Predictive analytics plays a vital role in risk management by enabling proactive measures for system safety and reliability. When considering possible predictive methodologies for the creation of failure and fault prediction systems in electric motorbikes, Bayesian Networks (BNs) stand out for their ability to model complex and dynamic systems through conditional probability calculations and continuous updates. This paper presents a systematic literature review to explore recent and contemporary methodologies, alongside trends, in Bayesian Networks for fault prediction and prevention, with a focus on electric vehicle applications. Following PRISMA guidelines, 12 studies were analysed to identify core methodologies and peripheral support systems for each of their implementations. Key findings reveal a shift toward dynamic and hybrid Bayesian models, emphasizing real-time fault detection, scalability, and reliability growth through varied approaches. Alongside recent developments and tendencies, accompanying challenges were also identified, including increased complexity and deployment costs. The review also highlights opportunities for more complex integrations between both peripheral systems themselves and core methods, as well as gaps in early deployment strategies for Bayesian Networks in data-limited scenarios.