Scholarlink

<p>A critical challenge in personalized medicine is identifying the optimal drug treatment for individual patients based on their unique biological profiles. Recent advancements have produced a surge in drug prioritization algorithms using patient omics data, yet few undergo rigorous in vivo testing. There is a need for an efficient, objective framework to determine the smallest experimental cohort capable of producing statistically robust results. OptiRanker, which offers a statistical simulation framework that perturbs algorithmic predictions with controlled noise and evaluates their ability to preserve true performance ranking using minimal cohorts, aims to address that issue. OptiRanker demonstrated that through weighted mean squared error (WMSE) and Spearman correlation against baseline rankings, accurate predictor rankings can be recovered across the simulated conditions tested while substantially reducing experimental scale. In in silico validation using a fixed evaluation set of 36 drug IC50s and 798 cell-line models and 3 published algorithms, the full predictor ranking was recovered with as few as six individuals and one drug for the dataset used. Overall, OptiRanker offers a reproducible, exploratory approach to optimize in vivo validation trials of drug prioritization algorithms. By focusing on the smallest statistically robust experimental designs, it addresses a key bottleneck in translating computational models into clinically actionable tools. The Python code is available at: <a href="https://github.com/OhadLandau/OptiRanker">https://github.com/OhadLandau/OptiRanker</a>.</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

OptiRanker: a simulation and optimization framework for efficient in vivo validation of drug prioritization algorithms

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年7月15日

查看全文
  • Ohad Landau,
  • Kartheeswaran Thangathurai,
  • Shai Magidi,
  • Angel Porgador,
  • Eitan Rubin

摘要

A critical challenge in personalized medicine is identifying the optimal drug treatment for individual patients based on their unique biological profiles. Recent advancements have produced a surge in drug prioritization algorithms using patient omics data, yet few undergo rigorous in vivo testing. There is a need for an efficient, objective framework to determine the smallest experimental cohort capable of producing statistically robust results. OptiRanker, which offers a statistical simulation framework that perturbs algorithmic predictions with controlled noise and evaluates their ability to preserve true performance ranking using minimal cohorts, aims to address that issue. OptiRanker demonstrated that through weighted mean squared error (WMSE) and Spearman correlation against baseline rankings, accurate predictor rankings can be recovered across the simulated conditions tested while substantially reducing experimental scale. In in silico validation using a fixed evaluation set of 36 drug IC50s and 798 cell-line models and 3 published algorithms, the full predictor ranking was recovered with as few as six individuals and one drug for the dataset used. Overall, OptiRanker offers a reproducible, exploratory approach to optimize in vivo validation trials of drug prioritization algorithms. By focusing on the smallest statistically robust experimental designs, it addresses a key bottleneck in translating computational models into clinically actionable tools. The Python code is available at: https://github.com/OhadLandau/OptiRanker.