Estimating boundary conditions of pharyngeal bolus movement by neural network inversion

Dysphagia is a common clinical symptom causing life-threatening pulmonary complications, such as choking and aspiration pneumonia. The forces applied to the bolus are the most important indicators while difficult to estimate by using current tools for the clinical evaluation of dysphagia. Pharyngeal bolus modeling using computational fluid dynamics can provide information about the forces which can be used to ensure safe swallowing. In this paper, a neural network inversion technique is used to help specify velocity boundary conditions at the glossopalatal junction, which are required to model pharyngeal bolus movement based on finite element method