This review systematically outlines research works in the integrated use of neuroimaging and machine learning to predict outcomes from mental disorders, since diagnosing and treating such conditions is very complicated. It puts into perspective neurobiomarker-based predictive models, different approaches to machine learning comprising support vector machines, random forests, and deep learning, and determines the need for early, effective interventions. The chapter reviews state-of-the-art contributions consisting of structural, functional, and diffusion tensor imaging (DTI), in addition to the use of supervised and unsupervised learning methodologies. Key findings include the predictive power of specific neuroimaging modalities and machine learning models with respect to mental health disorders such as schizophrenia, depression, bipolar disorder, and autism spectrum disorder. Concerns regarding interpretability, generalizability, and clinical applicability are discussed in relation to ethical considerations. This review focused on how multimodal neuroimaging, combined with machine learning, enabled improved diagnostic precision and treatment responses to form a basis for recommendations for future research in improving personalized interventions in mental health.