Selected article for: "training set and validation training"

Author: Liu, Yuliang; Zhang, Quan; Zhao, Geng; Liu, Guohua; Liu, Zhiang
Title: Deep Learning-Based Method of Diagnosing Hyperlipidemia and Providing Diagnostic Markers Automatically
  • Document date: 2020_3_11
  • ID: 1r4gm2d4_39
    Snippet: In this study, our work is the first systematic study on the auxiliary diagnostic system that used human hematological data to automatically diagnose hyperlipidemia and provide the relevant diagnostic basis (automatically prompt diagnostic markers). Experimental results show that our attention deep learning algorithm can not only accurately and automatically diagnose hyperlipidemia but also automatically provide the diagnostic markers of hyperlip.....
    Document: In this study, our work is the first systematic study on the auxiliary diagnostic system that used human hematological data to automatically diagnose hyperlipidemia and provide the relevant diagnostic basis (automatically prompt diagnostic markers). Experimental results show that our attention deep learning algorithm can not only accurately and automatically diagnose hyperlipidemia but also automatically provide the diagnostic markers of hyperlipidemia and the importance of different diagnostic markers. As shown in Figure 7 , the model achieved good and similar performance on both the training set and the validation set, and the model achieved 94% ACC with a completely independent test dataset. Therefore, this phenomenon can be proved that our model has good generalization ability, and it can still achieve better A B Figure performance in the facing of data that does not exist in the training set. As shown in Figures 8 and 9 , the model achieved 97.48% AUC, 92% specificity and 96% sensitivity. It can be proved that the model not only achieves better diagnostic accuracy but also has the good distinguishing ability and high reliability in the facing of different health conditions. An AI system which can auxiliary diagnosis of disease can alleviate the problem of uneven distribution of medical resources and improve the medical level in areas where medical resources are scarce. At the same time, the auxiliary diagnosis system can also speed up the patient's medical treatment process and enhance the patient's medical experience. Because the AI system proposed in this paper does not have the segment of manual feature extraction, it has higher comprehensiveness and objectivity, and reduces the dependence of diagnostic results in the professional level of doctors.

    Search related documents:
    Co phrase search for related documents
    • AI system and diagnosis system: 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21
    • AI system and diagnostic accuracy: 1, 2, 3, 4
    • AI system and diagnostic basis: 1
    • AI system and diagnostic result: 1
    • AI system and diagnostic system: 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21
    • AI system and feature extraction: 1, 2
    • AI system and generalization ability: 1, 2
    • AI system and good generalization ability: 1
    • AI system and medical level: 1
    • AI system and paper propose: 1, 2, 3, 4, 5, 6
    • AI system and paper propose AI system: 1, 2
    • AI system and set validation: 1, 2
    • AI system and test dataset: 1, 2, 3, 4, 5
    • AI system and training set: 1
    • AI system and uneven distribution: 1
    • attention deep learning algorithm and automatically accurately hyperlipidemia diagnose: 1
    • attention deep learning algorithm and automatically hyperlipidemia diagnose: 1, 2
    • attention deep learning algorithm and auxiliary diagnosis: 1, 2
    • attention deep learning algorithm and auxiliary diagnosis system: 1, 2