Scholarlink

<p>Given the challenges of personalized depression treatment, such as individual differences, diverse medications, long treatment cycles, and adverse interactions. We proposed a self-adaptive data-driven evolutionary algorithm based on random forest feature selection and incremental Gaussian process regression (SADDEA-RFFS-IGPR). The algorithm integrates feature selection, surrogate modeling, and adaptive adjustment. We evaluated SADDEA-RFFS-IGPR on samples from three depression subtypes through model benchmark testing, comparison with same type algorithms, and ablation experiments. Results showed it achieved the best surrogate performance, an average ranking below 1.1 and statistically significant improvements (<i>p-value</i> &lt; 0.05). Performance declined when any component was removed. In addition, analysis of variance (ANOVA) identified key factors. These findings confirmed the algorithm’s effectiveness, and potential in optimizing personalized treatment strategies.</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

Self-adaptive data-driven evolutionary algorithm based on random forest feature selection and incremental Gaussian process regression on personalized antidepressant medication research

  • 期刊论文

  • 发表日期：

    2025年7月5日

查看全文
  • Ruxin Zhao,
  • Hongtan Zhang,
  • Chang Liu,
  • Yulin Xie,
  • Yue Cao,
  • Yang Shi

摘要

Given the challenges of personalized depression treatment, such as individual differences, diverse medications, long treatment cycles, and adverse interactions. We proposed a self-adaptive data-driven evolutionary algorithm based on random forest feature selection and incremental Gaussian process regression (SADDEA-RFFS-IGPR). The algorithm integrates feature selection, surrogate modeling, and adaptive adjustment. We evaluated SADDEA-RFFS-IGPR on samples from three depression subtypes through model benchmark testing, comparison with same type algorithms, and ablation experiments. Results showed it achieved the best surrogate performance, an average ranking below 1.1 and statistically significant improvements (p-value < 0.05). Performance declined when any component was removed. In addition, analysis of variance (ANOVA) identified key factors. These findings confirmed the algorithm’s effectiveness, and potential in optimizing personalized treatment strategies.