Type 2 diabetes (T2D) is a chronic disease characterised by a range of dysfunctions in glycaemic regulation. These dysfunctions are known to include insulin resistance (IR), hyper-insulin secretion, hypo-insulin secretion and altered hepatic glucose balance, all in the course of developing the disease. IR, in particular, is a condition in which the circulating insulin is less effective in lowering the glucose levels in blood. Insulin hypersecretion is most associated with pre-diabetes, but can sometimes occur in early diabetes in concert with sufficient IR. Insulin hyposecretion, on the other hand, is most often present in longer-term diabetes and is a result of reduction in β-cell mass. Hence, the ability to accurately monitor and diagnose these stages of progression would offer unique insight and clinical opportunity. As an individual progresses towards T2D, the amount of insulin required to deal with the glucose loads increases. This outcome is ultimately driven by low insulin sensitivity (SI = IR-1). Additionally, T2D is said to have a lower insulin secretion capability, and thus, resulted in consistently increase glucose levels in the blood. Thus, more specifically, precisely observing and understanding the metabolic disorder as changes in both SI and endogenous insulin secretion (UN) may provide further insight into the heterogeneous etiology of type 2 diabetes, and clinical intervention opportunities. Although, several test protocols and mathematical modelling strategies have been developed to quantify these key aspects of T2D, particularly in SI and UN, the goal of this thesis is to find how to effectively improve the precision and clinical utility of these model-based assessments when assessing the SI and UN. This thesis focuses on minimising the identification error or accurately identify the SI value particularly for individual with established T2D. In addition, this thesis also develops and analyses a proportional-derivative (PD) control model that may potentially be able to replace the conventional and accepted methods for estimating the participant-specific UN profile, which are not precise and thus introduce error. In particular, many modelling strategies use fasting glucose (G0) as basal glucose concentration (GB) when assessing the insulin sensitivity. With the assumption of GB = G0, most of the model-based SI assessment able to produce a highly correlated of an SI value to gold standard euglycaemic hyperinsulinaemic clamp (EIC). However, some of the model-based like dynamic insulin sensitivity and secretion test (DISST), was developed in a relatively healthy, normoglycaemic cohort. Thus, the assumption of GB = G0 might be untrue as prior studies have suggested that GB and G0 should be treated differently particularly for T2D individuals. Hence, the outcomes of identifying GB potentially provide accurate assessment of SI value, in particular, for pre-diabetes individual, are investigated and quantified for the first time. It is understandable that UN plays a leading role in glucose homeostasis. Pathological changes in UN can enable early diagnosis of metabolic dysfunction before the emergence of type 2 diabetes. A PD control model that defines UN as a function of glucose concentration is proposed and analysed to provide further insight and modelling capability for this prediabetic state. In addition, it offers the ability to add precision to estimating SI and additional diagnostics around UN. Thus, finally, the proposed PD UN model is further analysed to provide more information in determining each participant’s glycemic condition. The characterised gains of derivative control, 𝜙𝐷 and proportional control, 𝜙𝑃 are used in identifying and discriminating the UN profile for each metabolic state. Hence, the outcome will potentially improve the overall identification of UN profile.