Graph neural networks for clinical risk prediction based on electronic health records: A survey.

OBJECTIVE: This study aims to comprehensively review the use of graph neural networks (GNNs) for clinical risk prediction based on electronic health records (EHRs). The primary goal is to provide an overview of the state-of-the-art of this subject, highlighting ongoing research efforts and identifying existing challenges in developing effective GNNs for improved prediction of clinical risks. METHODS: A search was conducted in the Scopus, PubMed, ACM Digital Library, and Embase databases to identify relevant English-language papers that used GNNs for clinical risk prediction based on EHR data. The study includes original research papers published between January 2009 and May 2023. RESULTS: Following the initial screening process, 50 articles were included in the data collection. A significant increase in publications from 2020 was observed, with most selected papers focusing on diagnosis prediction (n = 36). The study revealed that the graph attention network (GAT) (n = 19) was the most prevalent architecture, and MIMIC-III (n = 23) was the most common data resource. CONCLUSION: GNNs are relevant tools for predicting clinical risk by accounting for the relational aspects among medical events and entities and managing large volumes of EHR data. Future studies in this area may address challenges such as EHR data heterogeneity, multimodality, and model interpretability, aiming to develop more holistic GNN models that can produce more accurate predictions, be effectively implemented in clinical settings, and ultimately improve patient care.

Randomised, siteless study to compare systematic atrial fibrillation screening using enrichment by a risk prediction model with standard care in a Swedish population aged ≥ 65 years: CONSIDERING-AF study design.

INTRODUCTION: Atrial fibrillation (AF) is the most common arrhythmia and confers an increased risk of mortality, stroke, heart failure and cognitive decline. There is growing interest in AF screening; however, the most suitable population and device for AF detection remains to be elucidated. Here, we present the design of the CONSIDERING-AF (deteCtiON and Stroke preventIon by moDEl scRreenING for Atrial Fibrillation) study. METHODS AND ANALYSIS: CONSIDERING-AF is a randomised, controlled, siteless, non-blinded diagnostic superiority trial with four parallel groups and a primary endpoint of identifying AF during a 6-month study period set in Region Halland, Sweden. In each group, 740 individuals aged≥65 years will be included. The primary objective is to compare the intervention of AF screening enrichment using a risk prediction model (RPM), followed by 14 days of a continuous ECG patch, with no intervention (standard care). Primary outcome is defined as the incident AF recorded in the Region Halland Information Database after 6 months as compared with standard care. Secondary endpoints include the difference in incident AF between groups enriched or not by the RPM, with and without an invitation to 14 days of continuous ECG recording, and the proportions of oral anticoagulation treatment in the four groups. ETHICS AND DISSEMINATION: This study has ethical approval from the Swedish Ethical Review Authority. Results will be published in peer-reviewed international journals. TRIAL REGISTRATION NUMBER: NCT05838781.