Operation opportunity prediction model for patient with acute necrotizing pancreatitis

Abstract

The invention discloses an operation opportunity prediction model for a patient with acute necrotizing pancreatitis, and the model comprises the following steps: S100, collecting and cleaning case data which comprises admission data and hospitalization data of the patient; S200, sorting the case data to obtain preprocessed data, wherein the sequence form of the preprocessed data is {Variables, Time}; S300, carrying out one-hot coding on the category type variables in the preprocessed data by using an Embedding mechanism, and then mapping the category type variables into a real vector space; S400, normalizing the data obtained in the step S300, introducing an LSTM model which is a multi-task model, wherein the output layer of the LSTM model is provided with four outcome indexes: 48-hour death after the patient is hospitalized, the number of remaining hospitalized days, whether organ failure occurs or not and whether an operation is performed or not. The optimal operation opportunity ofeach patient with acute pancreatitis necrosis can be determined according to the characteristics of the patient, so the prognosis condition of the patient is improved.

Description

technical field [0001] The invention relates to the field of neural networks, in particular to a model for predicting operation timing of patients with necrotizing acute pancreatitis. Background technique [0002] At present, there is almost no prediction model for the timing of surgery for patients with necrotizing acute pancreatitis in my country; as we all know, the time series characteristics and variable characteristics of different diseases are different. exists and cannot be applied directly. Previous models of disease prediction mainly focused on traditional statistical methods. Due to the heterogeneity of data, most of them were difficult to consider complex time information and simulate the real disease development status of patients. [0003] Domestic and foreign guidelines suggest that the timing of surgery for patients with acute pancreatitis and necrosis should be delayed as far as possible after four weeks, but they do not give specific recommendations on whic...

AI Technical Summary

Problems solved by technology

Not all patients with acute pancreatitis and necrosis can wait until four weeks later for surgery, and patients with acute pancreatitis and necrosis who can wait until four weeks later have the problem of when is the best timing for surgery

Images