Considering that there is relatively abundant literature on the theoretical research of dynamic Bayesian networks

but there is still ample research space for their practical applications

this article systematically demonstrates the process of constructing a complex vital sign network based on real-time medical data

and predicts the long-term status of unknown indicators (implicit indicators) that are difficult to observe based on observed patient physiological indicators (explicit indicators). That is to say

the modeling method shown in this article can help medical staff better monitor and predict possible abnormal fluctuations in physiological indicators of patients

and thus help doctors provide more accurate diagnostic results. Specifically

this article first uses the maximum minimum mountain climbing method to build an initial dynamic Bayesian network structure and uses the maximum likelihood estimation method to determine the probability parameters in the network

Finally

this article uses interface algorithms to predict the most likely state values of other indicators in the later stages based on the monitoring results of simulated observation indicators at the initial time. At the same time

the results showed that although there was no significant linear relationship between the eight main indicators such as lymphocyte concentration and pulse

from a clustering perspective

the similarity between other indicators except pulse

creatinine concentration

and red blood cell concentration was relatively high. In addition

if the patient's initial pulse frequency and red blood cell concentration are at level 2 and level 4

respectively

the likelihood of their creatinine concentration at level 2 is highest after two stages. If the patient's non-invasive systolic and diastolic blood pressure at the initial stage are at level 3 and level 2

respectively

then their hemoglobin concentration at the current stage is most likely to be at level 3

and their red blood cell concentration is most likely to be at level 2 after the two stages.