A deep attention-based encoder for the prediction of type 2 diabetes longitudinal outcomes from routinely collected health care data
Por:
Manzini, E, Vlacho, B, Franch-Nadal, J, Escudero, J, Génova, A, Reixach, E, Andrés, E, Pizarro, I, Mauricio, D, Perera-Lluna, A
Publicada:
15 may 2025
Ahead of Print:
1 feb 2025
Resumen:
Recent evidence indicates that Type 2 Diabetes Mellitus (T2DM) is a complex and highly heterogeneous disease involving various pathophysiological and genetic pathways, which presents clinicians with challenges in disease management. While deep learning models have made significant progress in helping practitioners manage T2DM treatments, several important limitations persist. In this paper we propose DARE, a model based on the transformer encoder, designed for analyzing longitudinal heterogeneous diabetes data. The model can be easily fine-tuned for various clinical prediction tasks, enabling a computational approach to assist clinicians in the management of the disease. We trained DARE using data from over 200,000 diabetic subjects from the primary healthcare SIDIAP database, which includes diagnosis and drug codes, along with various clinical and analytical measurements. After an unsupervised pre-training phase, we fine-tuned the model for predicting three specific clinical outcomes: i) occurrence of comorbidity, ii) achievement of target glycemic control (defined as glycated hemoglobin <7%) and iii) changes in glucose-lowering treatment. In cross-validation, the embedding vectors generated by DARE outperformed those from baseline models (comorbidities prediction task AUC = 0.88, treatment prediction task AUC = 0.91, HbA1c target prediction task AUC = 0.82). Our findings suggest that attention-based encoders improve results with respect to different deep learning and classical baseline models when used to predict different clinical relevant outcomes from T2DM longitudinal data.
Filiaciones:
Manzini, E:
Univ Politecn Cataluna, Inst Recerca & Innovacio Salut IRIS, Barcelona Tech, B2SLab, Barcelona, Spain
Networking Biomed Res Ctr Subject Area Bioengn Bio, Madrid, Spain
Inst Recerca St Joan Deu, Barcelona, Spain
Vlacho, B:
Fundacio Inst Univ Recerca Atencio Primaria Salut, DAP Cat Grp, Unitat Suport Recerca Barcelona, Barcelona, Spain
Inst Salud Carlos III, CIBER Diabet & Associated Metab Dis CIBERDEM, Madrid, Spain
Franch-Nadal, J:
Fundacio Inst Univ Recerca Atencio Primaria Salut, DAP Cat Grp, Unitat Suport Recerca Barcelona, Barcelona, Spain
Inst Salud Carlos III, CIBER Diabet & Associated Metab Dis CIBERDEM, Madrid, Spain
Inst Catala Salut, Primary Hlth Care Ctr Raval Sud, Gerencia Atencio Primaria, Barcelona, Spain
Escudero, J:
Evidenze Hlth, Barcelona, Spain
Génova, A:
Evidenze Hlth, Barcelona, Spain
Reixach, E:
Generalitat Catalunya, Fundacio TIC Salut Social, Dept Salut, Barcelona, Spain
Andrés, E:
Generalitat Catalunya, Fundacio TIC Salut Social, Dept Salut, Barcelona, Spain
Pizarro, I:
IEDIS, Madrid, Spain
Mauricio, D:
Fundacio Inst Univ Recerca Atencio Primaria Salut, DAP Cat Grp, Unitat Suport Recerca Barcelona, Barcelona, Spain
Inst Salud Carlos III, CIBER Diabet & Associated Metab Dis CIBERDEM, Madrid, Spain
Hosp Santa Creu & Sant Pau, Dept Endocrinol & Nutr, IR St Pau, Barcelona, Spain
Univ Vic Cent Univ Catalonia, Dept Med, Vic, Spain
Perera-Lluna, A:
Univ Politecn Cataluna, Inst Recerca & Innovacio Salut IRIS, Barcelona Tech, B2SLab, Barcelona, Spain
Networking Biomed Res Ctr Subject Area Bioengn Bio, Madrid, Spain
Inst Recerca St Joan Deu, Barcelona, Spain
hybrid
|