The curative potential of herbal medicine on diabetes and diabetic retinopathy : a pharmacological study with machine learning intelligence

Abstract

Diabetes is a complicated disease composed of metabolic disorders mainly caused by hyperglycaemia. Diabetic retinopathy (DR) is one of the pivotal complications of diabetic mellitus, which is mainly resulted from long-term hyperglycemia. DR progression can ultimately lead to the degeneration of retinal neurons and vasculatures. Unfortunately, although conventional means, including laser photocoagulation and vitrectomy surgery, have exhibited anti-DR effects, the therapeutic effect is limited. Accumulating studies have reported therapeutic effects of herbal medicines on diabetic retinopathy. The US FDA hold a positive attitude on the approval of the clinical application of medicinal herbs. Herbal medicines may be the potential candidate for treating diabetic retinopathy. Ginger-Coptis formula (GCF) is a herbal formula composed of Zingiber officinale (Ginger) and Coptidis Rhizoma (Coptis), which was documented to improve high blood glucose and disorder of lipid metabolism. Our study may be the first attempt to clarify the therapeutic effect of GCF and its bioactivities on diabetic retinopathy. Therefore, the first aim of this study is to identify an effective anti-DR herbal medicine.

Due to the multi-components and multi-targets nature of Chinese Medicine, conventional approaches may be time-consuming and use abundant manpower to individually analysis the bioactive compounds in a single herb or herbal formulae. Therefore, it is necessary to establish a method that can identify the contributive compounds in herbs in an efficient way. Given the interdisciplinary scope of life science and mathematics, the interaction between chemical profile and biological activity may be measured via the combination of machine learning analysis and experimental validation. The machine learning-related analysis may enable us to predict and validate the critical bioactive component efficiently and accurately. Hereby, as the second aim of this study, we would like to establish a feasible method that can disclose the contributive components in a herb or herbal formulae for treating diseases. The anti-DR compounds in ginger-coptis formula will be identified by the machine learning approaches and experimental validation.

In this study, using systematical experimental validation and R-language based machine-learning approaches, we made the first attempt to establish a novel “Mathematic-Biological Evaluation” approach for identifying the key contributive compounds in ginger-coptis formula for treating DR, in which 6-shogaol in GCF is potentially the most contributive ingredient. 6-shogaol is mainly effective in 3 therapeutic aspects, including the reverse of hyperglycaemia, the degeneration of retinal vasculature, and neuronopathy.