A deep-learning-based unsupervised model on esophageal manometry using variational autoencoder
作者：
Highlights：
• A variational auto-encoder was developed with 32-thousand swallows of manometry data.
• Generative capability was compared among models with different latent space and loss.
• Key motility patterns were found well-encoded by 4-dimensional latent variables.
• Discriminative analysis revealed clusters consistent with clinical impression.
• Understanding of swallow-level data will guide the development of study-level models.
摘要
•A variational auto-encoder was developed with 32-thousand swallows of manometry data.•Generative capability was compared among models with different latent space and loss.•Key motility patterns were found well-encoded by 4-dimensional latent variables.•Discriminative analysis revealed clusters consistent with clinical impression.•Understanding of swallow-level data will guide the development of study-level models.
论文关键词：High-resolution manometry,Artificial intelligence,Esophageal diagnosis,Generative modeling
论文评审过程：Received 12 June 2020, Revised 19 October 2020, Accepted 28 December 2020, Available online 5 January 2021, Version of Record 10 January 2021.
论文官网地址：https://doi.org/10.1016/j.artmed.2020.102006
