Artificial intelligence to enhance clinical value across the spectrum of cardiovascular healthcare

Abstract

Artificial intelligence (AI) is increasingly being utilized in healthcare. This article provides clinicians and researchers with a step-wise foundation for high-value AI that can be applied to a variety of different data modalities. The aim is to improve the transparency and application of AI methods, with the potential to benefit patients in routine cardiovascular care. Following a clear research hypothesis, an AI-based workflow begins with data selection and pre-processing prior to analysis, with the type of data (structured, semi-structured, or unstructured) determining what type of pre-processing steps and machine-learning algorithms are required. Algorithmic and data validation should be performed to ensure the robustness of the chosen methodology, followed by an objective evaluation of performance. Seven case studies are provided to highlight the wide variety of data modalities and clinical questions that can benefit from modern AI techniques, with a focus on applying them to cardiovascular disease management. Despite the growing use of AI, further education for healthcare workers, researchers, and the public are needed to aid understanding of how AI works and to close the existing gap in knowledge. In addition, issues regarding data access, sharing, and security must be addressed to ensure full engagement by patients and the public. The application of AI within healthcare provides an opportunity for clinicians to deliver a more personalized approach to medical care by accounting for confounders, interactions, and the rising prevalence of multi-morbidity.

Keywords: Artificial intelligence; Healthcare; Management; Treatment.

Graphical Abstract

Step-wise workflow to improve the value of artificial intelligence in clinical research. AI, artificial intelligence.

Figure 1

Artificial intelligence framework for clinical application. An overview of a framework to apply artificial intelligence, with standardized assessment and reporting of data acquisition, data pre-processing, and machine learning. These steps are interlinked with evaluation and validation to provide clinical value in real-world applications

Figure 2

Examples of analytic steps based on data modality. Examples of pre-processing and machine-learning approaches based on the type of data available. CNN, convolutional neural network; DNN, deep neural network; LSTM, long short-term memory recurrent network; O2PLS, orthogonal two-way PLS; PCA, principal component analysis; PLS, partial least squares regression; SVM, support vector machine

Figure 3

Case studies on how to improve phenotyping of patients. (A) Case study 1 using an artificial intelligence pipeline to cluster patients across nine clinical trials in patients with heart failure, reduced ejection fraction, and concomitant atrial fibrillation. Circles represent the average mortality risk (size relative to the number of patients in that cluster), with odds ratios for the efficacy of beta-blockers vs. placebo for all-cause mortality, and radar plots summarizing descriptors for each cluster compared with the cohort average (open orange circles at the centre of each radar plot). Reproduced and amended from. (B) Case study 2 showing the gene–gene interaction network of the top genes based on integrating transcriptomics and inflammation markers in patients with primary sclerosing cholangitis-inflammatory bowel disease and ulcerative colitis, externally validating the group sparse two-way orthogonal partial least squares approach. (C) Case study 4 depicting receiver operating characteristic curves for incident heart failure prediction in routine clinical practice, demonstrating superiority of the deep neural network model derived from digital electrocardiograms compared with clinical parameters (P < 0.00001). AF, atrial fibrillation; ECG, electrocardiogram; EHR, electronic healthcare records

Figure 4

Case studies on how to incorporate novel investigations. (A) Case study 5 demonstrating ambulatory time-series data obtained from consumer wearable devices in two patients with atrial fibrillation. The blue line indicates minute-to-minute heart rate, in relation to the orange bars showing physical activity measured by step count over a 24 h period. Both patients had permanent atrial fibrillation and heart failure, the same level of symptoms (New York Heart Association Class III) and were treated with optimal medical therapy including heart rate control at the time of data capture (from the RATE-AF clinical trial). (B) Case study 6 showing the pipeline for signal processing of photoplethysmography signals using a smartphone camera, using deep learning (convolutional neural networks on pre-processed signals) and machine learning (features ranking and support vector machine on standardized features) to obtain clean signals for analysis and prediction of vascular ageing. (C) Case study 7 using dynamic neural network segmentation of volumetric cardiac magnetic resonance imaging to train a denoising autoencoder for stratification of survival in patients with pulmonary hypertension. AF, atrial fibrillation; CNN, convolutional neural network; HF, heart failure; PPG, photoplethysmography.