Since its emergence, the highly virulent coronavirus disease 2019 (COVID-19) has infected over 50 million people worldwide, and its cardiovascular (CV) manifestations have become an important cause of morbidity and mortality. Patients with pre-existing diseases, particularly diabetes and obesity, are more likely to experience severe manifestation of the infection.(1-3) The prevalence of CV complications is also higher in patients with immunosuppression, chronic obstructive pulmonary disease (COPD), asthma, advanced age, male sex, hypertension, and renal insufficiency.(4,5) The high prevalence of myocardial injury in patients with COVID-19 highlights the need for better tools to prognosticate and guide management – such tools can either be clinical or laboratory-based, imaging techniques or a
combination of these. Artificial intelligence (AI) and machine learning (ML) can have significant utility in the application of these tools to better diagnose CV involvement, provide risk scoring systems, and to select appropriate therapy.
Overview of AI and ML
AI is a general term that signifies the use of mathematical algorithms which give machines the ability to perform complex tasks and make predictions without having to make as many assumptions about the underlying data compared to traditional statistical models. ML is a category of AI in which a machine learns to perform a task or make decisions from available data without being explicitly programmed. ML methods can complement and expand upon conventional statistical methods. Statistical methods primarily allow one to explore relationships between a limited number of variables and require that several strong criteria be met. In contrast, ML methods can identify features in the data, make predictions, and provide algorithms to understand broader patterns from large, heterogeneous datasets without such binding or strong limitations. (6,7) Although this approach does not involve traditional statistical inference, it can provide superior and more robust algorithms that more accurately predict and classify disease progression. The two major techniques and approaches in AI and ML include supervised and unsupervised learning. In supervised learning, a general algorithm, such as a neural network, is used to approximate a complex mathematical relationship between the input data and expected outputs and classifies an
observation into one or more categories or outcomes. In unsupervised learning, we feed only the inputs and the machine determines an algorithm that it uses to subsequently identify hidden patterns or structures within the dataset.(8)
Potential Cardiovascular applications of AI During the COVID-19 Pandemic
Because AI systems are capable of handling massive amounts of data at a time, they can be scaled to make mass diagnoses and predictions matching the exponential pandemic curve. Although pulmonary manifestations predominate in most
patients with COVID-19, involvement of other organs is common, with acute cardiac injury occurring in up to 28% of patients.(9) CV involvement can manifest with a variety of clinical presentations, even amongst individuals with similar risk factors. Additionally, these cardiac complications can occur precipitously at any point during hospitalization, and even have been reported as late complications that occur after improvement in a patient’s respiratory status.(10) This poses a significant challenge when it comes to identifying CV complications as they may be obscured by the pulmonary picture or distributive shock that may accompany the acute illness in COVID-19. AI applications for detecting CV complications with inputs from the heart rate, blood pressure, oxygen saturation levels, temperature curves, electrocardiographic monitoring, and imaging techniques can make computeraided assessments and diagnoses faster because AI can be scaled exponentially.
A few ongoing clinical trials investigating COVID-19 and its effects on the CV system using AI are listed in Table 1. There are a number of studies evaluating AI tools to generate predictive models that can assess COVID-19 related CV complications with
echocardiography.(11,12) One study at Johns Hopkins University is collecting data from ~300 COVID-19 patients to train an ML algorithm that can predict cardiac events in advance and risk stratify patients on initial presentation.(12) The inclusion of AI and ML models in echocardiography is very promising, as they can accurately identify various echocardiographic parameters rapidly, and predict outcomes, without the limitation of inter-operator variability and experience. Several other studies are ongoing and are expected to yield promising leads for further discovery and validation – these have been modeled on prior experience in stable CVD patients, especially with noninvasive imaging. The scale of the pandemic allows us to apply the lessons and algorithms learned from these prior datasets to see if meaningful new insights can be gleaned in both hospitalized and discharged COVID-19 patients.
Conclusion
The current COVID-19 pandemic poses a major and imminent challenge for healthcare systems, and there is increasing awareness of its CV manifestations and the adverse impact that this has on prognosis. Early and reliable personalized risk prediction represents a major unmet clinical need and may allow for better clinical decision making and more effective resource allocation. ML could potentially help clinicians make a faster and more accurate diagnosis, better risk-stratify patients and improve their outcomes. ML is expected to yield profound new insights into patterns that have not been evident based on heuristic human intelligence from application to prior cardiovascular pathologies, but whether it will meet the hype will await properly designed trials.
Disclosure statement:
Dr. Zimmerman has no financial disclosures to report. Dr. Jahan has no financial disclosures to report. Dr. Kalra has no financial disclosures to report.
References:
1. Shi S, Qin M, Shen B, et al. Association of Cardiac Injury with Mortality in Hospitalized Patients with COVID-19 in Wuhan, China. JAMA Cardiol. 2020:1-8. doi:10.1001/jamacardio.2020.0950
2. Wang D, Hu B, Hu C, et al. Clinical Characteristics of 138 Hospitalized Patients with 2019 Novel Coronavirus-Infected Pneumonia in Wuhan, China. JAMA - J Am Med Assoc. 2020;323(11):1061-1069. doi:10.1001/jama.2020.1585
3. Huang C, Wang Y, Li X, et al. Clinical features of patients infected with 2019 novel coronavirus in Wuhan, China. Lancet. 2020;395(10223):497-506. doi:10.1016/S0140-6736(20)30183-5
4. Zheng YY, Ma YT, Zhang JY, Xie X. COVID-19 and the cardiovascular system. Nat Rev Cardiol. 2020;17(5):259-260. doi:10.1038/s41569-020-0360-5
5. Mossa-Basha M, Meltzer CC, Kim DC, Tuite MJ, Kolli KP, Tan BS. Radiology Department Preparedness for COVID-19: Radiology Scientific Expert Review Panel. Radiology. 2020;296(2):E106-E112. doi:10.1148/radiol.2020200988
6. Al’Aref SJ, Anchouche K, Singh G, et al. Clinical applications of machine learning in cardiovascular disease and its relevance to cardiac imaging. Eur Heart J. 2019;40(24):1975-1986. doi:10.1093/eurheartj/ehy404
7. Johnson KW, Torres Soto J, Glicksberg BS, et al. Artificial Intelligence in Cardiology. J Am Coll Cardiol. 2018;71(23):2668-2679. doi:10.1016/j.jacc.2018.03.521
8. Lopez-Jimenez F, Attia Z, Arruda-Olson AM, et al. Artificial Intelligence in Cardiology: Present and Future. Mayo Clin Proc. 2020;95(5):1015-1039. doi:10.1016/j.mayocp.2020.01.038
9. Atri D, Siddiqi HK, Lang JP, Nauffal V, Morrow DA, Bohula EA. COVID-19 for the Cardiologist: Basic Virology, Epidemiology, Cardiac Manifestations, and Potential Therapeutic Strategies. JACC Basic to Transl Sci. 2020;5(5):518-536. doi:10.1016/j. jacbts.2020.04.002
10. Fried JA, Ramasubbu K, Bhatt R, et al. The variety of cardiovascular presentations of COVID-19. Circulation. 2020:1930-1936. doi:10.1161/CIRCULATIONAHA.120.047164
11. Mayo Clinic Uses Artificial Intelligence to Help Assess Cardiac Danger From COVID-19. Diagnostic and Interventional Cardiology. https://www.dicardiology.com/content/mayo-clinic-uses-artificialintelligence-help-assess-cardiac-danger-covid-19. Published 2020.
12. Stempniak M. Machine learning uses imaging to predict heart damage in COVID-19 patients. Radiology Business News. 2020.
13. Claude Deharo J. QT-Logs : a Clinical Study to Monitor Cardiac Safety, With Artificial Intelligence for QT Interval Analysis of ECG Data From Smartwatches, in Patients Receiving Hydroxychloroquine Treatment for COVID-19. Clin Trials. 2020.
14. Italiano IA. Predictors of Cardiovascular Risk in Covid-19 Patients During Acute Disease and at Short Term Follow-up. ClinicalTrials. https://clinicaltrials.gov/ct2/show/record/NCT04371289term=cardiovascular&cond=artificial+intelligence+and+COVID19&draw=2&rank=1. Published 2020.
15. Twerenbold R, Kuster Pfister G. COVID-19 Survival - The COVIVA Study (COVIVA). Clinical Trials. https://clinicaltrials.gov/ct2/show/
NCT04366765?term=CARDIOLOGY&cond=ARTIFICIAL+INTELLIGENCE&draw=2&rank=5. Published 2020.
16. Shah J, Ako E, Al-Hussaini A, Ng FS, Patel K, ]. Artificial Intelligenceassisted Diagnosis and Prognostication in COVID-19 Using Electrocardiograms (AI-COV-19). Clinical Trials. https://clinicaltrials.gov/ct2/show/study/NCT04510441term=artificial+intelligence&cond=Covid19&draw=2&rank=5. Published 2020.
17. Meder B. CoronaWatch - Early Detection of Cardiovascular Risks i COVID-19 Via SmartWatch. Clinical Trials.