Optical FRET biosensors allow for real-time biomarker detection for healthcare applications. In this thesis, I discuss the development of real-time FRET sensors for insulin (Project I) and calcium (Project II) for applications in diabetes management and improved clinical practices, respectively. Both biosensors are based on the structures of native proteins that interact specifically with either insulin or calcium in real-time and undergo conformational changes. These conformational changes are harnessed with fluorescent proteins and small molecules to generate a FRET response. For the insulin biosensor (Project I), the sensor is modelled on insulin binding domains for the human insulin receptor (InR), whereas the calcium sensor is developed from an existing troponin C-based FRET biosensor, Twitch-2B. Using a combination of rational design, high diversity random mutagenesis phage display libraries and a high-throughput FRET screening platform, libraries of InR variants are screened to identify a potential insulin sensor with specific dose-dependent insulin binding and FRET response. Project II involves the optimization of the Twitch-2B FRET calcium sensor for integration in implantable devices. Towards this goal, we develop a FRET construct using SNAP- and CLIP-tags that can be stably expressed and is not prone to photobleaching.