Alzheimer's disease (AD) is a neurological disorder with complicated pathogenesis. The approved AD drugs cannot block or reverse the pathologic progression of AD. In this study, a method based on Logistic Matrix Factorization and Similarity Network Fusion (MLMFSNF) is proposed for screening out the Traditional Chinese medicines (TCMs) and active ingredients targeting AD targets. Firstly, TCMs for AD are obtained from the AD drug reviews, the active ingredients and related targets are collected from various databases. Secondly, the similarity networks are constructed by an improved Gaussian interaction profile kernel and other metrics for active ingredients and targets. The synthesized similarity networks are integrated based on similarity network fusion (SNF). The filling of missing activity ingredient-target associations is achieved by the logistic matrix factorization. Finally, the association scores between active ingredients and targets are calculated and ranked. We screen out TCMs for AD by the logistic function transformation. The results demonstrated that the MLMFSNF algorithm is effective for association prediction.

中文翻译：

---

基于Logistic矩阵分解和相似性网络融合的AD中医研究


阿尔茨海默病 （AD） 是一种发病机制复杂的神经系统疾病。批准的 AD 药物不能阻断或逆转 AD 的病理进展。本研究提出了一种基于 Logistic 矩阵分解和相似性网络融合 （MLMFSNF） 的方法，用于筛选出针对 AD 靶点的中药 （TCM） 和活性成分。首先，从 AD 药物综述中获得治疗 AD 的中药，从各种数据库中收集活性成分和相关靶点。其次，通过改进的高斯交互剖面核和活性成分和靶标的其他指标构建相似性网络。合成的相似性网络基于相似性网络融合 （SNF） 进行集成。缺失活性成分-目标关联的填充是通过 logistic matrix factorization 实现的。最后，计算活性成分与靶点之间的关联分数并对其进行排序。我们通过 logistic 函数转换筛选出 AD 的中医。结果表明，MLMFSNF 算法对关联预测是有效的。