Innovations in hardware and control have created a new class of orthoses or exoskeletons to augment and assist human movement. These designs enable forces and torques to be applied to nearly any segment of the body. However, predicting how an individual will adapt their movement in response to external assistance remains incredibly challenging. These predictions are challenging for unimpaired individuals, let alone for individuals after neurologic injury, such as in cerebral palsy or stroke. Even for a “simple” case such as single degree-of-freedom ankle foot orthoses, we often fail to predict how a given device will alter or improve an individual’s movement. In this seminar, we will discuss these challenges as well as new methods that may assist in optimizing orthoses after neurologic injury. In particular, I will discuss how machine learning can help us to learn from past prescriptions, while also using musculoskeletal modeling, muscle synergy analysis, and ultrasound imaging to quantify neuromuscular adaptations and inform orthotic design.