Year: 2020
Type: article
Abstract: Advances in digital health and particularly in artificial intelligence (AI) have led us very close to the true implementation of personalized medicine. The year 2020 has brought an exponential increase of studies using various forms of AI, from supervised machine learning to unsupervised deep learning, with applications across all domains of cardiovascular medicine. AI is now moving from research to implementation, affecting all aspects of clinical cardiology. The studies bringing AI close to clinical practice span from fast clinical and biochemical data analysis and interpretation of results to image analysis, electrocardiogram (ECG) interpretation, arrhythmia detection, or even the use of face recognition to diagnose cardiovascular diseases. We review some of the most exciting development if the field of AI in cardiology, published from fall of 2019 up until now. The studies highlighted in this article give only a small glimpse into this booming field, creating more anticipation for what will come to clinical practice in the coming years. Digital health and particularly artificial intelligence (AI) are getting fast ground during the last few years in cardiovascular diagnostics and therapeutics. Indeed, the number of publications using various AI techniques has been increased by >20-fold from 2010 to 2020. Since last year’s European Society of Cardiology (ESC) congress, the role of AI in cardiovascular medicine had been highlighted as the next frontier in cardiovascular diagnostics, paving the way to the implementation of personalized strategies in cardiovascular therapeutics.1 In a similar line, the last American Heart Association 2020 meeting also had a session entitled ‘Hype or Hope? Artificial Intelligence and Machine Learning in Imaging’, reminding us the importance paid by the major clinical cardiovascular medicine societies in this field. Indeed, issues like algorithm transparency and data open access transparency were key issues introduced. The concept of using digital innovation and particularly AI and Big Data to optimize treatments in clinical trials and eventually in clinical practice was brought up as a fundamental aspect of digital health of the future. The introduction of AI in research but also in clinical practice is mainly driven by the technological advances in the handling and analysis of big data. AI is referred to the ability of a machine to execute tasks characteristic of human intelligence, such as problem solving or pattern recognition, and it is typically characterized by the element of positive or negative reinforcement as part of the learning process, similar to what typically happens with human learning. Indeed, machine learning refers to the ability of computers to improve their knowledge without being explicitly programmed to do so; so the machines can identify patterns in digital data and make generalizations, learning from their observations.2 Unsupervised deep learning is used to build convolutional neuronal networks (CNNs) that recognize features in digital data, not visible to the human eye. These data can be clinical information, images, ECGs or even standard ‘selfies’ taken using smartphone cameras. Management of arrhythmias has always been a challenge, especially when we have to deal with subclinical conditions such as paroxysmal atrial fibrillation, which often have stroke as their first presentation. Indeed, including clinical risk factors into a machine-learning algorithm was recently found to identify patients at risk for atrial fibrillation in a primary care population of >600k individuals in the DISCOVER registry in the UK.3 That algorithm could achieve negative predictive value of 96.7% and sensitivity to detect atrial fibrillation of 91.8%. In another landmark study published by the Mayo Clinic last year,4 it seems now possible that, by using a CNN to screen standard 12 lead ECGs for characteristics not visible to the eye of the clinician, we can detect subclinical paroxysmal atrial fibrillation from sinus rhythm ECGs, achieving AUC as high as 0.9. This study was conducted in a population of >180k individuals with >450k ECGs included in the training set, >64k ECGs in the internal validation dataset, and >130k in the testing dataset. Algorithms like this could completely transform population screening for atrial fibrillation and will most likely enable timely administration of anticoagulant treatment to prevent cardioembolic stroke. The astonishing size of this dataset gives a clear example of how deep learning should be performed, to yield reproducible, practice-changing tools. Algorithms like this will soon be available on our portable ECGs in the clinic. One of the major problems of deep learning algorithms used for ECG interpretation is their susceptibility to adversarial examples, leading to consistently wrong classification of the test by detecting false patterns undetectable to the human eye. An elegant study by Han et al.5 has provided recently the tools needed to study the impact of these adversarial patterns in automated ECG classification and provides new opportunities to develop appropriate mitigation measures. The recent release of large, publically available ECG databases such as the PTB-XL (that includes ∼21k records from ∼19k patients)6 or the one from the Shaoxing Hospital Zhejiang University School of Medicine (∼10k patients)7 brings further optimism that, by increasing the variability and ethnic diversity of the training and validation datasets, these ECG applications are not far from clinical implementation. Further to the use of AI to detect atrial fibrillation, a recent study built a deep neural network to classify various types of ECGs using >2.3 m ECGs from >1.6 m patients, demonstrating a remarkable ability of these networks to provide accurate interpretation of these tests.8 Finally, in the Apple Heart Study,9 the use of smartphones was demonstrated to be a very effective way to detect patients with subclinical paroxysmal atrial fibrillation. That was a large study that included ∼420k participants followed up for a median of 117 days through their smartphones. The technology developed by apple identified 0.5% with potentially irregular pulse (34% of which were proven to have atrial fibrillation confirmed by ECG). Although the exact nature of the technology used in the smartphone is not available, this study demonstrates that large volumes of data can be collected even using standard smartphones or portable devices like apple watches, opening new opportunities for big data research and development of AI algorithms for timely detection of cardiac arrhythmias in asymptomatic individuals. Risk stratification plays a key role in designing the therapeutic strategies in heart failure, given that the expected survival inevitably affects the decision for device implantation.10 Currently, decision of implanting a defibrillator and/or applying cardiac resynchronization therapy (CRT) in patients with heart failure relies on well-defined clinical, electrophysiological and imaging characteristics.10,11 However, a recent study published in European Heart Journal (EHJ) earlier this year12 came to remind us that the prediction of responsiveness to CRT focused on mid- or long-term outcomes should be a key driver in decision-making. Indeed, Tokodi et al.12 used machine learning to help them build a risk score for the prediction of mortality following CRT. The score used information from medical history, physical examination, medication records, ECG, echocardiographic and laboratory data commonly obtained as part of routine hospital visits of patients with heart failure, and after it was trained in 1510 patients using a random forest algorithm, it has achieved a remarkable prognostic value for all-cause mortality, with AUC in ROC analysis ranging from 0.77 (in 1-year prediction) to 0.8 (in 5-year prediction). The risk calculator is now available for use (SEMMELWEIS-CRT Score, https://arguscognitive.com/crt, Figure 1).12 As the authors mention, this score could facilitate the prompt recognition of high-risk patients, guiding deployment of the appropriate prophylactic measures.13 It could also assist the patients and the families in making advance care decisions,13 while it could assist clinicians in deciding which patients are most suitable for CRT. The 12 most important predictors of all-cause mortality as assessed by the SEMMELWEIS-CRT score. The importance of each feature was quantified by calculating the decrease in the model’s performance (area under the receiver operating characteristic curve) after permuting its values (permutation feature importance method). The higher its value, the more important the feature is. As the values of feature importance were spread over a wide range (more orders of magnitude), base-10 logarithmic transformation was performed to facilitate plotting. CRT, cardiac resynchronization therapy; LVEF, left ventricular ejection fraction; NYHA, New York Heart Failure Association functional class (from Tokodi et al.12). The 12 most important predictors of all-cause mortality as assessed by the SEMMELWEIS-CRT score. The importance of each feature was quantified by calculating the decrease in the model’s performance (area under the receiver operating characteristic curve) after permuting its values (permutation feature importance method). The higher its value, the more important the feature is. As the values of feature importance were spread over a wide range (more orders of magnitude), base-10 logarithmic transformation was performed to facilitate plotting. CRT, cardiac resynchronization therapy; LVEF, left ventricular ejection fraction; NYHA, New York Heart Failure Association functional class (from Tokodi et al.12). The results of that study were in line with another recent study showing that the use of machine learning to integrate clinical data together with imaging characteristics can provide meaningful information about the future responsiveness of heart failure patients to CRT in a population of >1.1k patients from the MADIT-CRT study.14 CRT would have meaningful impact in patient’s prognosis. Before we reach at that stage though, it seems important to understand how to manage the high-risk individuals identified through such algorithms, given that most of the factors included into these models are non-modifiable.15 Randomized clinical trials are needed, to evaluate the clinical benefit and cost-effectiveness of applying such algorithms in clinical practice. The last year brought also new advances in the use of AI for the diagnosis of heart failure. Indeed, a CNN was trained based on paired ECGs and transthoracic echocardiograms from ∼45k patients and validated in an independent cohort of >52k patients.16 The ROC for detection of systolic dysfunction using this AI-enhanced ECG interpretation reached an AUC of 0.93. This impressive result confirms the notion that AI could extract invaluable information even from simple, low-cost tests like ECG, which could even be used as screening tests for the detection of subclinical heart failure in the community. This year has been extraordinary for medical imaging, as a wide range of AI-powered algorithms has been introduced in clinical care by the hardware vendors and software manufacturers. These algorithms range from image reconstruction, to automated segmentation and improvement of workflows, or even to the detection of imaging characteristics not visible to the human eye assisting diagnosis.17,18 The year 2020 is considered by many as the year of cardiac computed tomography angiography (CTA), as this has just been incorporated into the recent ESC guidelines as a first-line investigation for the management of chest pain.19 This approach came a few years after a similar recommendation was published in the UK NICE guidelines, but it is still more advanced compared to the US standard of care.20 Given the standardized way by which computed tomography (CT) images are captured, the modality is particularly attractive to machine-learning methods to improve segmentation and interpretation. Indeed, in a study by Al’Aref et al.21 from the CONFIRM registry, a population of >13k patients undergoing coronary calcium score measurements (CCS) was used to examine whether including CCS in a machine-learning model together with clinical risk factors could improve risk stratification. Indeed, adding CCS in a baseline model that included clinical risk factors resulted in ∼9% improvement in the ability to estimate the pre-test probability of obstructive coronary artery disease, with remarkable diagnostic accuracy. Particularly in the young patients (<65 years old), the algorithm improved the ability to detect coronary artery disease by ∼17%. However, it remains to be proven that machine learning performs better than simple statistical regression models, when risk factors are combined with results from tests like CCS.22 Further to the use of machine learning to integrate imaging with other datasets, the practical value of AI lies with the improvement of image analysis workflows.18 Automated segmentation of coronary atherosclerotic plaques, coronary calcification or even epicardial fat in CT makes image interpretation faster and more accurate and eliminates user-dependent variability.23 The true power of AI though comes from its ability to ‘see the invisible’. The field of radiomics allows extraction of thousands of different pieces of information from images, which provide information on the texture and composition of the tissue visualized (Figure 2). Indeed, further to the analysis of the composition and volume of coronary atherosclerotic plaques, it is now widely accepted that vascular inflammation causes changes in the composition and texture of perivascular fat, which activates lipolysis and increases its hydrophilic content around inflamed vascular structures.24 Visualizing perivascular fat using standard CTA allows the calculation of a metric of these changes, driven by the 3D changes in perivascular fat attenuation. An AI-derived biomarker that captures that biology, the fat attenuation index (FAI), has striking prognostic value25 that goes beyond atherosclerotic plaque characteristics,26 as demonstrated in ∼4000 patients from the outcomes In a recent published by et published in the the ability of perivascular fat to its texture and composition in to coming from the vascular was in the field of The concept of has been introduced in the cardiovascular as by using the of tissue in fat obtained from patients undergoing cardiac for and all features perivascular fat after to vascular the features from the CT images of the and by using machine learning built a to detect vascular inflammation perivascular and The new metric the index (Figure was for its performance in patients from the were followed up for years after their Indeed, had a remarkable prognostic value, as with had higher risk for a or cardiac with an AUC to detect will have the of was combined with the of high risk the patient’s risk for cardiac was higher than the (Figure As it was in the this technology could therapeutic this could be in the future in the of a diagnostic to deployment of or as an for clinical published this to the of this while the strategies for risk have been in a recent of the review published in This further validation in ethnic while its into a is to the of the analysis, which makes it to on standard clinical Artificial intelligence can be used to different types of information, from clinical and laboratory data, to imaging or other of information, to assist clinical diagnosis and (from et Artificial intelligence can be used to different types of information, from clinical and laboratory data, to imaging or other of information, to assist clinical diagnosis and (from et inflammation first changes in while at a stage it to perivascular and These changes can be in standard coronary CT angiography by a using a et have built an imaging that captures these changes That has striking prognostic value over and risk factors including coronary calcium as it was validated in the coronary artery attenuation (from et inflammation first changes in while at a stage it to perivascular and These changes can be in standard coronary CT angiography by a using a et have built an imaging that captures these changes That has striking prognostic value over and risk factors including coronary calcium as it was validated in the coronary artery attenuation (from et an of 2020 has been a year for the of earlier in how to neural networks to images and from echocardiograms to and extract such as ejection and The study by et is the most in the field, driven by AI in 2020. In that analysis from still segmentation to a deep learning approach through development of a algorithm that and information the neural networks for the of segmentation and achieve an for the of ejection on a that can help in of heart failure. The next is to be the of these to to improve of disease in large databases like the UK This approach could open new in applying deep learning in image interpretation for risk Indeed, this year the of the confirmed its to innovation for years to In a study just cardiac were used in machine-learning algorithms to the detection of imaging and in the new into the of factors and on cardiac and and them also with 2020 will be as the year when brought the As our knowledge about the disease, it clear that in the a vascular The for of large volumes of data collected from around the to facilitate the development of treatments to the disease brought to the the power of AI to give fast and Indeed, fast and accurate data has been in the of the to the European like the are data around the disease, together with from the and and The use of AI to these is expected to improve our on the and pattern of cardiovascular in patients with and evaluate the and clinical of patients with cardiovascular diseases. In AI algorithms have been used to integrate chest CT with clinical laboratory testing and to diagnose In a very recent that included patients with of which positive for the AI achieved an under the of to diagnose the disease without the of a sensitivity to a The use of learning methods to integrate of inflammation and and cardiac in was recently found to mortality with AUC These models could lead to and could have major impact in clinical in the coming In the the in applying machine learning to integrate and clinical is expected to cardiac AI and convolutional neural networks are being as that features that are into algorithms that give meaningful Indeed, in a landmark study just published in the a of has developed a deep convolutional neural network that coronary artery disease by by the patient’s (Figure included patients in their training dataset and in the test dataset. The algorithm had sensitivity and to detect coronary artery disease from the of the patients, with an AUC this be demonstrating to it on the and of the face to risk factors or the disease is it just the result of training the algorithm in an population that will not the test of the concept this study is medical be at into a or through the of an is in give health problems that would like to or about health issues are not of These issues will in the coming years. in the face with information in face pattern recognition, in prediction of coronary artery In tests AUC was as the decrease in algorithm performance after a In tests of the were highlighted by the algorithm as important for detecting In the test the positive were based on the in algorithm performance after positive was with of coronary artery disease in of the while was disease in of the (from et This technology can be used for screening in the (from and in the face with information in face pattern recognition, in prediction of coronary artery In tests AUC was as the decrease in algorithm performance after a In tests of the were highlighted by the algorithm as important for detecting In the test the positive were based on the in algorithm performance after positive was with of coronary artery disease in of the while was disease in of the (from et This technology can be used for screening in the (from and Further to the opportunities by AI, these also The results by most machine-learning algorithms often to in different Since these algorithms are often in the of a it is to understand and their and this could lead to results only to or clinical included into the training deep learning algorithms are also to adversarial examples, leading to consistently wrong classification of the by detecting false patterns undetectable to the human eye. The in machine learning the of these models to comes to the of these algorithms to clinical practice. This is by applying beyond training and internal validation inevitably to of the model’s also independent the independent test dataset should the population of but in a dataset independent of the training dataset from independent and/or The training part should be used for development of the model and for can use methods like random or prevent in performance the model should be the independent The of transparency on the true the training and the independent validation dataset often makes it to evaluate the of the published can be by following while open data would independent high in in the field. is that 2020 has been an extraordinary by the these for and with most of cardiovascular research to the data The ability of AI to extract and large volumes of data this field of cardiovascular medicine to its and we have major many aspects of clinical to automated image accurate cardiovascular risk prediction or recognition to screen for cardiac AI is now major part of cardiovascular medicine. The studies highlighted in this article give only a small glimpse into this booming field, creating more anticipation for what will come to clinical practice in the coming years. The European Society of Cardiology has the importance of the fast field of digital health and has it as a of cardiovascular medicine. The European Heart Journal is at the of the to high in AI the of AI into clinical new on digital health has recently been included in the to the of in the of cardiovascular medicine. In a new Digital has been into the Finally, the European has recently an to the use of AI algorithms as medical especially for risk AI algorithms will to as medical devices from next This approach is being by and and European and will have on the clinical implementation of developed AI cardiovascular risk that will be included in the clinical guidelines in the future. of is and of an is also of CT and is of to this The authors are to for in the is by the Heart and the for and is by and the is by the is by the
Source: European Heart Journal
Language: English
Subfield: Health Informatics
Field: Medicine
Domain: Health Sciences
Open Access status: hybrid
Estimated APC paid: $5,922
Awards: Developing novel biomarkers for cardiovascular risk stratification using computed tomography imaging of perivascular adipose tissue, G2259971274, Exploring new links between adipose tissue and vascular redox state: the role of wnt5a, 11βHSD1 inhibition: a distinctive acute-MI intervention for promotion of peri-infarct vascularisation, reduction of infarct expansion and prevention of heart failure, G5218525617, Using radiomics and artificial intelligence to predict cardio-embolic stroke
OpenAlex ID: W3121078634