Internet of Things (IoT) technologies are now becoming available in various complex and connected systems including vehicles. While IoT technologies result in high-dimensional data environments that provide an unprecedented opportunity for smart condition monitoring and predictive analytics of complex systems; they also pose critical challenges. Data for each complex system are collected from multiple streams that are often correlated and each data stream contains only partial information about the degradation process of the system. Second, data from one complex system may not be sufficient to extrapolate its degradation profile and it is critical to transfer knowledge between different connected systems. In this talk, we present a set of statistical and deep learning techniques for smart and connected systems to address challenges related to warranty data analytics, remaining useful life estimation, and safety of artificial intelligence systems.