{"doi":"10.1093/europace/euaf085.294","title":"Predicting dementia in patients with atrial fibrillation using machine learning of a large multimodal registry","abstract":"Abstract Background Growing evidence indicates a significant association between atrial fibrillation (AF) and dementia. This link has been compelling enough to shape recent AF management guidelines, which now provide a class II indication for catheter ablation to mitigate the risk of dementia. However, which AF patients may develop dementia remains poorly understood. Our objective was to bridge this knowledge gap by applying machine learning techniques to a large Community Registry of AF patients in which we studied a broad range of multimodal data. Method We studied 7,498 patients diagnosed with atrial fibrillation in a large U.S. Medical Network that includes academic and community medical centers. We applied statistical univariate feature selection followed by stepwise logistic regression to 104 clinical variables to develop a predictive model for dementia. In order to better understand the risk contributors, we ranked the features according to their importance in the predictive model. Results Patients were aged 77.9±14.7 years, 39.7% female, with body mass index (BMI) 27.4±7.0 kg/m2. An ICD code for dementia (codes F03, F02.8) was found in N=415. Using logistic regression, the best performing model achieved an AUC=0.83, sensitivity=0.73, specificity=0.74, with average precision=0.24 (Fig 1A). The top contributors to optimal model performance extended beyond CHADS2VASc to other cardiovascular and non-cardiovascular features (Fig 1B). Conclusion In a large AF registry, we developed a machine learning model that identified risk for dementia with high accuracy by incorporating 104 clinical variables. Further development of the model, and generalization to larger cohorts may better delineate dementia risk and identify combinations of characteristics most strongly associated with dementia. This approach could help identify personalized tools to guide AF ablation.","journal":"EP Europace","year":2025,"id":566583,"datarank":0.0,"base_score":0.0,"endowment":0.0,"self_citation_contribution":0.0,"citation_network_contribution":0.0,"self_endowment_contribution":0.0,"citer_contribution":0.0,"corpus_percentile":null,"corpus_rank":null,"citation_count":0,"citer_count":0,"citers_with_citation_signal":0,"citers_with_endowment":0,"datacite_reuse_total":0,"is_dataset":false,"is_dataset_confidence":0.9573,"is_data_producer":false,"deposit_databanks":null,"is_oa":true,"file_count":0,"downloads":0,"has_version_chain":false,"published_date":"2025-01-01","fair_score":null,"fair_percentile":null,"algorithm_id":"datarank_citation_only_1hop_v6","ranking_scope":"data_only","authors":[{"id":939279,"name":"Sabyasachi Bandyopadhyay","orcid":"0000-0003-4617-7832","position":1,"is_corresponding":false},{"id":533281,"name":"Prasanth Ganesan","orcid":"0000-0002-1885-0690","position":2,"is_corresponding":false},{"id":979400,"name":"H J Chang","orcid":null,"position":3,"is_corresponding":false},{"id":1009811,"name":"Renhai Feng","orcid":"0000-0001-7194-6889","position":4,"is_corresponding":false},{"id":1470638,"name":"Kathryn Brennan","orcid":null,"position":5,"is_corresponding":false},{"id":241956,"name":"Albert J. Rogers","orcid":"0000-0001-6585-534X","position":6,"is_corresponding":false},{"id":328098,"name":"Paul Clopton","orcid":"0000-0002-0642-0861","position":7,"is_corresponding":false},{"id":586080,"name":"Chad Brodt","orcid":"0000-0002-2790-5040","position":8,"is_corresponding":false},{"id":241959,"name":"Sanjiv M. Narayan","orcid":"0000-0001-7552-5053","position":9,"is_corresponding":false},{"id":1242702,"name":"Vivek Srivastava","orcid":"0000-0003-0503-1108","position":0,"is_corresponding":true}],"reference_count":0,"raw_metadata":null,"created_at":"2026-07-19T02:56:36.440192Z","pmid":null,"pmcid":null,"fwci":null,"citation_percentile":null,"influential_citations":0,"oa_status":null,"license":null,"views":0,"total_file_size_bytes":0,"version_count":0,"fair_f":null,"fair_a":null,"fair_i":null,"fair_r":null,"fair_zscore":null,"fair_rationale":null,"fair_model":null,"fair_agent_version":null,"fair_fulltext_source":null,"fair_has_llm":null,"fair_computed_at":null,"clinical_trials":[],"software_tools":[],"db_accessions":[],"linked_datasets":[],"topics":[]}