Scholarlink

Accurate cancer stage classification is crucial for tailored treatment strategies. Here, we propose a novel method utilizing Support Vector Machines (SVM) and Long Short-Term Memory (LSTM) networks to classify cancer stages from single-cell RNA sequencing (scRNA-seq) gene expression data. SVM handles both linear and nonlinear tasks, while LSTM captures temporal dependencies, exploiting the rich information in scRNA-seq data. Several evaluation metrics are utilized to compare the effectiveness of the models. To overcome the sparsity and large volume of scRNA-seq data, we utilize preprocessing and dimensionality reduction methods. Different kernels in SVM, including linear, polynomial, and radial basis functions (RBF), are explored to assess their impact on classification accuracy. Experimental results indicate room for improvement, encouraging further research to enhance the accuracy of our deep learning model. This study contributes to advancing the field of cancer stage classification using scRNA-seq data and underscores the potential of machine learning and deep learning models in clinical decision-making, paving the way for more personalized treatment approaches. Both SVM-Linear (avg. accuracy = 88.46%, ROC = 0.97) and LSTM (avg. accuracy = 87.90, ROC = 0.98) perform comparatively well during testing.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Leveraging SVM and LSTM Networks for Enhanced Cancer Stage Detection from Single-Cell RNA-Seq Data

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Gurupriya Takkar,
  • Khalid Raza

摘要

Accurate cancer stage classification is crucial for tailored treatment strategies. Here, we propose a novel method utilizing Support Vector Machines (SVM) and Long Short-Term Memory (LSTM) networks to classify cancer stages from single-cell RNA sequencing (scRNA-seq) gene expression data. SVM handles both linear and nonlinear tasks, while LSTM captures temporal dependencies, exploiting the rich information in scRNA-seq data. Several evaluation metrics are utilized to compare the effectiveness of the models. To overcome the sparsity and large volume of scRNA-seq data, we utilize preprocessing and dimensionality reduction methods. Different kernels in SVM, including linear, polynomial, and radial basis functions (RBF), are explored to assess their impact on classification accuracy. Experimental results indicate room for improvement, encouraging further research to enhance the accuracy of our deep learning model. This study contributes to advancing the field of cancer stage classification using scRNA-seq data and underscores the potential of machine learning and deep learning models in clinical decision-making, paving the way for more personalized treatment approaches. Both SVM-Linear (avg. accuracy = 88.46%, ROC = 0.97) and LSTM (avg. accuracy = 87.90, ROC = 0.98) perform comparatively well during testing.