Abstract: Skin diseases affect people's health and quality of life because of their high incidence, difficult diagnosis and apparent harm, coupled with insufficient medical resources. In recent years, with the development of computer-aided diagnosis (CAD) technology, single-modality CAD approaches have broken the limitations of traditional methods, such as strong subjectivity, and high missed-diagnosis and misdiagnosis rate, but failed to leverage the multi-modal information in real clinical scenarios. Multi-modality CAD methods help artificial intelligence models learn the clinical representations in a more complex and comprehensive manner, aiding dermatologists in making a more accurate diagnosis of skin diseases. This article introduces different types of skin lesion data commonly used in CAD methods, summarizes the single-modality/multi-modality methods based on related works in the field of CAD systems of skin diseases, and predicts possible future development trends of CAD technology, thus providing insights for mitigating the challenge on the diagnosis of skin diseases.