2-Dimensional Echocardiographic Global Longitudinal Strain With Artificial Intelligence Using Open Data From a UK-Wide Collaborative.

BACKGROUND: Global longitudinal strain (GLS) is reported to be more reproducible and prognostic than ejection fraction. Automated, transparent methods may increase trust and uptake. OBJECTIVES: The authors developed open machine-learning-based GLS methodology and validate it using multiexpert consensus from the Unity UK Echocardiography AI Collaborative. METHODS: We trained a multi-image neural network (Unity-GLS) to identify annulus, apex, and endocardial curve on 6,819 apical 4-, 2-, and 3-chamber images. The external validation dataset comprised those 3 views from 100 echocardiograms. End-systolic and -diastolic frames were each labelled by 11 experts to form consensus tracings and points. They also ordered the echocardiograms by visual grading of longitudinal function. One expert calculated global strain using 2 proprietary packages. RESULTS: The median GLS, averaged across the 11 individual experts, was -16.1 (IQR: -19.3 to -12.5). Using each case's expert consensus measurement as the reference standard, individual expert measurements had a median absolute error of 2.00 GLS units. In comparison, the errors of the machine methods were: Unity-GLS 1.3, proprietary A 2.5, proprietary B 2.2. The correlations with the expert consensus values were for individual experts 0.85, Unity-GLS 0.91, proprietary A 0.73, proprietary B 0.79. Using the multiexpert visual ranking as the reference, individual expert strain measurements found a median rank correlation of 0.72, Unity-GLS 0.77, proprietary A 0.70, and proprietary B 0.74. CONCLUSIONS: Our open-source approach to calculating GLS agrees with experts' consensus as strongly as the individual expert measurements and proprietary machine solutions. The training data, code, and trained networks are freely available online.

Automated Left Ventricular Dimension Assessment Using Artificial Intelligence Developed and Validated by a UK-Wide Collaborative.

BACKGROUND: requires training and validation to standards expected of humans. We developed an online platform and established the Unity Collaborative to build a dataset of expertise from 17 hospitals for training, validation, and standardization of such techniques. METHODS: The training dataset consisted of 2056 individual frames drawn at random from 1265 parasternal long-axis video-loops of patients undergoing clinical echocardiography in 2015 to 2016. Nine experts labeled these images using our online platform. From this, we trained a convolutional neural network to identify keypoints. Subsequently, 13 experts labeled a validation dataset of the end-systolic and end-diastolic frame from 100 new video-loops, twice each. The 26-opinion consensus was used as the reference standard. The primary outcome was precision SD, the SD of the differences between AI measurement and expert consensus. RESULTS: In the validation dataset, the AI's precision SD for left ventricular internal dimension was 3.5 mm. For context, precision SD of individual expert measurements against the expert consensus was 4.4 mm. Intraclass correlation coefficient between AI and expert consensus was 0.926 (95% CI, 0.904-0.944), compared with 0.817 (0.778-0.954) between individual experts and expert consensus. For interventricular septum thickness, precision SD was 1.8 mm for AI (intraclass correlation coefficient, 0.809; 0.729-0.967), versus 2.0 mm for individuals (intraclass correlation coefficient, 0.641; 0.568-0.716). For posterior wall thickness, precision SD was 1.4 mm for AI (intraclass correlation coefficient, 0.535 [95% CI, 0.379-0.661]), versus 2.2 mm for individuals (0.366 [0.288-0.462]). We present all images and annotations. This highlights challenging cases, including poor image quality and tapered ventricles. CONCLUSIONS: Experts at multiple institutions successfully cooperated to build a collaborative AI. This performed as well as individual experts. Future echocardiographic AI research should use a consensus of experts as a reference. Our collaborative welcomes new partners who share our commitment to publish all methods, code, annotations, and results openly.

The CXCL16 A181V mutation selectively inhibits monocyte adhesion to CXCR6 but is not associated with human coronary heart disease.

OBJECTIVE: The chemokine CXCL16 serves as a scavenger receptor for oxidized low-density lipoprotein and as an adhesion molecule and chemoattractant for cells expressing the receptor CXCR6. A commonly occurring CXCL16 allele has been described containing 2 nonsynonymous single-nucleotide polymorphisms in complete linkage disequilibrium, although the effects on CXCL16 function are unknown. Here, we examined the effect of the single-nucleotide polymorphisms on CXCL16 function and assessed the association of the mutant allele with coronary heart disease (CHD). METHODS AND RESULTS: Both wild-type and mutant T123V181-CXCL16 were readily expressed in vitro and were similarly functional in assays of oxidized low-density lipoprotein scavenging and chemotaxis. However, unlike wild-type CXCL16, T123V181-CXCL16 was unable to promote adhesion of CXCR6(+) cells. Findings were confirmed ex vivo, with monocytes from donors homozygous for the T123V181 allele unable to facilitate adhesion of CXCR6 transfectants. In the London Life Sciences Prospective Population cohort (n = 2797), we found that the T123V181 allele was not associated with protection or susceptibility to CHD (adjusted odds ratio, 1.01; 95% CI, 0.95 to 1.10; P = 0.74). CONCLUSIONS: CXCL16-mediated cell adhesion plays at best a modest role in CHD, and the scavenging and chemotactic properties of the chemokine are more likely to be more important in disease pathogenesis.

Variability in the concentrations of intravenous drug infusions prepared in a critical care unit.

OBJECTIVE: To quantify the variability in the concentration of drug infusions prepared on an intensive care unit and establish whether there was a relationship between the quality of syringe labeling and drug preparation. DESIGN: Audit carried out over 3 weeks in May 2006 and completed in May 2007. SETTING: The adult neurosciences critical care unit of a UK university teaching hospital. INTERVENTIONS: Daily collections of discarded syringes containing midazolam, insulin, norepinephrine, dopamine, potassium or magnesium. MEASUREMENTS AND RESULTS: Residual solutions in the syringes were sampled and the concentrations measured. Syringe labels were inspected and awarded a score for labeling quality based on an 11-point scale. A total of 149 syringes were analyzed. Six of the magnesium syringes contained 4-5 times too much Mg(2+), presumably because of confusion about converting millimoles to grams. The majority of the other infusions differed from the expected concentration by more than 10%. Magnesium infusions were least likely to be properly labeled (p= 0.012), and there was a positive correlation between quality of syringe labeling and drug preparation (p=0.002). After the introduction of a new electrolyte prescription chart, magnesium and potassium preparation significantly improved but there was still substantial variability. CONCLUSIONS: These findings present a strong argument for the use of pre-prepared syringes or standardized drug preparation and labeling systems. They also highlight once again the difficulties healthcare professionals encounter when dealing with different ways of expressing drug concentrations.