Human behavior is a multifaceted phenomenon that involves the interaction of the mind, the body, and the brain. Wearable and mobile technologies allow the continuous recording of longitudinal data of human behavior on a daily basis. Despite the tremendous opportunities resulting from the large volume of data being collected, the inherent diversity of human behavior still poses significant challenges in quantifying such data. Combined with the limited availability of human experts, computational models developed for conventional, well-established tasks might not work well for detecting and predicting subtle aspects of human behavior. This chapter discusses a new line of research that focuses on population-specific and personalized (PSP) computational models of human behavior, which make decisions by emphasizing the subset of data that is most similar to the subject of interest. Emerging work on signal processing and machine learning algorithms indicates that PSP models can outperform systems learned on the general population and also have advantages in terms of computational efficiency, generalizability, and interpretability. These are discussed in relation to health and well-being applications for promoting the understanding of human behavior and intervening on it.