Research On Diagnosis Of Imbalanced Breast Cancer Based On Contrastive Learning

The continuous improvement of medical technology and the efficiency of doctors’ diagnoses have always been at the heart of the country’s medical development.In China,breast cancer disease,as one of the three major cancers,is extremely serious.With the development of modern technology and information technology,the application of machine learning algorithms in the medical field,especially in the processing of medical data has taken the majority,and computer-aided diagnosis has become a popular direction of research and a major trend.The emergence of complementary diagnostic technology not only brings convenience to the examination and collation of patient information but also helps to improve the efficiency and accuracy of doctors in diagnosing patients.Breast cancer diagnostic technology can improve the accuracy of doctors’ detection of patients,reduce their diagnosis time,and enable patients to be treated quickly and effectively,thereby reducing the likelihood of their disease deteriorating and becoming cancerous.A breast cancer diagnosis is a data classification problem.Most of the current research on breast cancer disease diagnosis is based on machine learning algorithms,which work well for the classification of balanced data,but for the unbalanced data present in breast cancer disease,the classification results are often biased towards the majority class and the results are not ideal.Based on this,this paper proposes a breast cancer detection algorithm based on supervised contrast learning.The breast cancer detection algorithm based on supervised contrast learning mainly embeds supervised contrast learning into a multilayer perceptron.Firstly,the original positive and negative sample imbalance is balanced through data augmentation in supervised contrast learning,secondly,the data augmented samples are processed by the multilayer perceptron for classification work,and finally,the supervised contrast loss function and cross-entropy loss function are combined to form an overall loss to guide the model to fully learn to acquire data features.The performance of the multilayer perceptron algorithm based on supervised contrast learning is validated based on publicly available breast cancer datasets.The experimental results demonstrate that the multilayer perceptron based on supervised contrast learning can handle imbalanced data,and to a certain extent can improve the classification accuracy and efficiency of imbalanced data.Finally,the breast cancer disease diagnosis model constructed based on breast cancer disease data was applied to a breast cancer disease medical aid platform to realize a convenient,efficient,and accurate breast cancer disease medical aid system.