Machine learning (ML) is rapidly transforming the field of healthcare, thanks to the advancements in digitization of medical data. However, progress has been limited in neurological applications because of, a) the complex physiology of the brain and b) the lack of large, labeled datasets. This is particularly emphasized in complex diseases such as epilepsy and Alzheimer’s disease, where disease mechanisms are still actively investigated. In my thesis, I developed several clinical decision support tools in collaboration with the Mayo Clinic, Rochester, to assist neurologists with diagnostic and treatment decisions related to epilepsy and Alzheimer’s disease, all using a novel ML concept named domain-guided ML. The key idea of domain-guided ML is to utilize domain knowledge and labeled data as complementary means of supervision for designing and training machine learning models, particularly when labeled supervision is limited. In this talk, I will present some specific applications of domain-guided ML in neurology (e.g., localization of epileptic brain regions and diagnosis of cognitive decline and epilepsy) and some of my recent work on domain-guided self-supervised learning and natural language processing.