Disease prediction and risk assessment of internet of things medical data based on causal reasoning techniques

Abstract

With the opening of the big data era, the exponential growth of IoT medical data provides powerful data support for the medical and health care fields. To effectively utilize these data, the study innovatively utilizes direct method for learning a linear non-Gaussian acyclic model with graph capsule neural network and proposes a causal inference technique based on IoT medical data disease prediction and risk assessment model. The experimental results indicated that the proposed method of the study could effectively clarify the causal relationship between disease features and thus enhance the stability of the model compared with the popular disease prediction and risk assessment methods of the same type. The research model achieved 86.01%, 94.26%, 95.03%, 95.12%, and 94.21% classification accuracy for the causality maps of diabetes dataset, stroke dataset, heart failure dataset, heart disease dataset, and cardiovascular disease dataset, respectively. In addition, the research model also achieved 97.13%, 97.26%, and 77.89% for precision, recall, and F1 value on the test set, respectively. It had good disease prediction and risk assessment ability. The research results are of great significance in improving the accuracy and efficiency of disease prediction and risk assessment. It can not only provide strong technical support for the application in related fields, but also is expected to promote the further development and application of causal inference technique in disease prediction and risk assessment.