Receive a weekly summary and discussion of the top papers of the week by leading researchers in the field.

Radiology Radiology

AI Cardiac MRI Scar Analysis Aids Prediction of Major Arrhythmic Events in the Multicenter DERIVATE Registry.

In Radiology ; h5-index 91.0

Background Scar burden with late gadolinium enhancement (LGE) cardiac MRI (CMR) predicts arrhythmic events in patients with postinfarction in single-center studies. However, LGE analysis requires experienced human observers, is time consuming, and introduces variability. Purpose To test whether postinfarct scar with LGE CMR can be quantified fully automatically by machines and to compare the ability of LGE CMR scar analyzed by humans and machines to predict arrhythmic events. Materials and Methods This study is a retrospective analysis of the multicenter, multivendor CarDiac MagnEtic Resonance for Primary Prevention Implantable CardioVerter DebrillAtor ThErapy (DERIVATE) registry. Patients with chronic heart failure, echocardiographic left ventricular ejection fraction (LVEF) of less than 50%, and LGE CMR were recruited (from January 2015 through December 2020). In the current study, only patients with ischemic cardiomyopathy were included. Quantification of total, dense, and nondense scars was carried out by two experienced readers or a Ternaus network, trained and tested with LGE images of 515 and 246 patients, respectively. Univariable and multivariable Cox analyses were used to assess patient and cardiac characteristics associated with a major adverse cardiac event (MACE). Area under the receiver operating characteristic curve (AUC) was used to compare model performances. Results In 761 patients (mean age, 65 years ± 11, 671 men), 83 MACEs occurred. With use of the testing group, univariable Cox-analysis found New York Heart Association class, left ventricle volume and/or function parameters (by echocardiography or CMR), guideline criterion (LVEF of ≤35% and New York Heart Association class II or III), and LGE scar analyzed by humans or the machine-learning algorithm as predictors of MACE. Machine-based dense or total scar conferred incremental value over the guideline criterion for the association with MACE (AUC: 0.68 vs 0.63, P = .02 and AUC: 0.67 vs 0.63, P = .01, respectively). Modeling with competing risks yielded for dense and total scar (AUC: 0.67 vs 0.61, P = .01 and AUC: 0.66 vs 0.61, P = .005, respectively). Conclusion In this analysis of the multicenter CarDiac MagnEtic Resonance for Primary Prevention Implantable CardioVerter DebrillAtor ThErapy (DERIVATE) registry, fully automatic machine learning-based late gadolinium enhancement analysis reliably quantifies myocardial scar mass and improves the current prediction model that uses guideline-based risk criteria for implantable cardioverter defibrillator implantation. ClinicalTrials.gov registration no.: NCT03352648 Published under a CC BY 4.0 license. Supplemental material is available for this article.

Ghanbari Fahime, Joyce Thomas, Lorenzoni Valentina, Guaricci Andrea I, Pavon Anna-Giulia, Fusini Laura, Andreini Daniele, Rabbat Mark G, Aquaro Giovanni Donato, Abete Raffaele, Bogaert Jan, Camastra Giovanni, Carigi Samuela, Carrabba Nazario, Casavecchia Grazia, Censi Stefano, Cicala Gloria, De Cecco Carlo N, De Lazzari Manuel, Di Giovine Gabriella, Di Roma Mauro, Focardi Marta, Gaibazzi Nicola, Gismondi Annalaura, Gravina Matteo, Lanzillo Chiara, Lombardi Massimo, Lozano-Torres Jordi, Masi Ambra, Moro Claudio, Muscogiuri Giuseppe, Nese Alberto, Pradella Silvia, Sbarbati Stefano, Schoepf U Joseph, Valentini Adele, Crelier Gérard, Masci Pier Giorgio, Pontone Gianluca, Kozerke Sebastian, Schwitter Juerg

2023-Mar-21

General General

Paradoxical decrease of imitation performance with age in children.

In British journal of psychology (London, England : 1953)

Imitation development was studied in a cross-sectional design involving 174 primary-school children (aged 6-10), focusing on the effect of actions' complexity and error analysis to infer the underlying cognitive processes. Participants had to imitate the model's actions as if they were in front of a mirror ('specularly'). Complexity varied across three levels: movements of a single limb; arm and leg of the same body side; or arm and leg of opposite body sides. While the overall error rate decreased with age, this was not true of all error categories. The rate of 'side' errors (using a limb of the wrong body side) paradoxically increased with age (from 9 years). However, with increasing age, the error rate also became less sensitive to the complexity of the action. This pattern is consistent with the hypothesis that older children have the working memory (WM) resources and the body knowledge necessary to imitate 'anatomically', which leads to additional side errors. Younger children might be paradoxically free from such interference because their WM and/or body knowledge are insufficient for anatomical imitation. Yet, their limited WM resources would prevent them from successfully managing the conflict between spatial codes involved in complex actions (e.g. moving the left arm and the right leg). We also found evidence that action side and content might be stored in separate short-term memory (STM) systems: increasing the number of sides to be encoded only affected side retrieval, but not content retrieval; symmetrically, increasing the content (number of movements) of the action only affected content retrieval, but not side retrieval. In conclusion, results suggest that anatomical imitation might interfere with specular imitation at age 9 and that STM storages for side and content of actions are separate.

Ottoboni Giovanni, Toraldo Alessio, Proietti Riccardo, Cangelosi Angelo, Tessari Alessia

2023-Mar-21

body knowledge, children, development, double dissociation, imitation, meaningless action, movement complexity, working memory

Internal Medicine Internal Medicine

Machine learning to predict major bleeding during anticoagulation for venous thromboembolism: possibilities and limitations.

In British journal of haematology ; h5-index 64.0

Predictive tools for major bleeding (MB) using machine learning (ML) might be advantageous over traditional methods. We used data from the Registro Informatizado de Enfermedad TromboEmbólica (RIETE) to develop ML algorithms to identify patients with venous thromboembolism (VTE) at increased risk of MB during the first 3 months of anticoagulation. A total of 55 baseline variables were used as predictors. New data prospectively collected from the RIETE were used for further validation. The RIETE and VTE-BLEED scores were used for comparisons. External validation was performed with the COMMAND-VTE database. Learning was carried out with data from 49 587 patients, of whom 873 (1.8%) had MB. The best performing ML method was XGBoost. In the prospective validation cohort the sensitivity, specificity, positive predictive value and F1 score were: 33.2%, 93%, 10%, and 15.4% respectively. F1 value for the RIETE and VTE-BLEED scores were 8.6% and 6.4% respectively. In the external validation cohort the metrics were 10.3%, 87.6%, 3.5% and 5.2% respectively. In that cohort, the F1 value for the RIETE score was 17.3% and for the VTE-BLEED score 9.75%. The performance of the XGBoost algorithm was better than that from the RIETE and VTE-BLEED scores only in the prospective validation cohort, but not in the external validation cohort.

Mora Damián, Mateo Jorge, Nieto José A, Bikdeli Behnood, Yamashita Yugo, Barco Stefano, Jimenez David, Demelo-Rodriguez Pablo, Rosa Vladimir, Yoo Hugo Hyung Bok, Sadeghipour Parham, Monreal Manuel

2023-Mar-21

haemorrhage, machine learning, outcomes, pulmonary embolism, venous thrombosis

Cardiology Cardiology

Convolution Neural Network Algorithm for Shockable Arrhythmia Classification Within a Digitally Connected Automated External Defibrillator.

In Journal of the American Heart Association ; h5-index 70.0

Background Diagnosis of shockable rhythms leading to defibrillation remains integral to improving out-of-hospital cardiac arrest outcomes. New machine learning techniques have emerged to diagnose arrhythmias on ECGs. In out-of-hospital cardiac arrest, an algorithm within an automated external defibrillator is the major determinant to deliver defibrillation. This study developed and validated the performance of a convolution neural network (CNN) to diagnose shockable arrhythmias within a novel, miniaturized automated external defibrillator. Methods and Results There were 26 464 single-lead ECGs that comprised the study data set. ECGs of 7-s duration were retrospectively adjudicated by 3 physician readers (N=18 total readers). After exclusions (N=1582), ECGs were divided into training (N=23 156), validation (N=721), and test data sets (N=1005). CNN performance to diagnose shockable and nonshockable rhythms was reported with area under the receiver operating characteristic curve analysis, F1, and sensitivity and specificity calculations. The duration for the CNN to output was reported with the algorithm running within the automated external defibrillator. Internal and external validation analyses included CNN performance among arrhythmias, often mistaken for shockable rhythms, and performance among ECGs modified with noise to mimic artifacts. The CNN algorithm achieved an area under the receiver operating characteristic curve of 0.995 (95% CI, 0.990-1.0), sensitivity of 98%, and specificity of 100% to diagnose shockable rhythms. The F1 scores were 0.990 and 0.995 for shockable and nonshockable rhythms, respectively. After input of a 7-s ECG, the CNN generated an output in 383±29 ms (total time of 7.383 s). The CNN outperformed adjudicators in classifying atrial arrhythmias as nonshockable (specificity of 99.3%-98.1%) and was robust against noise artifacts (area under the receiver operating characteristic curve range, 0.871-0.999). Conclusions We demonstrate high diagnostic performance of a CNN algorithm for shockable and nonshockable rhythm arrhythmia classifications within a digitally connected automated external defibrillator. Registration URL: https://clinicaltrials.gov/ct2/show/NCT03662802; Unique identifier: NCT03662802.

Shen Christine P, Freed Benjamin C, Walter David P, Perry James C, Barakat Amr F, Elashery Ahmad Ramy A, Shah Kevin S, Kutty Shelby, McGillion Michael, Ng Fu Siong, Khedraki Rola, Nayak Keshav R, Rogers John D, Bhavnani Sanjeev P

2023-Mar-21

ECG, automated external defibrillator, convolution neural network, machine learning, ventricular arrhythmias

Surgery Surgery

Artificial Intelligence-enabled Decision Support in Surgery: State-of-the-art and Future Directions.

In Annals of surgery ; h5-index 104.0

OBJECTIVE : To summarize state-of-the-art artificial intelligence-enabled decision support in surgery and to quantify deficiencies in scientific rigor and reporting.

SUMMARY BACKGROUND DATA : To positively affect surgical care, decision-support models must exceed current reporting guideline requirements by performing external and real-time validation, enrolling adequate sample sizes, reporting model precision, assessing performance across vulnerable populations, and achieving clinical implementation; the degree to which published models meet these criteria is unknown.

METHODS : Embase, PubMed, and MEDLINE databases were searched from their inception to September 21, 2022 for articles describing artificial intelligence-enabled decision support in surgery that uses preoperative or intraoperative data elements to predict complications within 90 days of surgery. Scientific rigor and reporting criteria were assessed and reported according to PRISMA-ScR guidelines.

RESULTS : Sample size ranged from 163-2,882,526, with 8/36 articles (22.2%) featuring sample sizes of less than 2,000; seven of these eight articles (87.5%) had below-average (<0.83) area under the receiver operating characteristic (AUROC) or accuracy. Overall, 29 articles (80.6%) performed internal validation only, five (13.8%) performed external validation, and two (5.6%) performed real-time validation. Twenty-three articles (63.9%) reported precision. No articles reported performance across sociodemographic categories. Thirteen articles (36.1%) presented a framework that could be used for clinical implementation; none assessed clinical implementation efficacy.

CONCLUSIONS : Artificial intelligence-enabled decision support in surgery is limited by reliance on internal validation, small sample sizes that risk overfitting and sacrifice predictive performance, and failure to report confidence intervals, precision, equity analyses, and clinical implementation. Researchers should strive to improve scientific quality.

Loftus Tyler J, Altieri Maria S, Balch Jeremy A, Abbott Kenneth L, Choi Jeff, Marwaha Jayson S, Hashimoto Daniel A, Brat Gabriel A, Raftopoulos Yannis, Evans Heather L, Jackson Gretchen P, Walsh Danielle S, Tignanelli Christopher J

2023-Mar-21

General General

Gender Differences in the Nonspecific and Health-Specific Use of Social Media Before and During the COVID-19 Pandemic: Trend Analysis Using HINTS 2017-2020 Data.

In Journal of health communication ; h5-index 36.0

The use of social media has changed since the outbreak of coronavirus disease 2019 (COVID-19). However, little is known about the gender disparity in social media use for nonspecific and health-specific issues before and during the COVID-19 pandemic. Based on a gender difference perspective, this study aimed to examine how the nonspecific and health-specific uses of social media changed in 2017-2020. The data came from the Health Information National Trends Survey Wave 5 Cycle 1-4. This study included 10,426 participants with complete data. Compared to 2017, there were higher levels of general use in 2019 and 2020, and an increased likelihood of health-related use in 2020 was reported among the general population. Female participants were more likely to be nonspecific and health-specific users than males. Moreover, the relationship of gender with general use increased in 2019 and 2020; however, concerning health-related use, it expanded in 2019 but narrowed in 2020. The COVID-19 global pandemic led to increased use of social media, especially for health-related issues among males. These findings further our understanding of the gender gap in health communication through social media, and contribute to targeted messaging to promote health and reduce disparities between different groups during the pandemic.

Ye Linglong, Chen Yang, Cai Yongming, Kao Yi-Wei, Wang Yuanxin, Chen Mingchih, Shia Ben-Chang, Qin Lei

2023-Mar-21