The Dissociation Between Polarity and Emotional Tone as an Early Indicator of Cognitive Impairment: Second Round

Motivation: Obtaining mechanisms that allow identification of Alzheimer’s disease early is the subject of analysis by many researchers. The purpose is to obtain an early classifier that identifies Alzheimer’s disease, and thus contribute to improving the patient’s quality of life by applying appropriate therapies derived from early diagnosis. This work has the title of Second Round because it is the continuation of our previous results. Objective: To work with free conversations, to detect if polarity and tonality can be used to classify the phrases of those conversations and differentiate patients with Alzheimer’s. Methodology: Data from Charlotte and free interviews of patients with Alzheimer’s were used to calculate their correlation and thus determine the disconnection between the variables and the classification of Alzheimer’s patients. Results: 407 phrases from Charlotte and 432 phrases from Alzheimer’s were used in this study. A negative correlation showed the disconnection of the variables. It was more evident in Alzheimer’s than in Charlotte. The Bayes Net algorithm managed to classify Alzheimer’s with 84% F measure while J48 achieved 76% of this measure, with a Cross validation of 10 Folds, confirming our proposals described in previous works, for different conversations in this new study. Obtaining the mechanisms for the identification of Alzheimer’s disease is an object of analysis by many researchers. The point is to obtain an early classifier that identifies Alzheimer’s disease, to help improve the quality of life of patients and their families, by applying the appropriate therapies derived from early diagnosis. This work has the title of Second Round because it is the continuity of our previous results. However, these results cannot yet be defined as conclusive in their entirety, as they generate new questions and doubts exposed in the conclusions and future work sections.