In recent years, many cognitive care robots have been developed. There is also a form of the companion pet, and there are also robots that can be personified and talk with the elderly. Also, some telecommunications equipment can be replaced to help older people connect to the telecommunications community. However, the need for such robots is still unclear in the focus of this area of being able to interact, talk to, and connect with family and society. Although many studies in the existing literature that tell how to research to verify their efficacy, methods of defining effects can pave the way for future studies. This is what this article will show. The article included qualitative research searches and screenings, effect summaries, and research method analysis. The problem to be solved in this article is how to use and measure the effectiveness of conversational humanoids in nursing elderly patients with cognitive impairment. We also hope to provide some guidance for future research on cognitive impairment.