approach could be useful for rapid lung disease screening, and in cases where free‐breathing respiratory navigated approaches fail.

Solving the problem of object detection in complex and unstructured environments is crucial for enhancing the safety and efficiency of autonomous systems. This paper introduces a semantic segmentation model capable of accurate object detection in complex backgrounds by integrating multiple Convolutional Neural Networks (CNNs). The system incorporates two distinct segmentation models: an Encoder-Decoder architecture for acquiring abstract feature representations and a dilated convolutional branch to tackle variations in object sizes. The model employs a dynamic fusion mechanism based on confidence scores from each branch, allowing it to adapt to varying and dynamic situations. The model is evaluated on the Indian Driving Dataset (IDD), featuring unstructured road environments, and the Cityscape dataset. Comparative pixel-wise analysis shows the proposed model outperforming four other state-of-the-art segmentation models by $12.91\%$ on the IDD and by $19.7\%$ over the second-best model on the Cityscape dataset in terms of F1 score. Furthermore, an extensive ablation study validates the efficacy of the ensemble approach and underscores the effectiveness of categorical cross-entropy as the chosen loss function.

Background The Comprehensive Care for Joint Replacement (CJR) model is an alternative Medicare payment model for joint replacement that mandated participation by hospitals in randomly selected Metropolitan Statistical Areas (MSAs). On average, the program decreased inpatient length of stay and increased home discharge rates. It is unclear if these effects differed based on hospital ownership type, even though ownership may impact care redesign opportunities. Methods We used the 2014–2017 California Patient Discharge Datasets. The study included 113,590 hospitalizations for hip and knee joint replacement from 287 hospitals in the treated and control MSAs in California. The primary outcomes were inpatient length of stay and home discharge rates. Home discharge status included self-care, the use of home health, and hospice care at home. To determine whether the impact of the CJR model differed by hospital ownership type, we used event study, difference-in-differences (DID), and triple differences (DDD) models to estimate changes in health care services utilization in treated relative to control areas before versus after CJR implementation (April 2016) by hospital ownership type. Results Of the 113,590 hospitalizations, 51,708 (45.52%) were in treated MSAs and 61,882 (54.48%) were in control MSAs; 81,649 (71.88%) were from nonprofit hospitals, 20,247 (17.82%) were from for-profit hospitals, and 11,694 (10.29%) were from government-owned hospitals. DID analyses showed that after policy implementation, nonprofit and for-profit hospitals experienced a decrease in inpatient length of stay of 0.02 days (95% CI, -0.04 to -0.01) and 0.04 days (95% CI, -0.06 to -0.01), respectively, while government-owned hospitals experienced an increase by 0.11 days (95% CI, 0.04 to 0.18). For home discharge rates, nonprofit hospitals experienced an increase of 0.02 (95% CI, 0.01 to 0.03), while other hospitals did not show statistically significant changes. DDD analyses confirmed that inpatient length of stay increased in public compared to nonprofit hospitals in treated relative to control MSAs after policy implementation. Conclusions The impacts of the CJR program differed by hospital ownership type. Government-owned hospitals, with their unique financial circumstances, may have faced challenges that hindered the reductions in inpatient length of stay observed in other types of hospitals under the CJR Model.

Objectives: This study aims to synthesize current knowledge and outcomes related to pediatric auditory brainstem implantation (ABI) in children with severe inner ear malformations (IEMs). It highlights the clinical management practices, challenges, and potential future directions for consensus development in this field. Methods: A systematic review of findings presented at the Third International Pediatric ABI Symposium organized by the Hacettepe Cochlear Implant team between 3 and 5 September 2020 was conducted, incorporating data from 41 departments across 19 countries. Relevant clinical outcomes, imaging techniques, surgical approaches, and rehabilitation strategies were analyzed to identify key trends and variability in practices. Results: The review indicates that children receiving ABIs exhibit diverse auditory outcomes influenced by individual anatomical variations and developmental factors. Early implantation, particularly before the age of three, positively correlates with better auditory and language development. Multicenter experiences underscore the necessity of tailored decision-making, which considers both surgical candidacy and comprehensive rehabilitation resources. Discussion: The variability in outcomes emphasizes the need for improved consensus and guidelines regarding eligibility, surgical techniques, and multidisciplinary rehabilitation approaches. Notable complications and the necessity for thorough imaging assessments were also identified as critical components affecting clinical decisions. Conclusion: A formal consensus statement is warranted to standardize best practices in ABI management. This will not only enhance patient outcomes but also guide future research efforts to address the remaining challenges in the treatment of children with severe IEMs. Enhanced collaboration among team members will be pivotal in achieving these objectives.