A Hybrid Decision Tree and Deep Learning Approach Combining Medical Imaging and Electronic Medical Records to Predict Intubation Among Hospitalized Patients With COVID-19: Algorithm Development and Validation.

BACKGROUND: Early prediction of the need for invasive mechanical ventilation (IMV) in patients hospitalized with COVID-19 symptoms can help in the allocation of resources appropriately and improve patient outcomes by appropriately monitoring and treating patients at the greatest risk of respiratory failure. To help with the complexity of deciding whether a patient needs IMV, machine learning algorithms may help bring more prognostic value in a timely and systematic manner. Chest radiographs (CXRs) and electronic medical records (EMRs), typically obtained early in patients admitted with COVID-19, are the keys to deciding whether they need IMV. OBJECTIVE: We aimed to evaluate the use of a machine learning model to predict the need for intubation within 24 hours by using a combination of CXR and EMR data in an end-to-end automated pipeline. We included historical data from 2481 hospitalizations at The Mount Sinai Hospital in New York City. METHODS: CXRs were first resized, rescaled, and normalized. Then lungs were segmented from the CXRs by using a U-Net algorithm. After splitting them into a training and a test set, the training set images were augmented. The augmented images were used to train an image classifier to predict the probability of intubation with a prediction window of 24 hours by retraining a pretrained DenseNet model by using transfer learning, 10-fold cross-validation, and grid search. Then, in the final fusion model, we trained a random forest algorithm via 10-fold cross-validation by combining the probability score from the image classifier with 41 longitudinal variables in the EMR. Variables in the EMR included clinical and laboratory data routinely collected in the inpatient setting. The final fusion model gave a prediction likelihood for the need of intubation within 24 hours as well. RESULTS: At a prediction probability threshold of 0.5, the fusion model provided 78.9% (95% CI 59%-96%) sensitivity, 83% (95% CI 76%-89%) specificity, 0.509 (95% CI 0.34-0.67) F1-score, 0.874 (95% CI 0.80-0.94) area under the receiver operating characteristic curve (AUROC), and 0.497 (95% CI 0.32-0.65) area under the precision recall curve (AUPRC) on the holdout set. Compared to the image classifier alone, which had an AUROC of 0.577 (95% CI 0.44-0.73) and an AUPRC of 0.206 (95% CI 0.08-0.38), the fusion model showed significant improvement (P<.001). The most important predictor variables were respiratory rate, C-reactive protein, oxygen saturation, and lactate dehydrogenase. The imaging probability score ranked 15th in overall feature importance. CONCLUSIONS: We show that, when linked with EMR data, an automated deep learning image classifier improved performance in identifying hospitalized patients with severe COVID-19 at risk for intubation. With additional prospective and external validation, such a model may assist risk assessment and optimize clinical decision-making in choosing the best care plan during the critical stages of COVID-19.

Computed tomography aortic valve calcium scoring for the assessment of aortic stenosis progression.

OBJECTIVE: CT quantification of aortic valve calcification (CT-AVC) is useful in the assessment of aortic stenosis severity. Our objective was to assess its ability to track aortic stenosis progression compared with echocardiography. METHODS: Subjects were recruited in two cohorts: (1) a reproducibility cohort where patients underwent repeat CT-AVC or echocardiography within 4 weeks and (2) a disease progression cohort where patients underwent annual CT-AVC and/or echocardiography. Cohen's d-statistic (d) was computed from the ratio of annualised progression and measurement repeatability and used to estimate group sizes required to detect annualised changes in CT-AVC and echocardiography. RESULTS: A total of 33 (age 71±8) and 81 participants (age 72±8) were recruited to the reproducibility and progression cohorts, respectively. Ten CT scans (16%) were excluded from the progression cohort due to non-diagnostic image quality. Scan-rescan reproducibility was excellent for CT-AVC (limits of agreement -12% to 10 %, intraclass correlation (ICC) 0.99), peak velocity (-7% to +17%; ICC 0.92) mean gradient (-25% to 27%, ICC 0.96) and dimensionless index (-11% to +15%; ICC 0.98). Repeat measurements of aortic valve area (AVA) were less reliable (-44% to +28%, ICC 0.85).CT-AVC progressed by 152 (65-375) AU/year. For echocardiography, the median annual change in peak velocity was 0.1 (0.0-0.3) m/s/year, mean gradient 2 (0-4) mm Hg/year and AVA -0.1 (-0.2-0.0) cm2/year. Cohen's d-statistic was more than double for CT-AVC (d=3.12) than each echocardiographic measure (peak velocity d=0.71 ; mean gradient d=0.66; AVA d=0.59, dimensionless index d=1.41). CONCLUSION: CT-AVC is reproducible and demonstrates larger increases over time normalised to measurement repeatability compared with echocardiographic measures.

Different Lifestyle Interventions in Adults From Underserved Communities: The FAMILIA Trial.

BACKGROUND: The current trends of unhealthy lifestyle behaviors in underserved communities are disturbing. Thus, effective health promotion strategies constitute an unmet need. OBJECTIVES: The purpose of this study was to assess the impact of 2 different lifestyle interventions on parents/caregivers of children attending preschools in a socioeconomically disadvantaged community. METHODS: The FAMILIA (Family-Based Approach in a Minority Community Integrating Systems-Biology for Promotion of Health) study is a cluster-randomized trial involving 15 Head Start preschools in Harlem, New York. Schools, and their children's parents/caregivers, were randomized to receive either an "individual-focused" or "peer-to-peer-based" lifestyle intervention program for 12 months or control. The primary outcome was the change from baseline to 12 months in a composite health score related to blood pressure, exercise, weight, alimentation, and tobacco (Fuster-BEWAT Score [FBS]), ranging from 0 to 15 (ideal health = 15). To assess the sustainability of the intervention, this study evaluated the change of FBS at 24 months. Main pre-specified secondary outcomes included changes in FBS subcomponents and the effect of the knowledge of presence of atherosclerosis as assessed by bilateral carotid/femoral vascular ultrasound. Mixed-effects models were used to test for intervention effects. RESULTS: A total of 635 parents/caregivers were enrolled: mean age 38 ± 11 years, 83% women, 57% Hispanic/Latino, 31% African American, and a baseline FBS of 9.3 ± 2.4 points. The mean within-group change in FBS from baseline to 12 months was ∼0.20 points in all groups, with no overall between-group differences. However, high-adherence participants to the intervention exhibited a greater change in FBS than their low-adherence counterparts: 0.30 points (95% confidence interval: 0.03 to 0.57; p = 0.027) versus 0.00 points (95% confidence interval: -0.43 to 0.43; p = 1.0), respectively. Furthermore, the knowledge by the participant of the presence of atherosclerosis significantly boosted the intervention effects. Similar results were sustained at 24 months. CONCLUSIONS: Although overall significant differences were not observed between intervention and control groups, the FAMILIA trial highlights that high adherence rates to lifestyle interventions may improve health outcomes. It also suggests a potential contributory role of the presentation of atherosclerosis pictures, providing helpful information to improve future lifestyle interventions in adults.

Stress-Associated Neurobiological Pathway Linking Socioeconomic Disparities to Cardiovascular Disease.

BACKGROUND: Lower socioeconomic status (SES) associates with a higher risk of major adverse cardiac events (MACE) via mechanisms that are not well understood. OBJECTIVES: Because psychosocial stress is more prevalent among those with low SES, this study tested the hypothesis that stress-associated neurobiological pathways involving up-regulated inflammation in part mediate the link between lower SES and MACE. METHODS: A total of 509 individuals, median age 55 years (interquartile range: 45 to 66 years), underwent clinically indicated whole-body 18F-fluorodeoxyglucose positron emission tomography/computed tomography imaging and met pre-defined inclusion criteria, including absence of known cardiovascular disease or active cancer. Baseline hematopoietic tissue activity, arterial inflammation, and in a subset of 289, resting amygdalar metabolism (a measure of stress-associated neural activity) were quantified using validated 18F-fluorodeoxyglucose positron emission tomography/computed tomography methods. SES was captured by neighborhood SES factors (e.g., median household income and crime). MACE within 5 years of imaging was adjudicated. RESULTS: Over a median 4.0 years, 40 individuals experienced MACE. Baseline income inversely associated with amygdalar activity (standardized ß: -0.157 [95% confidence interval (CI): -0.266 to -0.041]; p = 0.007) and arterial inflammation (ß: -0.10 [95% CI: -0.18 to -0.14]; p = 0.022). Further, income associated with subsequent MACE (standardized hazard ratio: 0.67 [95% CI: 0.47 to 0.96]; p = 0.029) after multivariable adjustments. Mediation analysis demonstrated that the path of: ↓ neighborhood income to ↑ amygdalar activity to ↑ bone marrow activity to ↑ arterial inflammation to ↑ MACE was significant (ß: -0.01 [95% CI: -0.06 to -0.001]; p < 0.05). CONCLUSIONS: Lower SES: 1) associates with higher amygdalar activity; and 2) independently predicts MACE via a serial pathway that includes higher amygdalar activity, bone marrow activity, and arterial inflammation. These findings illuminate a stress-associated neurobiological mechanism by which SES disparities may potentiate adverse health outcomes.

Efficacy and safety assessment of a TRAF6-targeted nanoimmunotherapy in atherosclerotic mice and non-human primates.

Macrophage accumulation in atherosclerosis is directly linked to the destabilization and rupture of plaque, causing acute atherothrombotic events. Circulating monocytes enter the plaque and differentiate into macrophages, where they are activated by CD4+ T lymphocytes through CD40-CD40 ligand signalling. Here, we report the development and multiparametric evaluation of a nanoimmunotherapy that moderates CD40-CD40 ligand signalling in monocytes and macrophages by blocking the interaction between CD40 and tumour necrosis factor receptor-associated factor 6 (TRAF6). We evaluated the biodistribution characteristics of the nanoimmunotherapy in apolipoprotein E-deficient (Apoe-/-) mice and in non-human primates by in vivo positron-emission tomography imaging. In Apoe-/- mice, a 1-week nanoimmunotherapy treatment regimen achieved significant anti-inflammatory effects, which was due to the impaired migration capacity of monocytes, as established by a transcriptome analysis. The rapid reduction of plaque inflammation by the TRAF6-targeted nanoimmunotherapy and its favourable toxicity profiles in both mice and non-human primates highlights the translational potential of this strategy for the treatment of atherosclerosis.

Author Correction: Efficacy and safety assessment of a TRAF6-targeted nanoimmunotherapy in atherosclerotic mice and non-human primates.

In the version of this Article originally published, the surname of the author Edward A. Fisher was spelt incorrectly as 'Fischer'. This has now been corrected.

Rationale and Design of Family-Based Approach in a Minority Community Integrating Systems-Biology for Promotion of Health (FAMILIA).

BACKGROUND: The 2020 American Heart Association Impact Goal aims to improve cardiovascular health of all Americans by 20% while reducing deaths from cardiovascular disease and stroke by 20%. A large step toward this goal would be to better understand and take advantage of the significant intersection between behavior and biology across the entire life-span. In the proposed FAMILIA studies, we aim to directly address this major knowledge and clinical health gap by implementing an integrated family-centric health promotion intervention and focusing on the intersection of environment and behavior, while understanding the genetic and biologic basis of cardiovascular disease. METHODS: We plan to recruit 600 preschool children and their 600 parents or caregivers from 12-15 Head Start schools in Harlem, NY, and perform a 2:1 (2 intervention/1 control) cluster randomization of the schools. The preschool children will receive our intensive 37-hour educational program as the intervention for 4 months. For the adults, those in the "intervention" group will be randomly assigned to 1 of 2 intervention programs: an "individual-focused" or "peer-to-peer based." The primary outcome in children will be a composite score of knowledge (K), attitudes (A), habits (H), related to body mass index Z score (B), exercise (E), and alimentation (A) (KAH-BEA), using questionnaires and anthropometric measurements. For adults, the primary outcome will be a composite score for behaviors/outcomes related to blood pressure, exercise, weight, alimentation (diet) and tobacco (smoking; Fuster-BEWAT score). Saliva will be collected from the children for SNP genotyping, and blood will be collected from adults for RNA sequencing to identify network models and predictors of primary prevention outcomes. CONCLUSION: The FAMILIA studies seek to demonstrate that targeting a younger age group (3-5 years) and using a family-based approach may be a critical strategy in promoting cardiovascular health across the life-span.

Noninvasive assessment of hypoxia in rabbit advanced atherosclerosis using ¹8F-fluoromisonidazole positron emission tomographic imaging.

BACKGROUND: Hypoxia is an important microenvironmental factor influencing atherosclerosis progression by inducing foam-cell formation, metabolic adaptation of infiltrated macrophages, and plaque neovascularization. Therefore, imaging plaque hypoxia could serve as a marker of lesions at risk. METHODS AND RESULTS: Advanced aortic atherosclerosis was induced in 18 rabbits by atherogenic diet and double balloon endothelial denudation. Animals underwent (18)F-fluoromisonidazole positron emission tomographic and (18)F-fluorodeoxyglucose positron emission tomographic imaging after 6 to 8 months (atherosclerosis induction) and 12 to 16 months (progression) of diet initiation. Four rabbits fed standard chow served as controls. Radiotracer uptake of the abdominal aorta was measured using standardized uptake values. After imaging, plaque hypoxia (pimonidazole), macrophages (RAM-11), neovessels (CD31), and hypoxia-inducible factor-1α were assessed by immunohistochemistry.(18)F-fluoromisonidazole uptake increased with time on diet (standardized uptake value mean, 0.10±0.01 in nonatherosclerotic animals versus 0.20±0.03 [P=0.002] at induction and 0.25±0.03 [P<0.001] at progression). Ex vivo positron emission tomographic imaging corroborated the (18)F-fluoromisonidazole uptake by the aorta of atherosclerotic rabbits. (18)F-fluorodeoxyglucose uptake also augmented in atherosclerotic animals, with an standardized uptake value mean of 0.43±0.02 at induction versus 0.35±0.02 in nonatherosclerotic animals (P=0.031) and no further increase at progression. By immunohistochemistry, hypoxia was mainly located in the macrophage-rich areas within the atheromatous core, whereas the macrophages close to the lumen were hypoxia-negative. Intraplaque neovessels were found predominantly in macrophage-rich hypoxic regions (pimonidazole(+)/hypoxia-inducible factor-1α(+)/RAM-11(+)). CONCLUSIONS: Plaque hypoxia increases with disease progression and is present in macrophage-rich areas associated with neovascularization. (18)F-fluoromisonidazole positron emission tomographic imaging emerges as a new tool for the detection of atherosclerotic lesions.

Labeling galectin-3 for the assessment of myocardial infarction in rats.

BACKGROUND: Galectin-3 is a ß-galactoside-binding lectin expressed in most of tissues in normal conditions and overexpressed in myocardium from early stages of heart failure (HF). It is an established biomarker associated with extracellular matrix (ECM) turnover during myocardial remodeling. The aim of this study is to test the ability of (123)I-galectin-3 (IG3) to assess cardiac remodeling in a model of myocardial infarction (MI) using imaging techniques. METHODS: Recombinant galectin-3 was labeled with iodine-123 and in vitro binding assays were conducted to test (123)I-galectin-3 ability to bind to ECM targets. For in vivo studies, a rat model of induced-MI was used. Animals were subjected to magnetic resonance and micro-SPETC/micro-CT imaging two (2 W-MI) or four (4 W-MI) weeks after MI. Sham rats were used as controls. Pharmacokinetic, biodistribution, and histological studies were also performed after intravenous administration of IG3. RESULTS: In vitro studies revealed that IG3 shows higher binding affinity (measured as counts per minute, cpm) (p < 0.05) to laminin (2.45 ± 1.67 cpm), fibronectin (4.72 ± 1.95 cpm), and collagen type I (1.88 ± 0.53 cpm) compared to bovine serum albumin (BSA) (0.88 ± 0.31 cpm). Myocardial quantitative IG3 uptake (%ID/g) was higher (p < 0.01) in the infarct of 2 W-MI rats (0.15 ± 0.04%) compared to control (0.05 ± 0.03%). IG3 infarct uptake correlates with the extent of scar (r s = 1, p = 0.017). Total collagen deposition in the infarct (percentage area) was higher (p < 0.0001) at 2 W-MI (24.2 ± 5.1%) and 4 W-MI (30.4 ± 7.5%) compared to control (1.9 ± 1.1%). However, thick collagen content in the infarct (square micrometer stained) was higher at 4 W-MI (20.5 ± 11.2 µm(2)) compared to control (4.7 ± 2.0 µm(2), p < 0.001) and 2 W-MI (10.6 ± 5.1 µm(2), p < 0.05). CONCLUSIONS: This study shows, although preliminary, enough data to consider IG3 as a potential contrast agent for imaging of myocardial interstitial changes in rats after MI. Labeling strategies need to be sought to improve in vivo IG3 imaging, and if proven, galectin-3 might be used as an imaging tool for the assessment and treatment of MI patients.

Regression of inflammation in atherosclerosis by the LXR agonist R211945: a noninvasive assessment and comparison with atorvastatin.

OBJECTIVES: The aim of this study was to noninvasively detect the anti-inflammatory properties of the novel liver X receptor agonist R211945. BACKGROUND: R211945 induces reversal cholesterol transport and modulates inflammation in atherosclerotic plaques. We aimed to characterize with (18)F-fluorodeoxyglucose (FDG)-positron emission tomography (PET)/computed tomography (CT) and dynamic contrast-enhanced cardiac magnetic resonance (DCE-CMR) inflammation and neovascularization, respectively, in atherosclerotic plaques with R211945 treatment compared with atorvastatin treatment and a control. METHODS: Twenty-one atherosclerotic New Zealand white rabbits were divided into 3 groups (control, R211945 [3 mg/kg orally], and atorvastatin [3 mg/kg orally] groups). All groups underwent (18)F-FDG-PET/CT and DCE-CMR at baseline and at 1 and 3 months after treatment initiation. Concomitantly, serum metabolic parameters and histology were assessed. For statistical analysis, continuous DCE-CMR and PET/CT outcomes were modeled as linear functions of time by using a linear mixed model, whereas the histological data, animal characteristics data, and nonlinear regression imaging data were analyzed with a 2-tailed Student t test. RESULTS: (18)F-FDG-PET/CT detected a decrease in mean and maximum standard uptake values (SUV) over time in the R211945 group (both p = 0.001), indicating inflammation regression. The atorvastatin group displayed no significant change (p = 0.371 and p = 0.600, respectively), indicating no progression or regression. The control group demonstrated an increase in SUV (p = 0.01 and p = 0.04, respectively), indicating progression. There was a significant interaction between time and group for mean and maximum SUV (p = 0.0003 and p = 0.0016, respectively) . DCE-CMR detected a trend toward difference (p = 0.06) in the area under the curve in the atorvastatin group, suggesting a decrease in neovascularization. There was no significant interaction between time and group (p = 0.6350 and p = 0.8011, respectively). Macrophage and apolipoprotein B immunoreactivity decreased in the R211945 and atorvastatin groups (p < 0.0001 and p = 0.0004, respectively), and R211945 decreased oxidized phospholipid immunoreactivity (p = 0.02). CONCLUSIONS: Noninvasive imaging with (18)F-FDG-PET/CT and DCE-CMR and histological analysis demonstrated significant anti-inflammatory effects of the LXR agonist R211945 compared with atorvastatin. The results suggest a possible role for LXR agonists in the treatment of atherosclerosis.

Pioglitazone modulates vascular inflammation in atherosclerotic rabbits noninvasive assessment with FDG-PET-CT and dynamic contrast-enhanced MR imaging.

OBJECTIVES: We sought to determine the antiatherosclerotic properties of pioglitazone using multimethod noninvasive imaging techniques. BACKGROUND: Inflammation is an essential component of vulnerable or high-risk atheromas. Pioglitazone, a peroxisome proliferator-activated receptor-gamma agonist, possesses potent anti-inflammatory properties. We aimed to quantify noninvasively the anti-inflammatory effects of pioglitazone on atheroma using (18)F-fluorodeoxyglucose ((18)F-FDG) positron emission tomography (PET)/computed tomography (CT) and dynamic contrast-enhanced (DCE) magnetic resonance imaging (MRI). METHODS: Atherosclerotic plaques were induced in the aorta of 15 New Zealand white rabbits by a combination of a hyperlipidemic diet and 2 balloon endothelial denudations. Nine rabbits continued the same diet, whereas 6 rabbits received pioglitazone (10 mg/kg orally) in addition to the diet. Twelve animals underwent (18)F-FDG-PET/CT, and 15 animals underwent DCE-MRI at baseline, 1 month, and 3 months after treatment initiation. Concomitantly, serum metabolic parameters were monitored. After imaging was completed, aortic histologic analysis and correlation analysis were performed. RESULTS: The (18)F-FDG-PET/CT imaging detected an increase in average standardized uptake value in the control group (p < 0.01), indicating progressive inflammation, whereas stable standardized uptake values were observed in the treatment group, indicating no progression. The DCE-MRI analysis detected a significant decrease in the area under the curve for the pioglitazone group (p < 0.01). Immunohistologic examination of the aortas demonstrated a significant decrease in macrophage and oxidized phospholipid immunoreactivity in the pioglitazone group (p = 0.04 and p = 0.01, respectively) with respect to control animals, underlining the imaging results. Serum metabolic parameters showed no difference between groups. Strong positive correlations between standardized uptake value and macrophage density and between area under the curve and neovessels were detected (r(2) = 0.86 and p < 0.0001, and r(2) = 0.66 and p = 0.004, respectively). CONCLUSIONS: Both (18)F-FDG-PET/CT and DCE-MRI demonstrate noninvasively the anti-inflammatory effects of pioglitazone on atheroma. Both imaging methods seem suited to monitor inflammation in atherosclerosis.

Cross-sectional, prospective study of MRI reproducibility in the assessment of plaque burden of the carotid arteries and aorta.

BACKGROUND: The reliability of imaging techniques to assess early atherosclerosis remains unclear. We did a cross-sectional, prospective study to test reproducibility of MRI when imaging arteries, to assess risk of cardiovascular disease and correlations with age and sex. METHODS: Between January 2003 and December 2006 we performed black-blood MRI of both common carotid arteries and the thoracic descending aorta in patients with cardiovascular risk factors who were referred from clinics in New York, NY, USA. Mean wall area, wall thickness, lumen area, total vessel area, and ratio of the mean wall area to the mean total vessel area (WA/TVA) were manually measured. Reproducibility within and between readers was tested on subsets of images from randomly chosen patients. RESULTS: MRI was performed on 300 patients. Intrareader reproducibility, assessed in images from 20 patients, was high for all parameters (intraclass correlation coefficients >0.8), except WA/TVA ratio in the descending aorta. The inter-reader reproducibility, assessed in images from 187 patients, was acceptable (intraclass correlation coefficients >0.7) for the mean wall, lumen, and total vessel areas. Values for all MRI parameters in all vessels increased with increasing age for both sexes (all P <0.0005) but were always significantly higher in men than in women, except for aortic mean wall thickness and WA/TVA ratio in the carotid arteries. Mean wall area values correlated well between the carotid arteries and aorta, reflecting the systemic nature of atherosclerosis. CONCLUSIONS: Our findings support MRI as a reproducible measurement of plaque burden and demonstrate the systemic distribution of atherosclerosis.

Plaque imaging and characterization using magnetic resonance imaging: towards molecular assessment.

Identification of high-risk atherosclerotic lesions prone to rupture and thrombosis may greatly decrease the morbidity and mortality associated with atherosclerosis. High-resolution magnetic resonance imaging (MRI) has recently emerged as one of the most promising techniques for the non-invasive study of atherothrombotic disease, as it can characterize plaque composition and monitor its progression. The development of MRI contrast agents that specifically target components of the atherosclerotic plaque may enable non-invasive detection of high-risk lesions. This review discusses the use of high-resolution MRI for plaque detection and characterization and the potentials of "Molecular Imaging" using a variety of molecules present in atherosclerotic plaques that may serve as targets for specific contrast agents to allow the identification of high-risk atherosclerotic lesions in-vivo. Ultimately, such agents may enable treatment of "high-risk" patients prior to lesion progression and occurrence of complications.