With the global population increasing and the demographic shifting toward an aging society, the number of patients diagnosed with conditions such as peripheral neuropathies resulting from diabetes is expected to rise significantly. This growing health burden has emphasized the need for innovative solutions, such as brain-computer interfaces. brain-computer interfaces, a multidisciplinary field that integrates neuroscience, engineering, and computer science, enable direct communication between the human brain and external devices. In this study, we developed an autonomous diabetes therapeutic system that employs visually-induced electroencephalography devices to capture and decode event-related potentials using machine learning techniques. We present the visually-induced optogenetically-engineered system for therapeutic expression regulation (VISITER), which generates diverse output commands to control illumination durations. This system regulates insulin expression through optogenetically-engineered cells, achieving blood glucose homeostasis in mice. Our results demonstrate that VISITER effectively and precisely modulates therapeutic protein expression in mammalian cells, facilitating the rapid restoration of blood glucose homeostasis in diabetic mice. These findings underscore the potential for diabetic patients to manage insulin levels autonomously by focusing on target images, paving the way for a more self-directed approach to blood glucose control.