This bibliometric study analyses the development and research trends in control systems for robotic prostheses using artificial neural networks and electromyographic (EMG) signals. From an advanced search in the Scopus database, 222 relevant documents were collected and analysed using the Bibliometrix and VOSviewer tools. The results reveal a growing body of scientific literature on the integration of artificial intelligence into the myoelectric control of prostheses, highlighting key terms such as electromyography, machine learning, and prosthetics. Additionally, thematic clusters, emerging trends, and documents with high bibliographic impact were identified. The most productive sources include specialised journals in neuroengineering, rehabilitation, and robotics. This work provides an overview of the current state of the field, enabling future research to be directed towards the design of more precise, adaptive, and accessible control systems for users with amputations. Additionally, experimental tests were conducted with five healthy volunteers (20–32 years old), who performed three main gestures (opening, closing, and pinch) in six 5-s repetitions, with equivalent rest intervals. The system, based on an Arduino UNO, an EMG882 sensor, and a PCA9685 driver, achieved an average gesture detection accuracy of 87%, with an inter-subject variance of ±5%. These results validate the technical feasibility of the prototype and provide a solid initial foundation for future developments aimed at optimising hardware, improving classification algorithms, and integrating advanced myoelectric control strategies.