Scholarlink

Background <p>The prognosis of ovarian cancer is closely related to the degree of immune cell infiltration within the tumor microenvironment. However, current methods for assessing immune infiltration have certain subjective limitations. This study aimed to establish an objective assessment model based on machine learning and single-cell RNA sequencing data to provide a basis for the individualized immunotherapy of ovarian cancer.</p> Methods <p>This study integrated gene expression data from multiple public databases for ovarian cancer, including different histological subtypes and immune infiltration levels. We utilized single-cell RNA sequencing data to characterize immune cell populations with unprecedented resolution. After correcting for batch effects, we constructed machine learning models based on RandomForest and SVM to predict the immune infiltration status of samples at the single-cell level. The models were evaluated and optimized using cross-validation methods.</p> Results <p>Our machine learning models demonstrated high accuracy and robustness in predicting the immune infiltration status of ovarian cancer. The RandomForest model achieved an AUC of 0.88 on an independent test set, outperforming traditional immune scoring indices. Single-cell analysis revealed distinct immune cell subpopulations and their spatial distribution within tumors. The models also identified key gene features associated with immune infiltration at cellular resolution, providing clues for further understanding the immune microenvironment of ovarian cancer.</p> Conclusion <p>The machine learning-based approach for evaluating immune infiltration in ovarian cancer at the single-cell level can rapidly and objectively predict the immune status, and discover potential biomarkers and therapeutic targets. This method provides a new strategy and tool for achieving individualized immunotherapy for ovarian cancer.</p> <p><i>Clinical trial declaration</i>: This research is not a clinical trial and is exempt from clinical trial registration requirements.</p>

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 machine learning models to evaluate immune infiltration in the ovarian cancer microenvironment: a single-cell analysis approach

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年7月9日

查看全文
  • Jie Liu,
  • Baoguo Xia,
  • Bingxin Li,
  • Hui Liang

摘要

Background

The prognosis of ovarian cancer is closely related to the degree of immune cell infiltration within the tumor microenvironment. However, current methods for assessing immune infiltration have certain subjective limitations. This study aimed to establish an objective assessment model based on machine learning and single-cell RNA sequencing data to provide a basis for the individualized immunotherapy of ovarian cancer.

Methods

This study integrated gene expression data from multiple public databases for ovarian cancer, including different histological subtypes and immune infiltration levels. We utilized single-cell RNA sequencing data to characterize immune cell populations with unprecedented resolution. After correcting for batch effects, we constructed machine learning models based on RandomForest and SVM to predict the immune infiltration status of samples at the single-cell level. The models were evaluated and optimized using cross-validation methods.

Results

Our machine learning models demonstrated high accuracy and robustness in predicting the immune infiltration status of ovarian cancer. The RandomForest model achieved an AUC of 0.88 on an independent test set, outperforming traditional immune scoring indices. Single-cell analysis revealed distinct immune cell subpopulations and their spatial distribution within tumors. The models also identified key gene features associated with immune infiltration at cellular resolution, providing clues for further understanding the immune microenvironment of ovarian cancer.

Conclusion

The machine learning-based approach for evaluating immune infiltration in ovarian cancer at the single-cell level can rapidly and objectively predict the immune status, and discover potential biomarkers and therapeutic targets. This method provides a new strategy and tool for achieving individualized immunotherapy for ovarian cancer.

Clinical trial declaration: This research is not a clinical trial and is exempt from clinical trial registration requirements.