Deep Learning-Based Method of Diagnosing Hyperlipidemia and Providing Diagnostic Markers Automatically

Liu, Yuliang; Zhang, Quan; Zhao, Geng; Liu, Guohua; Liu, Zhiang

adens the application scope of the system, such as medical teaching (provide recommended diagnosis results and evidence to inexperienced physicians). Most importantly, the traditional method of researching diagnostic markers is often to observe the clinical manifestations of dozens or hundreds of patients artificially, and then find the diagnostic markers of diseases according to the method of statistics. The traditional methods mentioned above are often difficult to synthesize large quantities of data and have a long research cycle. 42, 43 Andrei M. Beliaevc et al used 96 patients samples to discover diagnostic markers of acute cholangitis. Akihiko Yuki et al found CADM1 is a diagnostic marker in early-stage mycosis fungoides with 58 cases. 44 Their research results have achieved good performance. Artificial analysis of limited data (dozens of samples) has the characteristics of one-sidedness and long research time, which undoubtedly increases the difficulty of researching diagnostic markers. The auxiliary diagnostic system proposed in this paper can automatically provide diagnostic markers by integrating a large amount of clinical data, which reduces the blindness of researching diagnostic markers and speeds up the discovery process of new diagnostic markers to a certain extent. In addition, automatic analysis of large quantities of samples can improve the reliability of the model and reduce the contingency caused by small quantities of samples.