Probabilistic domain-knowledge modeling of disorder pathogenesis for dynamics forecasting of acute onset
作者：
Highlights：
• Inter-patient variabilities and complex dependencies of the underlying pathogenetic mechanisms complicate modeling efforts.
• Pathogenetic domain knowledge was encoded as an interpretable machine-learning-based medical diagnostic model.
• Probabilistic graphical model fosters the implementation of pathogenetic modelling.
• Inference on the pathogenesis model improves medical queries and onset forecasting.
• Case studies of Paroxysmal Atrial Fibrillation and Obstructive Sleep Apnea pathogenesis were investigated.
摘要
•Inter-patient variabilities and complex dependencies of the underlying pathogenetic mechanisms complicate modeling efforts.•Pathogenetic domain knowledge was encoded as an interpretable machine-learning-based medical diagnostic model.•Probabilistic graphical model fosters the implementation of pathogenetic modelling.•Inference on the pathogenesis model improves medical queries and onset forecasting.•Case studies of Paroxysmal Atrial Fibrillation and Obstructive Sleep Apnea pathogenesis were investigated.
论文关键词：Disease pathogenesis modeling,Domain knowledge integration,Onset forecasting,Graphical,Probabilistic model,Bayesian networks
论文评审过程：Received 30 June 2020, Revised 1 March 2021, Accepted 22 March 2021, Available online 24 March 2021, Version of Record 4 May 2021.
论文官网地址：https://doi.org/10.1016/j.artmed.2021.102056
