For Type 2 diabetes, herbal medicine has been considered a helpful tool in preventing and treating the disease. Patients can be treated with precision by personalised herbal medicines. However, challenges exist of presenting herbal efficacies with scientific evidence using modern techniques. This thesis proposes a strategy for herbal medicines’ personalisation and recommendations, in order to build a bridge for the development of Type 2 diabetes treatments between Eastern and Western countries. The strategy is inspired by traditional Chinese theories of patient characteristic and syndrome. To identify these traditional concepts, biological analysis is presented using computer science approaches. Three groups of Type 2 diabetes patients are identified according to their likelihood of developing Type 2 diabetes. Accordingly, three herbal medicine products for the identified groups of patients are regularly selected by experts. Biological analysis of drug-target interactions is investigated for each group of patients, the metabolic processes and the involved pathways are analysed in depth. This thorough analysis may help us get insights into the mechanisms of the development of the disease. Comparisons of the three drug-target interaction networks are also made and I focused on differences among the networks. The differences among the biotargets of the herbal medicines may indicate potential mechanism associated to the incidence of Type 2 diabetes .

From my studies, five metabolic genes associated to the herbal formulae are identified to be the priority genes that affect the incidence of Type 2 diabetes. They are HMGC, ALDH2, ALDH7A1, HK1 and ACSL1. Single-nucleotide polymorphisms that are associated to Type 2 diabetes from the five targets are demonstrated to make local healthcare tests implementable. Meanwhile, differences of biotarget specifications and pathways that are targeted among the three herbal medicines may indicate the sub-mechanism for curing Type 2 diabetes. Consequently, personalising Type 2 diabetes treatment may be necessary. After this theoretical analysis, an open sourced, clinical data is identified. Hierarchical agglomerative clustering algorithm is applied to find the patterns for mRNA performances and target associations. The cluster result shows that the genetic measurement of the five targets is differentiable among the 24 samples. This differentiation may be able to further indicate that the different sub-mechanisms for Type 2 diabetes incidence may exist, and treatments for preventing and curing Type 2 diabetes should be designed individually.

To enable a precise treatment for Type 2 diabetes and an individual supervision by healthcare recommendations, this thesis demonstrates a comprehensive system, which directly applies the lab-based analysis to a real application. The system is designed to map herbal medicine products onto individual biological specifications, gaining the information about the herbal product by scanning the QR code on the products. By uploading individual biological data into a personal mobile device, output of herbal product recommendations is automatically generated. A wireless sensor network is illustrated for the recommendation communications. The network can help patients, doctors, and hospitals in using, sharing, and securing the data.