Drug classification plays a crucial role in contemporary drug discovery, design, and development. Determining the Anatomical Therapeutic Chemical (ATC) classes for new drugs is a laborious, costly, and intricate process, often requiring multiple clinical trial phases. Computational models offer significant benefits by accelerating drug evaluation, reducing complexity, and lowering costs; however, challenges persist in the drug classification system. To address this, a literature survey of computational models used for predicting ATC classes was conducted, covering research from 2008 to 2024. This study reviews numerous research articles on drug classification, focusing on drug descriptors, data sources, tasks, computational methods, model performance, and challenges in predicting ATC classes. It also examines the evolution of computational techniques and their application in identifying ATC classes. Finally, the study highlights open problems and research gaps, suggesting areas for further investigation in ATC class prediction. CCS Concepts: Applied computing → Life and medical sciences → Bioinformatics

中文翻译：

---

预测药物解剖治疗化学类别研究的综合调查


药物分类在当代药物发现、设计和开发中起着至关重要的作用。确定新药的解剖治疗化学 （ATC） 类别是一个费力、昂贵且复杂的过程，通常需要多个临床试验阶段。计算模型通过加速药物评估、降低复杂性和降低成本来提供显著的好处;然而，药物分类系统仍然存在挑战。为了解决这个问题，对用于预测 ATC 类别的计算模型进行了文献调查，涵盖了 2008 年至 2024 年的研究。本研究回顾了大量关于药物分类的研究文章，重点关注药物描述符、数据源、任务、计算方法、模型性能以及预测 ATC 类别的挑战。它还研究了计算技术的演变及其在识别 ATC 类中的应用。最后，该研究强调了开放的问题和研究空白，提出了 ATC 类别预测的进一步研究领域。CCS 概念：生命科学和医学→应用计算 → 生物信息学