Application of large models in wearable device-based cardiovascular risk monitoring

Text matching method was used to identify the corresponding Framingham risk factors from each cardiovascular disease examination report, and the corresponding risk coefficient given by the Framingham heart risk score scale was used as the Framingham risk coefficient. Words similar to each Framingham risk factor were collected as similar risk factors, and the self-attention agency model was used to find corresponding similar risk factors from each cardiovascular disease examination report, and the corresponding similar risk coefficient was given. Text semantic analysis method was used to find out the risk factors from each cardiovascular disease examination report, and the corresponding risk coefficients were calculated. The three types of risk factors were identified and the basic information of patients were spliced and integrated to generate the initial risk analysis report, and the medical large language model was used to adjust the initial risk analysis report through the fine-tuning of prompt words, and the final risk analysis report was generated.