A study of ChatGPT in facilitating Heart Team decisions on severe aortic stenosis.

BACKGROUND: Multidisciplinary Heart Teams (HTs) play a central role in the management of valvular heart diseases. However, the comprehensive evaluation of patients' data can be hindered by logistical challenges, which in turn may affect the care they receive. AIMS: This study aimed to explore the ability of artificial intelligence (AI), particularly large language models (LLMs), to improve clinical decision-making and enhance the efficiency of HTs. METHODS: Data from patients with severe aortic stenosis presented at HT meetings were retrospectively analysed. A standardised multiple-choice questionnaire, with 14 key variables, was processed by the OpenAI Chat Generative Pre-trained Transformer (GPT)-4. AI-generated decisions were then compared to those made by the HT. RESULTS: This study included 150 patients, with ChatGPT agreeing with the HT's decisions 77% of the time. The agreement rate varied depending on treatment modality: 90% for transcatheter valve implantation, 65% for surgical valve replacement, and 65% for medical treatment. CONCLUSIONS: The use of LLMs offers promising opportunities to improve the HT decision-making process. This study showed that ChatGPT's decisions were consistent with those of the HT in a large proportion of cases. This technology could serve as a failsafe, highlighting potential areas of discrepancy when its decisions diverge from those of the HT. Further research is necessary to solidify our understanding of how AI can be integrated to enhance the decision-making processes of HTs.

ChatGPT takes on the European Exam in Core Cardiology: an artificial intelligence success story?

Chat Generative Pre-trained Transformer (ChatGPT) is currently a trending topic worldwide triggering extensive debate about its predictive power, its potential uses, and its wider implications. Recent publications have demonstrated that ChatGPT can correctly answer questions from undergraduate exams such as the United States Medical Licensing Examination. We challenged it to answer questions from a more demanding, post-graduate exam-the European Exam in Core Cardiology (EECC), the final exam for the completion of specialty training in Cardiology in many countries. Our results demonstrate that ChatGPT succeeds in the EECC.

[Can smart devices and AI in cardiology improve clinical practice?] / Les dispositifs intelligents et l'IA en cardiologie peuvent-ils améliorer la pratique clinique ?

Remote monitoring is becoming increasingly popular among healthcare professionals and patients for diagnosing and treating heart disease. Several smart devices connected to smartphones have been developed and validated in recent years, but their clinical use is still limited. Significant advances in the field of artificial intelligence (AI) are also revolutionizing several fields, yet the impact that these innovations could have on routine clinical practice is still unknown. We review the evidence and uses of the main smart devices currently available as well as the latest applications of AI in the field of cardiology, with the aim to ultimately evaluate the potential of this technology to transform modern clinical practice.

Le monitorage à distance devient de plus en plus populaire parmi les praticiens de la santé pour le diagnostic et la surveillance des maladies cardiaques. Plusieurs dispositifs « intelligents ¼ et connectés aux smartphones ont été développés et validés durant ces dernières années, mais leur utilisation clinique est toujours limitée. Bien que les progrès de l'intelligence artificielle (IA) soient en train de révolutionner plusieurs domaines, l'impact que ces innovations pourront avoir dans le monde médical est toujours inconnu. Le but de cet article est de passer en revue les principaux dispositifs disponibles et de comprendre les applications actuelles de l'IA en cardiologie, afin de mieux saisir dans quelle mesure ils sont susceptibles de transformer notre pratique clinique quotidienne.

Deep learning-based prediction of future myocardial infarction using invasive coronary angiography: a feasibility study.

BACKGROUND: Angiographic parameters can facilitate the risk stratification of coronary lesions but remain insufficient in the prediction of future myocardial infarction (MI). AIMS: We compared the ability of humans, angiographic parameters and deep learning (DL) to predict the lesion that would be responsible for a future MI in a population of patients with non-significant CAD at baseline. METHODS: We retrospectively included patients who underwent invasive coronary angiography (ICA) for MI, in whom a previous angiogram had been performed within 5 years. The ability of human visual assessment, diameter stenosis, area stenosis, quantitative flow ratio (QFR) and DL to predict the future culprit lesion (FCL) was compared. RESULTS: In total, 746 cropped ICA images of FCL and non-culprit lesions (NCL) were analysed. Predictive models for each modality were developed in a training set before validation in a test set. DL exhibited the best predictive performance with an area under the curve of 0.81, compared with diameter stenosis (0.62, p=0.04), area stenosis (0.58, p=0.05) and QFR (0.67, p=0.13). DL exhibited a significant net reclassification improvement (NRI) compared with area stenosis (0.75, p=0.03) and QFR (0.95, p=0.01), and a positive nonsignificant NRI when compared with diameter stenosis. Among all models, DL demonstrated the highest accuracy (0.78) followed by QFR (0.70) and area stenosis (0.68). Predictions based on human visual assessment and diameter stenosis had the lowest accuracy (0.58). CONCLUSION: In this feasibility study, DL outperformed human visual assessment and established angiographic parameters in the prediction of FCLs. Larger studies are now required to confirm this finding.

ENTRYWISE EIGENVECTOR ANALYSIS OF RANDOM MATRICES WITH LOW EXPECTED RANK.

Recovering low-rank structures via eigenvector perturbation analysis is a common problem in statistical machine learning, such as in factor analysis, community detection, ranking, matrix completion, among others. While a large variety of bounds are available for average errors between empirical and population statistics of eigenvectors, few results are tight for entrywise analyses, which are critical for a number of problems such as community detection. This paper investigates entrywise behaviors of eigenvectors for a large class of random matrices whose expectations are low-rank, which helps settle the conjecture in Abbe et al. (2014b) that the spectral algorithm achieves exact recovery in the stochastic block model without any trimming or cleaning steps. The key is a first-order approximation of eigenvectors under the ℓ ∞ norm: u k ≈ A u k * λ k * , where {u k } and { u k * } are eigenvectors of a random matrix A and its expectation E A , respectively. The fact that the approximation is both tight and linear in A facilitates sharp comparisons between u k and u k * . In particular, it allows for comparing the signs of u k and u k * even if ‖ u k - u k * ‖ ∞ is large. The results are further extended to perturbations of eigenspaces, yielding new ℓ ∞-type bounds for synchronization ( ℤ 2 -spiked Wigner model) and noisy matrix completion.

Sample Complexity of the Boolean Multireference Alignment Problem.

The Boolean multireference alignment problem consists in recovering a Boolean signal from multiple shifted and noisy observations. In this paper we obtain an expression for the error exponent of the maximum A posteriori decoder. This expression is used to characterize the number of measurements needed for signal recovery in the low SNR regime, in terms of higher order autocorrelations of the signal. The characterization is explicit for various signal dimensions, such as prime and even dimensions.