The Mind-It Corpus: a Longitudinal Corpus of Electronic Messages Written by Older Adults with Incipient Alzheimer’s Disease and Clinically Normal Volunteers

In this article, we present the Mind-It project and the corpus we are currently collecting. The long-term aim of the project is to contribute to the preclinical detection of Alzheimer’s disease (AD) by developing a computer model that searches for linguistic changes that mark AD. To this end, we will automatically analyze the history of electronic messages, such as those communicated via WhatsApp, Messenger and e-mails, of clinically normal participants and AD patients. The literature about the automatic detection of AD using linguistic input has shown that productions from AD patients are automatically distinguishable from productions of normal older adults. Furthermore, case studies about authors who developed AD themselves suggest that their writing style progressively changes as a result of the disease. With respect to existing corpora containing linguistic materials from AD patients, the data that we collect will form a unique corpus; we are not aware of other resources featuring such longitudinal data. In this article, we argue how our project will contribute to the research on AD and discuss our considerations on collecting, processing and sharing the project’s data. We also speculate how the data could be used to develop an automated tool for preclinical detection of AD