Multi-modal fusion swallowing disorder recognition and evaluation system for disabled elders in old-age care institutions

Abstract

The invention provides a multi-modal fusion swallowing disorder recognition and evaluation system for disabled elders in old-age care institutions, which comprises a multi-modal data acquisition module, a data processing and analysis module and an evaluation and output module; the multi-modal data acquisition module realizes non-contact data acquisition of the whole swallowing process, and collects the whole swallowing process data in a non-contact way through high-definition vision, 3D depth quantization, infrared thermal imaging and meter entry units; after CNN feature extraction, weighting and decision-making level fusion algorithm, improved ResNet-LSTM and random forest model analysis, a personalized report with a risk score of 0-100 and SSA / FOIS level is generated, and the medium and high risks are automatically warned. The system breaks through the limitation of single mode, has high evaluation accuracy, reduces the missed diagnosis rate by more than 60 percent, adapts to multiple scenarios of institutions, is convenient to operate, and helps to promote the smart old-age care on a large scale.