Scholarlink

In many regions of the world, malaria is still a serious public health concern that requires prompt and precise diagnosis in order to be effectively treated. The generalization capacities of AI-based diagnosis models trained on thin blood smear pictures are assessed in this work. To evaluate their suitability for use in various clinical settings, a variety of datasets and AI methodologies were examined. Important issues were noted, including model robustness, dataset diversity, and site-specific biases. In order to provide greater clinical utility, the suggested strategies—which include incremental learning and transfer learning—strive to enhance these models’ generalization performance.

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

Assessing Generalization Capabilities of AI Models for Malaria Diagnosis Using Blood Smear Images

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Md. Sakibur Rahman,
  • Fahima Afrin Nidha,
  • Al Samiul Haque Rabbi,
  • Md. Tanvir Islam Mahin,
  • Sifat Khan Shishir,
  • Md. Parvezur Rahman Mahin,
  • K. M. Safin Kamal,
  • Ahmed Wasif Reza

摘要

In many regions of the world, malaria is still a serious public health concern that requires prompt and precise diagnosis in order to be effectively treated. The generalization capacities of AI-based diagnosis models trained on thin blood smear pictures are assessed in this work. To evaluate their suitability for use in various clinical settings, a variety of datasets and AI methodologies were examined. Important issues were noted, including model robustness, dataset diversity, and site-specific biases. In order to provide greater clinical utility, the suggested strategies—which include incremental learning and transfer learning—strive to enhance these models’ generalization performance.