Automated analysis of limited echocardiograms: Feasibility and relationship to outcomes in COVID-19.

Background: As automated echocardiographic analysis is increasingly utilized, continued evaluation within hospital settings is important to further understand its potential value. The importance of cardiac involvement in patients hospitalized with COVID-19 provides an opportunity to evaluate the feasibility and clinical relevance of automated analysis applied to limited echocardiograms. Methods: In this multisite US cohort, the feasibility of automated AI analysis was evaluated on 558 limited echocardiograms in patients hospitalized with COVID-19. Reliability of automated assessment of left ventricular (LV) volumes, ejection fraction (EF), and LV longitudinal strain (LS) was assessed against clinically obtained measures and echocardiographic findings. Automated measures were evaluated against patient outcomes using ROC analysis, survival modeling, and logistic regression for the outcomes of 30-day mortality and in-hospital sequelae. Results: Feasibility of automated analysis for both LVEF and LS was 87.5% (488/558 patients). AI analysis was performed with biplane method in 300 (61.5%) and single plane apical 4- or 2-chamber analysis in 136 (27.9%) and 52 (10.7%) studies, respectively. Clinical LVEF was assessed using visual estimation in 192 (39.3%), biplane in 163 (33.4%), and single plane or linear methods in 104 (21.2%) of the 488 studies; 29 (5.9%) studies did not have clinically reported LVEF. LV LS was clinically reported in 80 (16.4%). Consistency between automated and clinical values demonstrated Pearson's R, root mean square error (RMSE) and intraclass correlation coefficient (ICC) of 0.61, 11.3% and 0.72, respectively, for LVEF; 0.73, 3.9% and 0.74, respectively for LS; 0.76, 24.4ml and 0.87, respectively, for end-diastolic volume; and 0.82, 12.8 ml, and 0.91, respectively, for end-systolic volume. Abnormal automated measures of LVEF and LS were associated with LV wall motion abnormalities, left atrial enlargement, and right ventricular dysfunction. Automated analysis was associated with outcomes, including survival. Conclusion: Automated analysis was highly feasible on limited echocardiograms using abbreviated protocols, consistent with equivalent clinically obtained metrics, and associated with echocardiographic abnormalities and patient outcomes.

Human versus Artificial Intelligence-Based Echocardiographic Analysis as a Predictor of Outcomes: An Analysis from the World Alliance Societies of Echocardiography COVID Study.

BACKGROUND: Transthoracic echocardiography is the leading cardiac imaging modality for patients admitted with COVID-19, a condition of high short-term mortality. The aim of this study was to test the hypothesis that artificial intelligence (AI)-based analysis of echocardiographic images could predict mortality more accurately than conventional analysis by a human expert. METHODS: Patients admitted to 13 hospitals for acute COVID-19 who underwent transthoracic echocardiography were included. Left ventricular ejection fraction (LVEF) and left ventricular longitudinal strain (LVLS) were obtained manually by multiple expert readers and by automated AI software. The ability of the manual and AI analyses to predict all-cause mortality was compared. RESULTS: In total, 870 patients were enrolled. The mortality rate was 27.4% after a mean follow-up period of 230 ± 115 days. AI analysis had lower variability than manual analysis for both LVEF (P = .003) and LVLS (P = .005). AI-derived LVEF and LVLS were predictors of mortality in univariable and multivariable regression analysis (odds ratio, 0.974 [95% CI, 0.956-0.991; P = .003] for LVEF; odds ratio, 1.060 [95% CI, 1.019-1.105; P = .004] for LVLS), but LVEF and LVLS obtained by manual analysis were not. Direct comparison of the predictive value of AI versus manual measurements of LVEF and LVLS showed that AI was significantly better (P = .005 and P = .003, respectively). In addition, AI-derived LVEF and LVLS had more significant and stronger correlations to other objective biomarkers of acute disease than manual reads. CONCLUSIONS: AI-based analysis of LVEF and LVLS had similar feasibility as manual analysis, minimized variability, and consequently increased the statistical power to predict mortality. AI-based, but not manual, analyses were a significant predictor of in-hospital and follow-up mortality.

Ventricular Changes in Patients with Acute COVID-19 Infection: Follow-up of the World Alliance Societies of Echocardiography (WASE-COVID) Study.

BACKGROUND: COVID-19 infection is known to cause a wide array of clinical chronic sequelae, but little is known regarding the long-term cardiac complications. We aim to report echocardiographic follow-up findings and describe the changes in left (LV) and right ventricular (RV) function that occur following acute infection. METHODS: Patients enrolled in the World Alliance Societies of Echocardiography-COVID study with acute COVID-19 infection were asked to return for a follow-up transthoracic echocardiogram. Overall, 198 returned at a mean of 129 days of follow-up, of which 153 had paired baseline and follow-up images that were analyzable, including LV volumes, ejection fraction (LVEF), and longitudinal strain (LVLS). Right-sided echocardiographic parameters included RV global longitudinal strain, RV free wall strain, and RV basal diameter. Paired echocardiographic parameters at baseline and follow-up were compared for the entire cohort and for subgroups based on the baseline LV and RV function. RESULTS: For the entire cohort, echocardiographic markers of LV and RV function at follow-up were not significantly different from baseline (all P > .05). Patients with hyperdynamic LVEF at baseline (>70%), had a significant reduction of LVEF at follow-up (74.3% ± 3.1% vs 64.4% ± 8.1%, P < .001), while patients with reduced LVEF at baseline (<50%) had a significant increase (42.5% ± 5.9% vs 49.3% ± 13.4%, P = .02), and those with normal LVEF had no change. Patients with normal LVLS (<-18%) at baseline had a significant reduction of LVLS at follow-up (-21.6% ± 2.6% vs -20.3% ± 4.0%, P = .006), while patients with impaired LVLS at baseline had a significant improvement at follow-up (-14.5% ± 2.9% vs -16.7% ± 5.2%, P < .001). Patients with abnormal RV global longitudinal strain (>-20%) at baseline had significant improvement at follow-up (-15.2% ± 3.4% vs -17.4% ± 4.9%, P = .004). Patients with abnormal RV basal diameter (>4.5 cm) at baseline had significant improvement at follow-up (4.9 ± 0.7 cm vs 4.6 ± 0.6 cm, P = .019). CONCLUSIONS: Overall, there were no significant changes over time in the LV and RV function of patients recovering from COVID-19 infection. However, differences were observed according to baseline LV and RV function, which may reflect recovery from the acute myocardial injury occurring in the acutely ill. Left ventricular and RV function tends to improve in those with impaired baseline function, while it tends to decrease in those with hyperdynamic LV or normal RV function.

Echocardiographic Correlates of In-Hospital Death in Patients with Acute COVID-19 Infection: The World Alliance Societies of Echocardiography (WASE-COVID) Study.

BACKGROUND: The novel severe acute respiratory syndrome coronavirus-2 virus, which has led to the global coronavirus disease-2019 (COVID-19) pandemic is known to adversely affect the cardiovascular system through multiple mechanisms. In this international, multicenter study conducted by the World Alliance Societies of Echocardiography, we aim to determine the clinical and echocardiographic phenotype of acute cardiac disease in COVID-19 patients, to explore phenotypic differences in different geographic regions across the world, and to identify parameters associated with in-hospital mortality. METHODS: We studied 870 patients with acute COVID-19 infection from 13 medical centers in four world regions (Asia, Europe, United States, Latin America) who had undergone transthoracic echocardiograms. Clinical and laboratory data were collected, including patient outcomes. Anonymized echocardiograms were analyzed with automated, machine learning-derived algorithms to calculate left ventricular (LV) volumes, ejection fraction, and LV longitudinal strain (LS). Right-sided echocardiographic parameters that were measured included right ventricular (RV) LS, RV free-wall strain (FWS), and RV basal diameter. Multivariate regression analysis was performed to identify clinical and echocardiographic parameters associated with in-hospital mortality. RESULTS: Significant regional differences were noted in terms of patient comorbidities, severity of illness, clinical biomarkers, and LV and RV echocardiographic metrics. Overall in-hospital mortality was 21.6%. Parameters associated with mortality in a multivariate analysis were age (odds ratio [OR] = 1.12 [1.05, 1.22], P = .003), previous lung disease (OR = 7.32 [1.56, 42.2], P = .015), LVLS (OR = 1.18 [1.05, 1.36], P = .012), lactic dehydrogenase (OR = 6.17 [1.74, 28.7], P = .009), and RVFWS (OR = 1.14 [1.04, 1.26], P = .007). CONCLUSIONS: Left ventricular dysfunction is noted in approximately 20% and RV dysfunction in approximately 30% of patients with acute COVID-19 illness and portend a poor prognosis. Age at presentation, previous lung disease, lactic dehydrogenase, LVLS, and RVFWS were independently associated with in-hospital mortality. Regional differences in cardiac phenotype highlight the significant differences in patient acuity as well as echocardiographic utilization in different parts of the world.

A Machine Learning Approach for the Automated Interpretation of Plasma Amino Acid Profiles.

BACKGROUND: Plasma amino acid (PAA) profiles are used in routine clinical practice for the diagnosis and monitoring of inherited disorders of amino acid metabolism, organic acidemias, and urea cycle defects. Interpretation of PAA profiles is complex and requires substantial training and expertise to perform. Given previous demonstrations of the ability of machine learning (ML) algorithms to interpret complex clinical biochemistry data, we sought to determine if ML-derived classifiers could interpret PAA profiles with high predictive performance. METHODS: We collected PAA profiling data routinely performed within a clinical biochemistry laboratory (2084 profiles) and developed decision support classifiers with several ML algorithms. We tested the generalization performance of each classifier using a nested cross-validation (CV) procedure and examined the effect of various subsampling, feature selection, and ensemble learning strategies. RESULTS: The classifiers demonstrated excellent predictive performance, with the 3 ML algorithms tested producing comparable results. The best-performing ensemble binary classifier achieved a mean precision-recall (PR) AUC of 0.957 (95% CI 0.952, 0.962) and the best-performing ensemble multiclass classifier achieved a mean F4 score of 0.788 (0.773, 0.803). CONCLUSIONS: This work builds upon previous demonstrations of the utility of ML-derived decision support tools in clinical biochemistry laboratories. Our findings suggest that, pending additional validation studies, such tools could potentially be used in routine clinical practice to streamline and aid the interpretation of PAA profiles. This would be particularly useful in laboratories with limited resources and large workloads. We provide the necessary code for other laboratories to develop their own decision support tools.

Using Machine Learning to Aid the Interpretation of Urine Steroid Profiles.

BACKGROUND: Urine steroid profiles are used in clinical practice for the diagnosis and monitoring of disorders of steroidogenesis and adrenal pathologies. Machine learning (ML) algorithms are powerful computational tools used extensively for the recognition of patterns in large data sets. Here, we investigated the utility of various ML algorithms for the automated biochemical interpretation of urine steroid profiles to support current clinical practices. METHODS: Data from 4619 urine steroid profiles processed between June 2012 and October 2016 were retrospectively collected. Of these, 1314 profiles were used to train and test various ML classifiers' abilities to differentiate between "No significant abnormality" and "?Abnormal" profiles. Further classifiers were trained and tested for their ability to predict the specific biochemical interpretation of the profiles. RESULTS: The best performing binary classifier could predict the interpretation of No significant abnormality and ?Abnormal profiles with a mean area under the ROC curve of 0.955 (95% CI, 0.949-0.961). In addition, the best performing multiclass classifier could predict the individual abnormal profile interpretation with a mean balanced accuracy of 0.873 (0.865-0.880). CONCLUSIONS: Here we have described the application of ML algorithms to the automated interpretation of urine steroid profiles. This provides a proof-of-concept application of ML algorithms to complex clinical laboratory data that has the potential to improve laboratory efficiency in a setting of limited staff resources.

Phenolic metabolites of anthocyanins modulate mechanisms of endothelial function.

Anthocyanins are reported to have vascular bioactivity, however their mechanisms of action are largely unknown. Evidence suggests that anthocyanins modulate endothelial function, potentially by increasing nitric oxide (NO) synthesis, or enhancing NO bioavailability. This study compared the activity of cyanidin-3-glucoside, its degradation product protocatechuic acid, and phase II metabolite, vanillic acid. Production of NO and superoxide and expression of endothelial NO synthase (eNOS), NADPH oxidase (NOX), and heme oxygenase-1 (HO-1) were established in human vascular cell models. Nitric oxide levels were not modulated by the treatments, although eNOS was upregulated by cyanidin-3-glucoside, and superoxide production was decreased by both phenolic acids. Vanillic acid upregulated p22(phox) mRNA but did not alter NOX protein expression, although trends were observed for p47(phox) downregulation and HO-1 upregulation. Anthocyanin metabolites may therefore modulate vascular reactivity by inducing HO-1 and modulating NOX activity, resulting in reduced superoxide production and improved NO bioavailability.

Anthocyanins remain stable during commercial blackcurrant juice processing.

UNLABELLED: It remains important to establish the stability of anthocyanins throughout commercial processing in order to maintain the bioactivity of the processed food/s. The present study aimed to assess the recovery and formation of anthocyanins and their free phenolic acid degradation products during the commercial processing of blackcurrant juice concentrate. A bench-scale processing model was also established to allow for alteration of predefined parameters to identify where commercial processes could be modified to influence anthocyanin yield. No significant loss in anthocyanins was observed throughout the commercial processing of blackcurrants, from whole berry through milling, to pectin hydrolysis and sodium bisulphite addition (P ≥ 0.7). No significant loss in anthocyanins was observed following the subsequent processing of pressed juice, through pasteurization, decantation, filtration, and concentration (P ≥ 0.9). Similarly, the bench-scale model showed no significant losses in anthocyanin content except during pasteurization (22%± 0.7%, P < 0.001). In the full-factorial Design of Experiment model analysis, only sodium bisulphite concentration had an impact on anthocyanin recovery, which resulted in an increase (23% to 27%; P < 0.001) in final anthocyanin concentration. No phenolic degradation products (free protocatechuic acid or gallic acid derived from cyanidin and delphinin species, respectively) were identified in any processed sample when compared to authentic analytical standards, analyzed by ultra-performance liquid chromatography DAD. PRACTICAL APPLICATION: This article provides crucial data directly applicable to commercial juice processing, such as improving anthocyanin yield and practical considerations for anthocyanin stability and degradation. This aspect is particularly pertinent considering the current commercial interest in anthocyanin-derived phenolic acids and their health-related benefits. Further research and development targets in the area of commercial juice product development are identified.

Anthocyanin-derived phenolic acids form glucuronides following simulated gastrointestinal digestion and microsomal glucuronidation.

SCOPE: Current research indicates that anthocyanins are primarily degraded to form phenolic acid products. However, no studies have yet demonstrated the metabolic conjugation of these anthocyanin-derived phenolic acids in humans. METHODS AND RESULTS: Within the present study, a simulated gastrointestinal digestion model was used to evaluate the potential degradation of anthocyanins post-consumption. Subsequently, cyanidin (Cy) and pelargonidin and their degradation products, protocatechuic acid and 4-hydroxybenzoic acid, were incubated in the presence of human liver microsomes to assess their potential to form hepatic glucuronide conjugates. For structural conformation, phenolic glucuronides were chemically synthesised and compared to the microsomal metabolites. During the simulated gastric digestion, anthocyanin glycosides (200 µM) remained stable however their aglycone derivatives were significantly degraded (20% loss), while during subsequent pancreatic/intestinal digestion only pelargonidin-3-glucoside remained stable while cyanidin-3-glucoside (30% loss) and Cy and pelagonidin aglycones were significantly degraded (100% loss, respectively). Following microsomal metabolism, pelargonidin formed 4-hydroxybenzoic acid, which was further metabolised (65%) to form two additional glucuronide conjugates, while Cy formed protocatechuic acid, which was further metabolised (43%) to form three glucuronide conjugates. CONCLUSIONS: We propose that following ingestion, anthocyanins may be found in the systemic circulation as free or conjugated phenolic acids, which should be a focus of future dietary interventions.