The model of medical diagnosis based on machine adaptive learning
Xiaoyan Lu, Xiangshen Li · 2010 Sixth International Conference on Natural Computation · 2010
In order to achieve an accurate model of computer-aided diagnosis, an adaptive machine learning method based on fuzzy mathematics was proposed and used to develop the model. In this method, a fixed-step search algorithm was adopted to train the machine to establish the model, which is optimized through adjusting related parameters and continually increasing the value of object parameter (OPT) until reaching the satisfactory terminal conditions. A sample was selected in order to test the effectiveness of the method. Comparing the results between expert's diagnosis and model-based diagnosis in 146 acute cerebral infarction patients in Department of Neurology in the First Affiliated Hospital of Shanxi Medical University, the discrepancies are minimal. Furthermore, the accuracy rate is up to 90.7% in diagnosis of 86 outpatients. As the result, the model constructed by the method is effective and can be used in the practical medical diagnosis.