Susana Arias Tapia, Rafael Martinez-Tomás, Héctor F Gómez Alvardo, Jose Barbosa Corbacho, Azizuddin Khan, Adriana Reinoso and Victor Hernandez del Salto
Universidad Técnica Particular de Loja-Ecuador, Departamento de Ciencias de la Computación y Electrónica, San Cayetano S/N, 1101608
Introduction: In order better understand the power of emotions on a thinking brain, we need to consider the way in which the brain has evolved. The fact that the concept of an emotional brain is newer than the concept of a rational brain clearly reveals the relationship between thought and feelings. When we are trapped by desire or anger, we are under the influence of the limbic system, which points to two basic elements: learning and memory. Emotions can be processed in the brain, which helps to determine whether it is being affected by any illnesses. These behavioral changes refer to a series of symptoms that frequently occur in patients, which reflect changes in behavior, and personality changes in family members or other people such as: depression, anxiety, mania, exaltation or euphoria, hallucinations, delirium, abandonment, wandering, repeated questions, false recognition and sleep disturbance are the most common symptoms of patient symptoms, which affect their personality. In addition, the patients become inexpressive, and just provide pat answers . In dementia, for example, apathy is a very important symptom –an early indicator- since this is something that limits patients' well-being, that is, it does not matter if they eat, shower, or get out of bed. Dementia is also noted in the capacity for retaining information when there is no guidance by others, or when the patient knows that something is happening to them and is losing control, thus creating anxiety . Our work proposes a methodology to try to identify the early development of the illness. We believe that by analyzing the emotional changes in patients in both video technology and writing samples, we can identify the origins of mental illnesses. Thus, when there is a correlation between polarity, text and emotional semantic orientation, we can determine whether the patient is suffering from abnormalities.
Keyworks: The analysis of writing in patients has been analyzed in many research studies. In fact, one can analyze the way in which patients write since the mechanized system of writing is often affected in these types of patients [3, 4]. The idea is to discover the exact point of correlation with the aim of discovering alterations in memory in a correct way . It has been possible to develop conversations with various types of dementia patients. This provides our hypothesis with meaning seeing that we believe that there are individual characteristics of a patient's writing, which are different in polarity and semantic orientation in relation to other patients . Our proposal is based on supervised methods where we use dictionaries that help to assign polarity and emotion to a text. In addition, we propose that this classification may be corroborated by cluster methods to establish which emotion is experienced first in a text. It is from here that we get the importance of polarity and its analysis for our study in which we try to identify whether there exists a characteristic that relates to the polarity with the emotion in these types of patients. Text mining has been applied to extract useful information to detect dementia in non-structured texts. It has been applied in psychopathology, patient perspectives, registers and medical literature. These results support our work as we have used text mining algorithms (supervised training) to identify the lack of correlation between variables such as polarity and emotion. In this second section we describe our proposal, which commences with an analysis of the corpus and ends with the polarity, and its relation with emotion.
Test Methodology: Our job is to see whether there exists a lack of correlation between polarity and emotion. The patient can display certain mental problems. Furthermore, when this technique is used with old age pensioners, our methodology would help to determine a possible advancement of the illness. In the following diagram, we have the following proposals:
Figure 1: Methodology for polarity and emotion.
Current Work in the Field: By using audio transcriptions from the University of North Carolina, we were able to comprise a group with a simple of 23 patients and another 23 normal patients (young people), i.e. where we applied the analysis of text analysis of emotions using the programming language "Python" and the Rose guide (Emotion Words), which enables us to extract emotions from a text, and its polarity. The results have various significant differences for the recognition of facial expressions among the group of patients with Dementia, and the group of normal patients. The emotions that are most common in a person with dementia are: sadness, happiness, anger, fear and their various combinations are affected beginning with slight or minor displays of these emotions. When the recognition of fear is altered, there are moderate states of the illness, as well as in meaning of the polarity according to the experimentation carried out for the recognition of facial emotions, which produce variables in various conversations - specifically in studies carried out for the recognition of facial identity. This effect is such that some studies support the hypothesis that the mechanisms responsible for the discrimination of facial identity reaches the point of being suppressed, i.e. in which the patients can recognize human faces but not their facial expressions. In graph 1, we can see the sample of 23 audio recordings where the emotions in these types of patients always remain visible. The patients showed alterations in the recognition of sadness (40,48%), fear (20,24%). Moreover, our study coincides with other results found by other authors, which found greater alterations in emotions such as sadness and fear. There also existed the same emotions of happiness in some other contexts (23,1%). The results confirm that the patients with dementia display alterations in the processing of emotions, which are found more in negative emotions such as those previously mentioned.
Graph 1: Detailed Graph of the analysis and recognition of emotions in Patients.
Conclusion: The initial results show that it is possible to identify a lack of correlation between polarity and emotion. We are currently experimenting with 20 more patients, among which we have included Alzheimer disease. In addition, we have included the Kappa Index with the aim of obtaining an adequate scorecard for the training and validity of the results.
 C. Jung, Psychology of Dementia Praecox, Princenton University, 2015.
 A. Ford y O. Almeida,«Phsychological treatment for depression and anxiety associeted with dementia and mild cognitive impairment,» The British Journal of Pshychiatry, vol. 4, pp. 286-287, 2015.
 M. Slavin, J. Phillips, J. Bradshaw, K. A. Hall y I. Presnell, «Consistency of handwriting movements in dementia of the Alzheimer's type: A comparison with Huntington's and Parkinson's diseases,» Journal of the International Neuropsychological, vol. 5, no. 1, pp. 20-25, 2000.
 E. Onofri, M. Mercuri, T. Archer, R. Max Rap, F. Massoni y S. Ricci, «Effect of cognitive fluctuation of handwriting in alzheimer's patient: a case study,» Acta Medica Mediterranea, vol. 31, no. 751, 2015.
 M. Reuber, «Using Conversation Analysis in the memory clinic: a pilot study,» Royal Hallanshire Hospital, England, 2014.
 D. Blackburn y M. Reuber, «Using conversation analysis to help diagnose dementia,» Aging, Dementia, Cognitive, and Behavioral Neurology ePosters, no. 14, p. 84, 2015.