The application discloses a cardiovascular
disease home
rehabilitation nursing method and
system, and belongs to the technical field of intelligent
medical treatment. The method synchronously collects physiological data, action postures, voice emotions, electronic medical records and other multi-type and multi-
modal data of patients and completes preprocessing; a multi-
modal cognitive
knowledge graph covering
disease types,
nursing operations,
drug management, complication treatment and patient dynamic state entities is built; a structure
causal model is built relying on the
knowledge graph, an optimal
nursing scheme is screened through
causal reasoning and counterfactual
simulation, a recommended nursing path is output,
rehabilitation nursing interaction is completed by a virtual nurse, patient interaction data is collected, local model incremental training is carried out, gradients are encrypted and then uploaded to a
server for safe aggregation, a globally optimized model is iteratively generated, and an update is issued, so that nursing
recommendation model co-evolution is realized. The application combines multi-
modal cognitive graphs and
causal reasoning, accurately matches home
rehabilitation needs of cardiovascular patients, combines federal
encryption training to guarantee data privacy, continuously optimizes nursing recommendation strategies, and improves the intelligence, individualization and safety of home
rehabilitation nursing.