Artificial Intelligence in Cardiology

Abstract

Artificial intelligence and machine learning are poised to influence nearly every aspect of the human condition, and cardiology is not an exception to this trend. This paper provides a guide for clinicians on relevant aspects of artificial intelligence and machine learning, reviews selected applications of these methods in cardiology to date, and identifies how cardiovascular medicine could incorporate artificial intelligence in the future. In particular, the paper first reviews predictive modeling concepts relevant to cardiology such as feature selection and frequent pitfalls such as improper dichotomization. Second, it discusses common algorithms used in supervised learning and reviews selected applications in cardiology and related disciplines. Third, it describes the advent of deep learning and related methods collectively called unsupervised learning, provides contextual examples both in general medicine and in cardiovascular medicine, and then explains how these methods could be applied to enable precision cardiology and improve patient outcomes.

Current state and future prospects of artificial intelligence in ophthalmology: a review

“…To date, advances have been made predominantly in weak AI, so AI will be unlikely to replace most medical specialists for the foreseeable future. 8,12,13 Continuing on from this, we expect that significant headway will be made under medical imaging, given deep learning's particular proclivity for image processing. 8 In addition, combining the strengths of human clinicians with the strengths of deep learning systems should reduce errors in diagnostics and therapeutics that are inherent in our current system.…”

Orthodontics in the era of big data analytics

“…Specialties such as oncology, cardiology, surgery and radiology have been at the forefront of adapting to the big data environment and have made massive strides in using some of the big data analytical methods and data sources to deliver personalized and precision care . Using different data sources and new age big data analytic tools, these specialties have challenged the long‐held traditional monolithic view of disease . Big data analytic tools have been successful in identifying personalized disease concepts and subtypes and delivering targeted aetiology‐based therapies .…”

“…Traditionally, clinical research has been centred on a model where there is a clearly defined hypothesis, primary outcomes, set of predictor variables selected on biological plausibility and experimental settings that are well controlled. The statistical methods that are used to test the study hypotheses, for example logistic regression for binomial outcomes and linear regression for continuous outcomes, work best under strict assumptions including lack of multicollinearity between variables, absence of endogeneity and model fitness . However, these assumptions do not hold when applied to real‐world data.…”

“…The machine‐learning approach, in contrast to traditional statistical approaches, is not based on a study hypothesis nor is it dependent upon strict data parameters and data assumptions . Instead, the machine‐learning approach examines patterns in data and can simultaneously account for multiple variables and interactions between variables and assumes that endogeneity is ubiquitous . Consequently, machine learning is the analytical approach of choice when we are inundated with big data drawn from multiple sources and in a wide range of formats.…”

“…Consequently, machine learning is the analytical approach of choice when we are inundated with big data drawn from multiple sources and in a wide range of formats. Machine learning has become increasingly utilized in health care in recent years and has been successful in identifying disease subtypes as well as providing insights into rare diseases and associated outcomes, predictive modelling, identification of never events and targeted therapies . Broadly, machine learning can be either supervised or unsupervised.…”