Shijie Sun

master

Chongqing University of Posts and Telecommunications

none

NLST (Learn more about this study)

NLST-963

Sep 28, 2022

Lung cancer risk prediction model based on demographic data and diagnostic data

Considering that most of the models published in the medical field are based on logistic regression, decision tree and support vector machine, only demographic data, smoking history, family history, chronic history and other data are considered in the data. These data types are not quantifiable data, which may not be conducive to the calculation of risk index, while most of the models in the computer field are based on clinical data. Pathological images Radiomics images or electronic case texts are used for risk assessment, missing demographic and other data, which limits the application of the model. Therefore, it is considered to combine demographic and other data with diagnostic data (such as medical images) to develop algorithms.

1. The risk prediction model is constructed, and the accuracy of the model is more than 85%

2. When the input attributes are reduced, the model can still perform well

3. The model can better show the process of risk changing with patient attributes