Conventional Metrics Research Articles

Medial temporal lobe gray matter microstructure alterations provide improved sensitivity to detect neurodegeneration associated with Alzheimer’s disease biomarkers

AbstractBackgroundDespite the strengths conventional structural imaging measures to detect atrophy, cross‐sectional relationships between neurodegeneration and Alzheimer’s disease (AD) biomarkers in early stages is limited. Neurite orientation dispersion and density imaging (NODDI) measures microstructural alterations that are unique from conventional structural metrics. Alterations in microstructure have been detected in individuals with AD compared to cognitively normal individuals, but there is sparse data regarding the relationship between these metrics and AD biomarkers. Here we sought to characterize how NODDI metrics in the medial temporal lobe (MTL) are associated with plasma AD biomarkers.MethodOne‐hundred and sixty‐seven cognitively normal (CN) individuals, 47 individuals with mild cognitive impairment, and 16 individuals with AD underwent diffusion and T1w imaging. T1w images were segmented using the ASHS‐T1 protocol to form anterior and posterior MTL ROIs. Multi‐shell DTI data were collected in 113 non‐collinear directions with 1.5mm isotropic voxels, were pre‐processed using a custom pipeline, and fit with voxel‐wise NODDI models. ANTs was used to linearly register NODDI and T1w images to allow transformation of MTL ROIs to diffusion space. Stepwise‐regression models examined relationships between anterior and posterior MTL NODDI metrics with plasma biomarkers in CN and impaired (MCI/AD) groups after accounting for MTL volume.ResultNODDI metrics of intracellular volume fraction (icVf) and orientation dispersion index (ODI) were decreased, and isometric volume fraction (isoVf) was increased in cognitively impaired compared to CN. Amyloid positivity was associated with decreased ODI and increased isoVf in cognitively impaired individuals. While posterior MTL volume predicted plasma p‐Tau‐181, inclusion of NODDI metrics subsumed this effect (F‐change>7.54, p<0.0067), and reduced ODI was the best predictor of plasma pTau‐181 within cognitively impaired individuals, while isoVf best predicted pTau‐181 across all individuals. Posterior MTL volume did not predict plasma GFAP, but inclusion of ODI and icVf both significantly predicted plasma GFAP in both CN and impaired individuals (F‐change>12.68, p<.001). Anterior MTL icVf predicted both pTau‐181 and GFAP.ConclusionMicrostructural alterations measured by NODDI appear more closely linked to AD‐related molecular measures and, thus, may provide a more sensitive cross‐sectional measure of neurodegeneration of the MTL compared to conventional macrostructural metrics.

Medial temporal lobe gray matter microstructure alterations provide improved sensitivity to detect neurodegeneration associated with Alzheimer’s disease biomarkers

AbstractBackgroundDespite the strengths conventional structural imaging measures to detect atrophy, cross‐sectional relationships between neurodegeneration and Alzheimer’s disease (AD) biomarkers in early stages is limited. Neurite orientation dispersion and density imaging (NODDI) measures microstructural alterations that are unique from conventional structural metrics. Alterations in microstructure have been detected in individuals with AD compared to cognitively normal individuals, but there is sparse data regarding the relationship between these metrics and AD biomarkers. Here we sought to characterize how NODDI metrics in the medial temporal lobe (MTL) are associated with plasma AD biomarkers.MethodOne‐hundred and sixty‐seven cognitively normal (CN) individuals, 47 individuals with mild cognitive impairment, and 16 individuals with AD underwent diffusion and T1w imaging. T1w images were segmented using the ASHS‐T1 protocol to form anterior and posterior MTL ROIs. Multi‐shell DTI data were collected in 113 non‐collinear directions with 1.5mm isotropic voxels, were pre‐processed using a custom pipeline, and fit with voxel‐wise NODDI models. ANTs was used to linearly register NODDI and T1w images to allow transformation of MTL ROIs to diffusion space. Stepwise‐regression models examined relationships between anterior and posterior MTL NODDI metrics with plasma biomarkers in CN and impaired (MCI/AD) groups after accounting for MTL volume.ResultNODDI metrics of intracellular volume fraction (icVf) and orientation dispersion index (ODI) were decreased, and isometric volume fraction (isoVf) was increased in cognitively impaired compared to CN. Amyloid positivity was associated with decreased ODI and increased isoVf in cognitively impaired individuals. While posterior MTL volume predicted plasma p‐Tau‐181, inclusion of NODDI metrics subsumed this effect (F‐change>7.54, p<0.0067), and reduced ODI was the best predictor of plasma pTau‐181 within cognitively impaired individuals, while isoVf best predicted pTau‐181 across all individuals. Posterior MTL volume did not predict plasma GFAP, but inclusion of ODI and icVf both significantly predicted plasma GFAP in both CN and impaired individuals (F‐change>12.68, p<.001). Anterior MTL icVf predicted both pTau‐181 and GFAP.ConclusionMicrostructural alterations measured by NODDI appear more closely linked to AD‐related molecular measures and, thus, may provide a more sensitive cross‐sectional measure of neurodegeneration of the MTL compared to conventional macrostructural metrics.

Diffusion along the perivascular space as a potential biomarker for glioma grading and isocitrate dehydrogenase 1 mutation status prediction.

The diffusion tensor image analysis along the perivascular space (DTI-ALPS) may have the potential to reflect glymphatic dysfunction in patients with glioma. The study aimed to determine the correlation of DTI-ALPS with glioma grade and isocitrate dehydrogenase 1 (IDH1) genotype and to then compare the ALPS index with other diffusion metrics. In this study, 81 patients with glioma and 31 healthy controls underwent magnetic resonance imaging (MRI) examination. The ALPS-index, fractional anisotropy (FA), mean diffusivity (MD), and mean kurtosis (MK) were calculated. Comparisons were made between the left and right hemispheres and between patients and controls. IDH1 status was compared after age adjustment. The diagnostic performance of each metric was assessed via receiver operating characteristic (ROC) analysis. In patients with glioma, the ALPS-index of the hemisphere ipsilateral to glioma was significantly lower than that of the hemisphere contralateral to glioma (1.417±0.177 vs. 1.478±0.165; P=0.002), and the bilateral ALPS-index values in patients were significantly decreased compared with those in healthy controls. The ALPS-index was significantly higher in lower-grade gliomas (LrGGs) than that in glioblastomas (GBMs) (1.495±0.151 vs. 1.320±0.159; P<0.001) and was significantly lower in IDH1-wild-type LrGGs than in IDH1-mutant LrGGs (1.400±0.185 vs. 1.530±0.123; P=0.036). FA, MD, and MK also showed significant differences between LrGGs and GBMs and between IDH1-mutant and IDH1-wild-type LrGGs (P<0.05). Furthermore, the combination of the ALPS-index with FA, MD, or MK, exhibited superior discrimination ability compared to each metric used alone. The ALPS-index combined with MD had the highest area under the curve (AUC) of 0.854 as compared to that of 0.614-0.807 for a single metric in glioma grading, while for IDH1 mutation prediction, this combination had the highest AUC of 0.861 as compared to that of 0.707-0.778 for a single metric. The reduced ALPS-index may reflect tumor-induced glymphatic system impairment, and the ALPS-index may be able to complement conventional diffusion metrics in glioma grading and IDH1 genotyping.

SQAT: a MATLAB-based toolbox for quantitative sound quality analysis

Literature suggests that psychoacoustic metrics, which are established by models firmly based on the human auditory system, are able to better approximate perceptual attributes of sounds in a quantitative manner than conventional metrics based on the sound pressure level. Therefore, psychoacoustic metrics are commonly used as sound quality indicators in practical applications, such as product design or environmental noise assessment, in which human perception plays an important role. However, their calculation is not straightforward and the non-availability of verified, well-documented, open-source implementations in a single environment poses a challenge to reproducible research. In this contribution, we introduce the sound quality analysis toolbox (SQAT), an open-source compilation of MATLAB codes containing the implementation of the following key psychoacoustic metrics: loudness (ISO 532-1:2017), sharpness (DIN 45692:2009), roughness (Daniel &amp; Weber's model), fluctuation strength (Osses et al. model), and tonality (Aures' model). A comprehensive description of the implemented psychoacoustic metrics is provided. Furthermore, a verification study is presented in order to demonstrate the reliability of the implementations, supporting the use of SQAT as a reliable tool for future works in the field of sound quality assessment based on psychoacoustic metrics.

Reduction in retinal microvascular perfusion during migraine attacks.

To determine if there are changes in structure and function of the retinal vasculature during and between migraine attacks using optical coherence tomography angiography (OCTA). Migraine attacks commonly include visual symptoms, but the potential role of the retina in these symptoms is not well understood. OCTA is a rapid, non-invasive imaging technique that is used to visualize the retinal microvasculature with high spatial resolution in a clinical setting. In this study we used OCTA to quantify different features of the retinal vasculature in patients with migraine during and between attacks, as well as in healthy controls (HCs). We performed a prospective cohort study of 37 patients with migraine with aura (MA) (median [interquartile range, IQR] age of 37 [14] years, 86% female) and 30with migraine without aura (MO) (median [IQR] age of 37 [17] years, 77% female) and 20 HCs (median [IQR] age of 35 [7] years, 50% female). Macular OCTA scans were obtained for all participants for the interictal analysis. In 12 MA and eight MO, scans were captured both during and outside of migraine attacks and five HCs had initial and repeat scans. In addition to analyzing the morphology of the foveal avascular zone, we calculated the vessel flux index (VFI), which is an indicator of retinal perfusion and conventional metrics (such as vessel area density) in the foveal and parafoveal regions. There was a significant difference in the parafoveal VFI in the ictal state between the groups (p = 0.009). During migraine attacks there was a significant reduction in the parafoveal region VFI in MA (-7%, 95% confidence interval [CI] -10% to -4%; p = 0.006) and MO (-7%, 95% CI -10% to -3%; p = 0.016) from their interictal baseline as compared to the change between repeat scans in HCs (2%, 95% CI -3% to 7%). Interictally, there was a mean (standard deviation [SD]) 13% (10%) (p = 0.003) lower blood perfusion in the MA group as compared to the MO group in the foveal region (mean [SD] 0.093 [0.023] vs. 0.107 [0.021], p = 0.003). Interictal analysis also revealed higher circularity in the superficial foveal avascular zone in the MA group compared with the MO group (mean [SD] 0.686 [0.088] vs. 0.629 [0.120], p = 0.004). In addition, interictal analysis of the patients with MA or MO and unilateral headache showed increased retinal vascular parameters consistent with greater perfusion in the eye ipsilateral to the side of the pain as compared with the contralateral eye. These results indicate that perfusion is reduced in MA and MO in the parafoveal retina during the ictal period. Interictally, the foveal retina in MA has reduced perfusion when compared to the foveal retina in MO. Patients with unilateral headache showed interictal asymmetry of retinal perfusion between eyes. These results indicate that changes in retinal perfusion could be a part of migraine pathophysiology, and that distinct retinal vascular signatures identified with OCTA could represent biomarkers for migraine.

Use of sentiment analysis for capturing hospitalized cancer patients' experience from free-text comments in the Persian language

PurposeToday, the Internet provides access to many patients' experiences, which is crucial in assessing the quality of healthcare services. This paper introduces a model for detecting cancer patients' opinions about healthcare services in the Persian language, both positive and negative.MethodTo achieve the objectives of this study, a combination of sentiment analysis (SA) and topic modeling approaches was employed. All pertinent comments made by cancer patients were collected from the patient feedback form of the Tehran University of Medical Science (TUMS) Cancer Institute (CI) in Iran, from March to October 2021. Conventional evaluation metrics such as accuracy, precision, recall, and F-measure were utilized to assess the performance of the proposed model.ResultThe experimental findings revealed that the proposed SA model achieved accuracies of 89.3%, 92.6%, and 90.8% in detecting patients' sentiments towards general services, healthcare services, and life expectancy, respectively. Based on the topic modeling results, the topic "Metastasis" exhibited lower sentiment scores compared to other topics. Additionally, cancer patients expressed dissatisfaction with the current appointment booking service, while topics such as "Good experience," "Affable staff", and "Chemotherapy" garnered higher sentiment scores.ConclusionThe combined use of SA and topic modeling offers valuable insights into healthcare services. Policymakers can utilize the knowledge obtained from these topics and associated sentiments to enhance patient satisfaction with cancer institution services.

Sensor-aware CapsNet: Towards trustworthy multisensory fusion for remaining useful life prediction

Multisensory data-driven remaining useful life (RUL) prediction based on deep learning techniques is gaining increasing popularity as it captures the degradation process of mechanical equipment more comprehensively, the prediction performance of which is generally higher. However, less emphasis has been placed on enhancing the trustworthiness of the model, especially in terms of reasonability and interpretability in uncertainty-aware prediction. Aiming to solve such a problem, a novel framework called a sensor-aware capsule neural network (SACN) is proposed for multisensory fusion in RUL prediction. Then a novel Monte Carlo-based dropout method based on the activation level of capsules was developed for uncertainty quantification. Furthermore, the proposed method allows for multisensory fusion interpretation to attribute prediction results and uncertainties to the relevant features across sensors. A case study on C-MAPSS dataset shows that the proposed method outperforms the popular deep learning-based methods. In particular, the SACN based on capsule dropout achieves both greater accuracy and better reasonability than Gaussian dropout, as evaluated by conventional accuracy metrics and the proposed uncertainty coverage score, and the relationship between input features and final results can be revealed through multisensory fusion interpretation, which evaluates the contributions of each sensor for the given input.

Perception-based noise assessment of a future blended wing body aircraft concept using synthesized flyovers in an acoustic VR environment—The ARTEM study

New aircraft concepts are currently being developed with the goal of less emissions of CO2 and noise. Remarkable noise reductions in long-range aircraft can only be expected from disruptive vehicle designs, new propulsion systems and specific low-noise technologies. In this paper, one such future vehicle design, a blended wing body (BWB) long-range aircraft, is described and studied with respect to sound levels on the ground, sound characteristics and noise annoyance. Virtual flyovers of different vehicle variants were synthesized and auralized in an acoustic VR environment, and investigated through psychoacoustic laboratory experiments. The applied methodology was successfully hierarchically validated by comparison with measurements of existing jet aircraft, assessing acoustical indices, time-frequency features, perceived plausibility, and induced noise annoyance. The perception-based evaluation of the BWB revealed that, while the BWB aircraft may initially be perceived as somewhat more unfamiliar, they are substantially less annoying than current tube-and-wing long-range aircraft of similar range and mission for take-offs as well as for landings. For the best BWB variant, noise annoyance was reduced by 4.3 units for departures and by 3.5 units for approaches on the 11-point scale. The main reason for these findings seems to be the acoustic shielding by the body of the extended fuselage, which was found to be an important factor in reducing sound levels in the order of 10–20 dB, and accordingly also to strongly reduce loudness. Additional low noise technologies and geared turbofan engines with a high bypass ratio further contributed to the reduction of noise annoyance of the BWB. A large part of the BWBs benefit could be explained by its lower sound levels, but additional benefits were found. The observed reduction in noise annoyance was found to be larger than what can be explained with conventional noise metrics. This benefit is probably due to more favorable sound characteristics compared to today's reference aircraft, such as less variation in time and less audible tones. The current study thus suggests that the studied BWB vehicle concept may substantially reduce noise annoyance on humans.

Ensemble of CNN models for classification of groundnut plant leaf disease detection

Plant diseases pose a significant threat to the world's nutrition and can have severe consequences for smallholder farmers who rely on a thriving groundnut crop for their livelihoods. Therefore, it is crucial to create an algorithm for early automated diagnosis of plant diseases. Soa comprehensive dataset of groundnut leaf images was created in collaboration with a pathologist, facilitating the automated identification of plant diseases. The field of image classification has found CNN to be quite effective. In this study, we create such a method for disease identification and classification by utilizing a tri-CNN architecture consisting of DenseNet169, Inception, and Xception —that have been pre-trained on the ImageNet dataset using two created non-linear equations on decision scores from the before mentioned base learners,the outlined ensemble methodology generates ultimate predictions for the test samples by incorporating the scores of four conventional metrics for evaluation, namely recall, precision, accuracy, and f1-score, obtained from the base learners. Hence mentioned CNN customized models are used to train our model for obtaining better results. The proposed approach evaluated on real world groundnut leaf dataset. The model that was put forth attained accuracy rates of 98.46 %.

Stochastic resonance impact signal detection method based on a novel single potential well model

Our research introduces a novel stochastic resonance (SR) model featuring a single potential well and develops a dedicated detection system designed to address the challenging problem of detecting impact signals within a highly noisy background. We begin by examining the limitations of conventional metrics, such as the cross-correlation coefficient and kurtosis index, in identifying nonperiodic impact signals, and subsequently introduce an improved metric. By harnessing parameter-adjusted SR, this innovative potential well model and metric is integrated to formulate an adaptive detection method for nonperiodic impact signals. This method automatically adjusts system parameters in response to the input signal. Subsequently, numerical simulations of the system is conducted so as to perform a comparative analysis with experimental results obtained from both asymmetric single potential well and periodic potential systems. Our findings conclusively demonstrate the enhanced effectiveness of our proposed method in detecting impact signals within a high-noise environment. Furthermore, the method provides more accurate estimates of both the intensity and precise location of the input impact signal from the output results.

Oil price movements predictions in Kingdom of Saudi Arabia using financial and macro-economic variables

By combining financial and macroeconomic factors with machine learning techniques, this research paper proposes a novel method for forecasting oil price movements in the Kingdom of Saudi Arabia (KSA). Traditional methods generally struggle to capture the complex dynamics and nonlinear linkages in the oil market, which makes accurate oil price forecasting vital for decision-making in numerous sectors. In this paper, we offer a machine learning framework that leverages financial elements like stock market indices, currency rates, and interest rates, as well as macroeconomic data like GDP growth, inflation rates, and energy consumption, as predictors of oil price movements. These factors were chosen because of their significance and importance to the ways in which the oil market in KSA functions. We use several different machine learning techniques to construct the prediction models, some of which are regression-based (such as linear regression or support vector regression) while others are ensemble models (such as random forests or gradient boosting). The models are tested and refined using historical data spanning a sizable period of time and covering a wide range of market circumstances and pricing movements. Evaluation of the prediction models is carried out using conventional metrics like mean-squared error, mean absolute error, and R-squared [Formula: see text], and their robustness is evaluated using sensitivity analysis and cross-validation methods. Incorporating financial and macroeconomic variables vastly enhances the accuracy of oil price predictions compared to models based purely on historical price data, as shown by the preliminary findings, which underline the superiority of our approach. The machine learning models exhibit nonlinear pattern capture and responsiveness to market fluctuations. Insights into the relative significance of various variables and their effect on oil price movements in KSA are also provided by the sensitivity analysis. By demonstrating the usefulness of machine learning methods for oil price forecasting, particularly in the context of Saudi Arabia, this research adds to the body of knowledge already available. Our findings have important policy and market implications for oil price forecasting, benefiting policymakers, energy market participants, and investors. To improve the models’ ability to forecast the future, researchers may want to consider including more variables in their analyses, such as geopolitical developments and technology advances.

Current use of cardiac MRI in animals

Cardiovascular magnetic resonance (CMR) imaging has evolved to become an indispensable tool in human cardiology. It is a non-invasive technique that enables objective assessment of myocardial function, size, and tissue composition. Recent innovations in magnetic resonance imaging scanner technology and parallel imaging techniques have facilitated the generation of parametric mapping to explore tissue characteristics, and the emergence of strain imaging has enabled cardiologists to evaluate cardiac function beyond conventional metrics. As veterinary cardiology continues to utilize CMR beyond the reference standard, clinical application of CMR will further expand our capabilities. This article describes the current use of CMR and adoption of more recent advances such as T1/T2 mapping in veterinary cardiology.

Right ventricular pressure-strain-volume loop-derived myocardial work for the noninvasive assessment of volume overload-induced dysfunction in patients with secondary tricuspid regurgitation

Abstract Development of secondary tricuspid regurgitation (TR) is associated with poor outcomes in patients with left-sided cardiac diseases. Severe TR imposes substantial volume overload (VO) on the right ventricle (RV), inducing a vicious circle of progressive RV remodelling and dysfunction. Although previous studies have established the value of speckle tracking echocardiography (STE)-derived strain and pressure-strain loop-derived myocardial work indices, these metrics neglect chamber geometry, limiting their utility in conditions accompanied by VO. Accordingly, we aimed to (1) incorporate instantaneous RV volume into myocardial work calculation to define a method capable of detecting VO-induced RV dysfunction using 3D echocardiography (3DE) and (2) to test its prognostic power in patients with secondary TR. We enrolled 195 patients (age: 64±15 years, 69% male) with left-sided cardiac diseases and followed them for a median of 42 months. The primary endpoint was all-cause mortality. TR severity was graded according to current guidelines and we compared patients with significant (moderate+) TR (sTR, n=85) versus non-significant TR (nsTR, n=110). All patients underwent 3DE to measure left (LV), and RV volumes and ejection fractions (EF). A reference RV pressure curve was constructed using previously acquired right heart catheterization data and was individually adjusted by non-invasively estimated RV pressures. To obtain RV pressure-strain loops and global myocardial work index (GMWI), RV pressure and 3DE-derived RV circumferential strain tracings were concatenated. We further adjusted GMWI to instantaneous RV volumes (GMWIV) and constructed pressure-strain-volume loops (Figure 1). Twenty-three patients (11.8%) died during follow-up (16 out of 85 in the sTR group, and 7 out of 110 in the nsTR group). There were no differences in LV and RV volumes and EFs between the sTR and nsTR groups. Interestingly, patients with sTR had significantly higher values of GMWI (sTR vs. nsTR; 667 [438-879] vs. 567 [414-770] mmHg%, p=0.01), whereas GMWIV (292 [169-467] vs. 282 [170-399] mmHg%/mL, p=0.57) did not show difference. We compared patients with versus without adverse outcomes within the sTR group. Patients with sTR who died, presented with significantly lower volume-adjusted GMWIV values (dead vs. alive; 158 [122-332] vs. 312 [207-502] mmHg%/mL, p=0.02), while GMWI did not differ. By Cox regression analysis, GMWIV (hazard ratio [95% CI]: 0.997 [0.994-1.000], p&amp;lt;0.05) was a significant predictor of all-cause mortality in the entire population, whereas GMWI was not. By integrating RV volume into the pressure-strain relationship, our novel metric could detect the detrimental effects of VO on RV function with prognostic implications in patients with secondary TR. Our approach can be able to capture VO-induced subclinical RV dysfunction that conventional echocardiographic metrics could not unveil.Figure 1

Significant disagreements between conventional functional metrics and 3D echocardiography-derived ejection fraction in the detection of right ventricular systolic dysfunction

Abstract Background The conventional echocardiographic parameters of the right ventricular (RV) systolic function (e.g., tricuspid annular plane systolic excursion - TAPSE, fractional area change – FAC, free wall longitudinal strain – FWLS) only partially portray the complex functional characteristics of the RV; therefore, they may fail to capture the full spectrum of RV dysfunction and associated adverse clinical outcomes. 3D echocardiography-derived RV ejection fraction (RVEF) is a well-validated and reproducible parameter that overcomes these limitations. Purpose We aimed to investigate the discordance between TAPSE, FAC, FWLS and RVEF in RV systolic function grading and associated outcomes. Methods Two retrospective databases from different centers including consecutive patients with various cardiac diseases who underwent clinically indicated 2D and 3D echocardiography (n=750) were analysed. Patients were followed up for the occurrence of all-cause mortality. RVEF was measured by a single, commercially available 3D software package. Guideline-recommended cut-off values were used to indicate RV systolic dysfunction (TAPSE&amp;lt;17 mm, FAC&amp;lt;35%, FWLS&amp;gt;-20%). RVEF &amp;lt;45% served as the "ground truth" of RV dysfunction. Results Among patients with normal RVEF (n=511), 109 (21%) had reduced TAPSE, 168 (33%) had reduced FAC, and 42 (8%) had reduced FWLS. Among patients with reduced RVEF (n=239), 109 (46%) had normal TAPSE, 61 (26%) had normal FAC, and 98 (41%) had normal FWLS. Correspondingly, sensitivity and specificity for discrimination of RV systolic dysfunction (RVEF&amp;lt;45%) were 66% and 70% for TAPSE, 71% and 76% for FAC, and 89% and 69% for FWLS on ROC analysis, respectively (Figure 1). During the median follow-up time of 3.5 years, 112 patients (15%) died. Using univariable Cox regression, TAPSE (HR [95% CI]: 0.911 [0.881 - 0.942]), FAC (0.940 [0.924 - 0.957]), FWLS (1.101 [1.071 - 1.133], and RVEF (0.928 [0.913 - 0.944]) were all significant predictors of mortality (p&amp;lt;0.001). Interestingly, combining the conventional functional measures based on how many of them indicate dysfunction shows that outcomes are the worst if at least two parameters are impaired and gradually better if only one or none of them (Kaplan-Meier curves, Figure 2, log-rank p&amp;lt;0.001). Conclusion Guideline-recommended cut-off values of conventional echocardiographic parameters of RV systolic function are only modestly associated with RVEF-based assessment. Impaired values of FWLS show the closest association with the RVEF cut-off. Our results emphasize a multiparametric approach in the assessment of RV function, especially, if 3D echocardiography is not available.Figure 1.Figure 2.

Abstract 18776: ECG-GPT: Automated Complete Diagnosis Generation From ECG Images Using Novel Vision-Text Transformer Model

Background: Artificial intelligence can automate the diagnosis of a multitude of rhythm and conduction disorders from ECGs, but is limited by a dependency on discretely defined labels. Moreover, most models also frequently rely on signal data, limiting their application in low-resource settings, where such models are needed most and ECGs are available in image form. Methods: 1.4 million 12-lead ECGs and accompanying cardiologist diagnosis statements from Yale were used to develop a model capable of generating ECG report statements. This encompassed a wide array of conditions. To develop a model capable of report generation across ECG layouts, 12-lead signal recordings were plotted randomly in multiple formats. The gold-standard diagnosis statements were processed using to remove identifiable information and standardize diagnosis reporting. An ECG-specific vision encoder-decoder model with a pretrained BEiT transformer encoder with a 384x384 pixel input size and a generative pretrained transformer, GPT-2, as a decoder was developed (B). Results: In a held-out test set of 100,000 ECGs, representing ECG interpretation with up to 255 tokens and over 41 distinct cardiovascular conditions, the model performed well across conventional metrics, with ROUGE-L and BLEU-1 scores of 0.695 and 0.502, respectively (C). The model’s clinical validity was assessed by computing agreement between cardiologist and model-generated diagnosis statements. The model generated text reports with accurate identification of diverse set of rhythm and conduction disorders, with AUROCs for predicting atrial fibrillation, right bundle branch block, and sinus tachycardia of 0.95, 0.94, and 0.96, respectively (D). Conclusions: Our vision encoder-decoder model generates an ECG report from ECG images in any layout. This approach represents an automated and accessible strategy for generating cardiologist-level reporting on ECGs, useful for low-resource settings.

Exploring the effectiveness of fiscal decentralization in environmental expenditure based on the CO2 ecological footprint in urban China

Under the fiscal decentralization of government environmental management, this paper investigates the relationship between local environmental protection expenditure (LEPE) and CO2 ecological footprint (CEF). Unlike conventional emissions-based greenhouse gas metrics, this research calculates per capita CEF for 253 Chinese cities, considering both carbon emissions and absorption. The dynamic spatial Durbin model demonstrates that LEPE not only reduces a city’s own CEF but also exerts a long-term influence on neighboring cities, signifying competitive dynamics among local governments in LEPE. This decentraliztion of environmental budget authority may yield adverse outcomes. Further analysis reveals an inverted U-shaped relationship between CEF and LEPE intensity, partly mirroring the environmental Kuznets curve. Different developmental stages should consider economic levels when allocating resources to environmental budgets. The low-carbon pilot policy strengthens LEPE, with varying effects across Chinese urban agglomerations, remaining consistent post-environmental protection tax introduction. These findings hold critical reference value for local policymakers aiming to collaboratively adjust market-oriented environmental policies.

The geography of urban innovation beyond patents only: New evidence on large and secondary cities in the United States

The geography of urban innovation in the United States has often been portrayed as dominated by a few large cities. However, emerging perspectives challenge this notion by highlighting the significance of secondary cities for innovation processes. Unfortunately, prevailing narratives and empirical evidence focus on technological invention only, measured by patents. This study aims to reevaluate the geography of urban innovation by considering various phases of the innovation process and incorporating a broader range of innovation types. In addition to conventional patent metrics, this research suggests incorporating trademarks and design rights as complementary data. The findings reveal that large cities still exhibit prominence when examining absolute counts of innovation. Instead, analysing intensities allows the identification of secondary cities that specialise in various innovation activities beyond technological invention only. Overall, the findings provide compelling evidence that urban innovation in the United States is not limited to a few large cities nor is it solely driven by technological invention.

Symposium 1

Proteomics has been a very informative tool in uncovering several biomarkers that may have clinical utility in managing patients with lupus nephritis (LN). Both non-targeted (mass spectrometry-based) and targeted (ligand-based) proteomics point to promising biomarkers in the urine of LN patients. Aptamer-based interrogation of 1129 proteins using the Somascan platform, followed by ELISA-validation in two independent cohorts has indicated that the urine proteins that best distinguish active LN from controls were ALCAM, PF-4, properdin, and VCAM-1 among African Americans, E-selectin, VCAM-1, BFL-1 and Hemopexin among Caucasians, and ALCAM, VCAM-1, TFPI and PF-4 among Asians (PMID: 32366845). Most of these urine proteins correlate significantly with disease activity indices and surpass conventional metrics in identifying active LN. Several elevated urinary molecules are also expressed within LN kidneys, based on single-cell RNA-seq analysis. Of these, ALCAM has been validated in a large multi-ethnic, multi-national cohort of LN subjects, and has also been shown to be a pathogenic driver and therapeutic target in LN (PMID: 34981775). Antibody array based screening of 1000 proteins using the Raybiotech Kiloplex platform has identified urine L-selectin, Angptl4, TGF[Formula: see text] 1, TPP1, TSP-1, FOLR2, PDGFR[Formula: see text], and PRX2, as markers of active renal lupus, outperforming conventional metrics, with L-selectin, Angptl4 and TGF[Formula: see text] 1 tracking well with concurrent or pending disease flares in LN (PMID: 32651195). Of these, urine L-selectin has been validated in large multi-ethnic/national cohorts of LN subjects (under review). Proteomic screens of LN urine using an electrochemiluminescence platform (Mesoscale/MSD) has identified IL-7, IL-12p40, IL-15, IP-10, and TARC as markers of active renal lupus that correlated significantly with disease activity (PMID: 30618193). Very recent global proteomic screens of LN urine using a proximity ligation assay platform from Olink has added several additional biomarker candidates that have been successfully ELISA-validated (under review). Given the plethora of urine biomarkers that have been identified using proteomics, it now becomes important to validate the clinical utility of these urine proteins, using multiple well-annotated patient cohorts, across ethnicities. Among the completed studies, we now know that some urine proteins are diagnostic of concurrent renal pathology LN Class or Activity Index or Chronicity index (e.g. CD163: PMID: 32351512; VCAM-1: PMID: 29076253; Angiostatin: PMID: 29076253; Hemopexin, Kim-1 and Lipoxcalin-2/NGAL; Renin and SAP; PMID: 20065116), while others exhibit prognostic value (e.g., ALCAM and VCAM-1; PMID: 31722419). Ongoing studies also show that most of the urine proteins interrogated in LN are not elevated in circulation (PMID: 36091001), but are expressed within LN kidneys, as ascertained by immunohistochemistry (Clinical Immunology, 2022) and Imaging Mass Cytometry (Tissue CyTOF). To facilitate monitoring of urinary biomarkers in outpatient clinics or to enable self-monitoring of disease status by the patient from the comfort of one’s home, we have recently engineered a smartphone-readable nanophosphor-based lateral flow assay for point of care monitoring of urine ALCAM (Frontiers in Immunology, 2022). The long-term goal is to decipher the clinical utility of different urine biomarkers in managing lupus and LN in the clinics, to empower LN patients with easy-to-use tools to monitor their own disease status, and to understand the mechanistic significance of the uncovered molecules in renal disease pathogenesis. Ongoing research on this front is summarized at: https://mohanlab.bme.uh.edu/ .

Interpretable machine learning model based on the systemic inflammation response index and ultrasound features can predict central lymph node metastasis in cN0T1-T2 papillary thyroid carcinoma.

It is arguable whether individuals with T1-T2 papillary thyroid cancer (PTC) who have a clinically negative (cN0) diagnosis should undergo prophylactic central lymph node dissection (pCLND) on a routine basis. Many inflammatory indices, including the neutrophil-to-lymphocyte ratio (NLR), platelet-to-lymphocyte ratio (PLR), monocyte-to-lymphocyte ratio (MLR), and systemic immune-inflammatory index (SII), have been reported in PTC. However, the associations between the systemic inflammation response index (SIRI) and the risk of central lymph node metastasis (CLNM) remain unclear. Retrospective research involving 1,394 individuals with cN0T1-T2 PTC was carried out, and the included patients were randomly allocated into training (70%) and testing (30%) subgroups. The preoperative inflammatory indices and ultrasound (US) features were used to train the models. To assess the forecasting factors as well as drawing nomograms, the least absolute shrinkage and selection operator (LASSO) and multivariate logistic regression were utilized. Then eight interpretable models based on machine learning (ML) algorithms were constructed, including decision tree (DT), K-nearest neighbor (KNN), support vector machine (SVM), artificial neural network (ANN), random forest (RF), extreme gradient boosting (XGBoost), light gradient boosting machine (LightGBM), and categorical boosting (CatBoost). The performance of the models was evaluated by incorporating the area under the precision-recall curve (auPR) and the area under the receiver operating characteristic curve (auROC), as well as other conventional metrics. The interpretability of the optimum model was illustrated via the shapley additive explanations (SHAP) approach. Younger age, larger tumor size, capsular invasion, location (lower and isthmus), unclear margin, microcalcifications, color Doppler flow imaging (CDFI) blood flow, and higher SIRI (≥0.77) were independent positive predictors of CLNM, whereas female sex and Hashimoto thyroiditis were independent negative predictors, and nomograms were subsequently constructed. Taking into account both the auROC and auPR, the RF algorithm showed the best performance, and superiority to XGBoost, CatBoost and ANN. In addition, the role of key variables was visualized in the SHAP plot. An interpretable ML model based on the SIRI and US features can be used to predict CLNM in individuals with cN0T1-T2 PTC.

Changepoint detection-assisted nonparametric clustering for unsupervised temporal sign segmentation

Temporal sign segmentation aims to temporarily divide continuous sign language into category-agnostic sign segments. One of the challenges to temporal sign segmentation is the paucity of frame-level annotations for sign language videos, which restricts the applicability of supervised and semi-supervised approaches. To address this challenge, we consider temporal sign segmentation as a clustering problem defined as the grouping of semantically similar frames of a given video, which allows us to choose unsupervised clustering techniques as promising alternatives. However, most unsupervised clustering techniques are parametric. In the context of temporal sign segmentation, they assume that the number of sign segments of a given continuous sign video is predefined, which is implausible. The primary contributions of this study are as follows: (1) We propose the first nonparametric clustering algorithm for unsupervised temporal sign segmentation. The main concept is to enhance the hierarchical graph-based clustering algorithm to be nonparametric by adopting cost and penalty functions of a changepoint detection algorithm to determine the optimal number of sign segments. Experimental results show that the performance of the proposed unsupervised method is comparable to that of the latest semi-supervised sign segmentation method in terms of several metrics. Moreover, the execution time of the proposed clustering method was less than 1 s, thereby ensuring its applicability. (2) We identify that the conventional metrics for temporal sign segmentation do not sufficiently address over-segmentation. To overcome the difficulty, we propose a new main metric to evaluate the performance of temporal sign segmentation, called the adjusted MF1B.