Scholarlink

<p>Amid the global COVID-19 pandemic, developing effective drugs to combat Coronavirus Disease (COVID-19) has become crucial. However, identifying potential side effects of these drugs remains a significant challenge in the pursuit of effective treatments. Recent advancements in computational models for pharmaceutical development have opened new possibilities for detecting such side effects. In response to the urgent need for effective COVID-19 drugs, this research introduces a novel methodology combining multi-label InceptionV3 and Bidirectional Long Short-Term Memory (IncV3-BLSTM). The experimental evaluations utilize chemical conformers derived from the stick structures of COVID-19 drugs. These conformers’ distinctive features are represented through RGB color channels, with feature extraction performed using InceptionV3, GlobalAveragePooling2D, and BLSTM layers. The results demonstrate the superior efficiency of the IncV3-BLSTM model, outperforming previous studies. Notably, the proposed model achieves a peak accuracy of 97.50% and a co-occurrence potential side effects detection rate of 82.16%. This research marks a significant advancement in modeling drug side effects, particularly for COVID-19 treatments.</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

IncV3-BLSTM: a multi-label inceptionV3-BLSTM model for predicting potential side effects of COVID-19 drugs

  • 期刊论文

  • 发表日期：

    2025年3月31日

查看全文
  • Pranab Das

摘要

Amid the global COVID-19 pandemic, developing effective drugs to combat Coronavirus Disease (COVID-19) has become crucial. However, identifying potential side effects of these drugs remains a significant challenge in the pursuit of effective treatments. Recent advancements in computational models for pharmaceutical development have opened new possibilities for detecting such side effects. In response to the urgent need for effective COVID-19 drugs, this research introduces a novel methodology combining multi-label InceptionV3 and Bidirectional Long Short-Term Memory (IncV3-BLSTM). The experimental evaluations utilize chemical conformers derived from the stick structures of COVID-19 drugs. These conformers’ distinctive features are represented through RGB color channels, with feature extraction performed using InceptionV3, GlobalAveragePooling2D, and BLSTM layers. The results demonstrate the superior efficiency of the IncV3-BLSTM model, outperforming previous studies. Notably, the proposed model achieves a peak accuracy of 97.50% and a co-occurrence potential side effects detection rate of 82.16%. This research marks a significant advancement in modeling drug side effects, particularly for COVID-19 treatments.