Acessibilidade / Reportar erro

Artificial Intelligence and Machine Learning in Cardiology - A Change of Paradigm

Keywords
Artificial Intelligence/trends; Computer Systems/trends; Machine Learning/trends; Cardiovascular Diseases; Echocardiography/trends; NucleAR Medicine/trends

A robot may not injure a human being or, through inaction,allow a human being to come to harm.

First Law of Robotics - Isaac Asimov

We are experiencing a change paradigm in modern life. With the presence of computers and intelligent machines everywhere, the predictions of science-fiction books from years ago gradually become reality; these are the times of pervasive computing. Among the computational most frequently mentioned tools in clinical studies and seen with enthusiasm by the scientific community is the Artificial Intelligence and consequently the machines that learn, which is best quoted in its original English form, Machine Learning. In general Artificial Intelligence is defined as the constellation of items (algorithms, robotics, neural networks) that allow a software to have intelligence properties that are comparable to those of a human being, among them learning from databases with minimal human interference.11 Forsting M. Hot Topics: Will Machine Learning Change Medicine? J Nucl Med. 2017;58(3):357-8.

Obermeyer and Emanuel recently wrote an editorial stating that Machine Learning has become widespread and imperative for solving complex problems in the various fields of science, and in the medical field its use will transform the practice.22 Obermeyer Z . Emanuel EJ. Predicting the future: big data, machine learning, and clinical medicine. N Engl J Med. 2016;375(13):1216-9. The use of artificial intelligence is evolving increasingly in cardiology and there are already excellent examples in several areas. Using a sophisticated learning system to electrocardiographic interpretation, Li et al. achieved that electrocardiographic patterns were automatically recognized with an accuracy of 88% for the classification of abnormal rhythms(33 Li Q, Rajagopalan C, Clifford GD. A machine learning approach to multi-level ECG signal quality classification. Comput Methods Programs Biomed . 2014;117(3):435-47.). One of the most important limitations of the system studied was the quality of the electrocardiographic signal for interpretation and learning, which highlights one of the essential characteristics of Machine Learning, that is the need for accurate and reproducible information for the formation of databases.22 Obermeyer Z . Emanuel EJ. Predicting the future: big data, machine learning, and clinical medicine. N Engl J Med. 2016;375(13):1216-9.,33 Li Q, Rajagopalan C, Clifford GD. A machine learning approach to multi-level ECG signal quality classification. Comput Methods Programs Biomed . 2014;117(3):435-47. As databases are usually produced from patients selected for their basic condition, one of the most important points for development is the creation of broader and generalizable databases that do not induce biases in the interpretation of findings, point in which industry is heavily investing at the moment.22 Obermeyer Z . Emanuel EJ. Predicting the future: big data, machine learning, and clinical medicine. N Engl J Med. 2016;375(13):1216-9.

In echocardiography, many studies are evaluating the use of Machine Learning in image interpretation, such as the Narula et al.44 Narula S, Shameer K, Salem Omar AM, Dudley JT, Sengupta PP. Machine-Learning Algorithms to Automate Morphological and Functional Assessments in 2D Echocardiography. J Am Coll Cardiol. 2016;68(21):2287-95. who, through a database of patients with hypertrophic cardiomyopathy and individuals with physiological hypertrophy who were submitted to Speckle Tracking, were able to create a computer system based on Machine Learnig that reached to assist inexperienced echocardiographers in distinguishing between the two conditions with excellent accuracy. Tajik,55 Tajik AJ. Machine Learning for Echocardiographic Imaging: Embarking on Another Incredible Journey. J Am Coll Cardiol.2016;68(21):2296-8. in an enthusiastic editorial, pointed out that Machine Learning should reduce or even eliminate the intra- and interobserver variability of echocardiographic exams and greatly reduce cognitive errors. At this point the use of artificial intelligence comes across medical ethics, because the doubts that can occur in cases of errors may be linked to the attribution of responsibilities: did the doctor fail or did the software fail? Experiences with the use of autopilots in aviation may serve as a basis for the ethical discussion that will occur, given that there is always at least one human being responsible for the flight even when using the modern devices of commercial aviation. Obermeyer and Emanuel22 Obermeyer Z . Emanuel EJ. Predicting the future: big data, machine learning, and clinical medicine. N Engl J Med. 2016;375(13):1216-9. point out that there will be a massive reduction in the need for doctors in situations where computers can be fed directly by digital information such as radiology and pathology because the large amount of digital information available will allow the formation of reliable databases that will lead to a performance of the machines superior to the human.

When artificial intelligence is employed in more complex clinical contexts, there is a still longer way to go. Austin et al.66 Austin PC, Tu J V, Ho JE, Levy D, Lee DS. Using methods from the data-mining and machine-learning literature for disease classification and prediction: a case study examining classification of heart failure subtypes. J Clin Epidemiol. 2013;66(4):398-407. used a Machine Learning and data mining system to evaluate and classify patients with heart failure and found that although the system was superior to conventional methods for predicting heart failure with preserved ejection fraction, there were no advantages over the traditional logistic regression. Liu et al.77 Liu N, Lee MAB, Ho AFW, Haaland B, Fook-Chong S, Koh ZX, et al. Risk stratification for prediction of adverse coronary events in emergency department chest pain patients with a machine learning score compared with the TIMI score. Int J Cardiol . 2014;177(3):1095-7. developed a system based on Machine Learning for prediction of adverse coronary events in patients with chest pain in the emergency room and compared it with the TIMI score. Although the performance of the new system is reliable for predicting mortality and cardiac events in 30 days, the authors themselves acknowledge that clinical decisions are dependent on factors that still can not be fully incorporated into the machines, one of which is the physicians experience.77 Liu N, Lee MAB, Ho AFW, Haaland B, Fook-Chong S, Koh ZX, et al. Risk stratification for prediction of adverse coronary events in emergency department chest pain patients with a machine learning score compared with the TIMI score. Int J Cardiol . 2014;177(3):1095-7.

In nuclear cardiology Arsanjani et al.88 Arsanjani R, Dey D, Khachatryan T, Shalev A, Hayes SW, Fish M, et al. Prediction of revascularization after myocardial perfusion SPECT by machine learning in a large population. J Nucl Cardiol. 2014;22(5):877-84. evaluated the use of the Machine Learning tool to predict myocardial revascularization from myocardial perfusion scintigraphy data, finding an accuracy comparable or even superior to that of experienced examiners in the interpretation of the scintigraphic examination. Garcia et al.,99 Garcia E V, Klein JL, Taylor AT. Clinical decision support systems in myocardial perfusion imaging. J Nucl Cardiol. 2014;21(3):427-39. in an excellent review on the subject, point out that clinical decision support and artificial intelligence systems serve as warnings for the cognitive bias of clinicians and reduce intra and interobserver variability, allowing to interpret the exams faster and with greater accuracy, as observed in studies in which the diagnostic interpretation of the examination by the computer is similar to that of the experts.1010 Arsanjani R, Xu Y, Dey D, Vahistha V, Shalev A, Nakanishi R, et al. Improved accuracy of myocardial perfusion SPECT for detection of coronary artery disease by machine learning in a large population. J Nucl Cardiol. 2013;20(4):553-62. We generally agree with this view and consider that the support to make the clinical decision and the improvement of diagnostic and prognostic performances should be encouraged and supported by physicians of the various specialties. The frequent concern about the eventual replacement of the physician by the machines is not substantiated by the facts. The medical profession is of a complexity and subjectivity that makes the task impossible in its entirety by the machines, at least at the present stage of knowledge. The proper use of computing allows for not only the improved of medical performance but also the quest for solidarity among patients with successful experiences in creating social networks for patients.1111 Medina EL, Mesquita CT, Loques Filho O. Healthcare social networks for patients with cardiovascular diseases and recommendation systems. Int J Cardiovasc Sci. 2016;29(1):80-5.

Only the study of the profound impact for the development and use of these tools can bring the answers to the questions that are now in the minds of doctors and their patients. The International Journal of Cardiovascular Sciences encourages its readers and contributors to send scientific papers on the subject for publication.

References

  • 1
    Forsting M. Hot Topics: Will Machine Learning Change Medicine? J Nucl Med. 2017;58(3):357-8.
  • 2
    Obermeyer Z . Emanuel EJ. Predicting the future: big data, machine learning, and clinical medicine. N Engl J Med. 2016;375(13):1216-9.
  • 3
    Li Q, Rajagopalan C, Clifford GD. A machine learning approach to multi-level ECG signal quality classification. Comput Methods Programs Biomed . 2014;117(3):435-47.
  • 4
    Narula S, Shameer K, Salem Omar AM, Dudley JT, Sengupta PP. Machine-Learning Algorithms to Automate Morphological and Functional Assessments in 2D Echocardiography. J Am Coll Cardiol. 2016;68(21):2287-95.
  • 5
    Tajik AJ. Machine Learning for Echocardiographic Imaging: Embarking on Another Incredible Journey. J Am Coll Cardiol.2016;68(21):2296-8.
  • 6
    Austin PC, Tu J V, Ho JE, Levy D, Lee DS. Using methods from the data-mining and machine-learning literature for disease classification and prediction: a case study examining classification of heart failure subtypes. J Clin Epidemiol. 2013;66(4):398-407.
  • 7
    Liu N, Lee MAB, Ho AFW, Haaland B, Fook-Chong S, Koh ZX, et al. Risk stratification for prediction of adverse coronary events in emergency department chest pain patients with a machine learning score compared with the TIMI score. Int J Cardiol . 2014;177(3):1095-7.
  • 8
    Arsanjani R, Dey D, Khachatryan T, Shalev A, Hayes SW, Fish M, et al. Prediction of revascularization after myocardial perfusion SPECT by machine learning in a large population. J Nucl Cardiol. 2014;22(5):877-84.
  • 9
    Garcia E V, Klein JL, Taylor AT. Clinical decision support systems in myocardial perfusion imaging. J Nucl Cardiol. 2014;21(3):427-39.
  • 10
    Arsanjani R, Xu Y, Dey D, Vahistha V, Shalev A, Nakanishi R, et al. Improved accuracy of myocardial perfusion SPECT for detection of coronary artery disease by machine learning in a large population. J Nucl Cardiol. 2013;20(4):553-62.
  • 11
    Medina EL, Mesquita CT, Loques Filho O. Healthcare social networks for patients with cardiovascular diseases and recommendation systems. Int J Cardiovasc Sci. 2016;29(1):80-5.

Publication Dates

  • Publication in this collection
    May-Jun 2017

History

  • Received
    13 Mar 2017
  • Reviewed
    12 Apr 2017
  • Accepted
    12 Apr 2017
Sociedade Brasileira de Cardiologia Avenida Marechal Câmara, 160, sala: 330, Centro, CEP: 20020-907, (21) 3478-2700 - Rio de Janeiro - RJ - Brazil
E-mail: revistaijcs@cardiol.br