The added value of temporal data and the best way to handle it: A use-case for atrial fibrillation using general practitioner data

• Published in: Computers in biology and medicine Vol. 171; p. 108097
• Main Authors: Bennis, Frank C., Aussems, Claire, Korevaar, Joke C., Hoogendoorn, Mark
• Format: Journal Article
• Language: English
• Published: United States Elsevier Ltd 01.03.2024; Elsevier Limited
• Subjects: Algorithms; Atrial fibrillation; Atrial Fibrillation - diagnosis; Benchmarks; Blood pressure; Cardiac arrhythmia; Chronic illnesses; CKConv; Clinical prediction; Convolutional neural networks; Datasets; Deep learning; Disease; Fibrillation; General practitioner; General Practitioners; Humans; Internal Medicine; Logistic Models; Machine learning; Neural networks; Neural Networks, Computer; Other; Patients; Performance evaluation; Predictions; Primary care; Regression analysis; Temporal; Clinical prediction; CKConv; Temporal; Convolutional neural networks; General practitioner; Atrial fibrillation
• ISSN: 0010-4825; 1879-0534; 1879-0534
• DOI: 10.1016/j.compbiomed.2024.108097

Temporal data has numerous challenges for deep learning such as irregularity of sampling. New algorithms are being developed that can handle these temporal challenges better. However, it is unclear how the performance ranges from classical non-temporal models to newly developed algorithms. Therefore, this study compares different non-temporal and temporal algorithms for a relevant use case, the prediction of atrial fibrillation (AF) using general practitioner (GP) data. Three datasets with a 365-day observation window and prediction windows of 14, 180 and 360 days were used. Data consisted of medication, lab, symptom, and chronic diseases codings registered by the GP. The benchmark discarded temporality and used logistic regression, XGBoost models and neural networks on the presence of codings over the whole year. Pattern data extracted common patterns of GP codings and tested using the same algorithms. LSTM and CKConv models were trained as models incorporating temporality. Algorithms which incorporated temporality (LSTM and CKConv, (max AUC 0.734 at 360 days prediction window) outperformed both benchmark and pattern algorithms (max AUC 0.723, with a significant improvement using the 360 days prediction window (p = 0.04). The difference between the benchmark and the LSTM or CKConv algorithm decreased with smaller prediction windows, indicating temporal importance for longer prediction windows. The CKConv and LSTM algorithm performed similarly, possibly due to limited sequence length. Temporal models outperformed non-temporal models for the prediction of AF. For temporal models, CKConv is a promising algorithm to handle temporal data using GP data as it can handle irregular data. •Atrial fibrillation can be predicted one year in advance from GP data.•Temporal data significantly improves the prediction of AF using neural networks.•CKConv and LSTM outperform benchmark algorithms for AF prediction.•CKConv and LSTM perform similar for predicting AF.