Rodrigo Bavaresco1, Jorge Barbosa2, Henrique Vianna3, Paulo Büttenbender4, Lucas Dias5. 1. Applied Computing Graduate Program (PPGCA), University of Vale do Rio dos Sinos (Unisinos), Av. Unisinos, São Leopoldo, Rio Grande do Sul 93.022-000, Brazil. Electronic address: rsimonb@unisinos.br. 2. Applied Computing Graduate Program (PPGCA), University of Vale do Rio dos Sinos (Unisinos), Av. Unisinos, São Leopoldo, Rio Grande do Sul 93.022-000, Brazil. Electronic address: jbarbosa@unisinos.br. 3. Applied Computing Graduate Program (PPGCA), University of Vale do Rio dos Sinos (Unisinos), Av. Unisinos, São Leopoldo, Rio Grande do Sul 93.022-000, Brazil. Electronic address: hdvianna@gmail.com. 4. Applied Computing Graduate Program (PPGCA), University of Vale do Rio dos Sinos (Unisinos), Av. Unisinos, São Leopoldo, Rio Grande do Sul 93.022-000, Brazil. Electronic address: insidy@gmail.com. 5. Applied Computing Graduate Program (PPGCA), University of Vale do Rio dos Sinos (Unisinos), Av. Unisinos, São Leopoldo, Rio Grande do Sul 93.022-000, Brazil. Electronic address: lucaspfsd@gmail.com.
Abstract
BACKGROUND AND OBJECTIVE: Psychotherapy is one of the most common pathways to help individuals address any mental disorders. However, the traditional method of assessing mental health has a margin for improvement. The recent advances in digital technology (e.g., smartphones and wearables) and machine learning techniques can support psychotherapy through the addition of psychophysiology. This paper presents RevitalMe, a context-aware model for assisting a psychotherapeutic understanding of human behavior, providing psychophysiological insights from real-life. METHODS: Five volunteers used RevitalMe's prototype in natural environments for eight days each. Ecological Momentary Assessment was used to collect individuals' stressful states, and to label real-life data. The Wilcoxon Signed-Rank Test was performed to verify a significant difference between the labeled states. Then, RevitalMe classified psychological states based on physiological measurements through machine learning, associating them with the behavior of the individual. After that, visual insights were generated through contexts processing and presented to psychotherapists as evidence of an individual's daily behavior and psychological state. Twelve psychotherapists evaluated the clinical acceptability of RevitalMe, answering six quantitative statements and two qualitative questions. Furthermore, a t-Test was performed to investigate clinical acceptability given therapy field and clinical years. RESULTS: The Wilcoxon Signed-Rank Test succeeds in proving that labeled states were statistically significant, and RevitalMe achieved an F1-Score of 75% in the binary classification of stressed states in natural environments. The evaluation showed clinical acceptability of 90%, composed by partial agreement of 62% and a total agreement of 28%. In this regard, the t-Test provided that the level of interest from cognitive-behavior therapists in psychophysiological insight was higher than that from psychodynamic therapists. CONCLUSIONS: The psychophysiological insights approximate cognitive-behavior psychotherapy to individual's behavior and daily events, focusing on assistance in mental healthcare.
BACKGROUND AND OBJECTIVE: Psychotherapy is one of the most common pathways to help individuals address any mental disorders. However, the traditional method of assessing mental health has a margin for improvement. The recent advances in digital technology (e.g., smartphones and wearables) and machine learning techniques can support psychotherapy through the addition of psychophysiology. This paper presents RevitalMe, a context-aware model for assisting a psychotherapeutic understanding of human behavior, providing psychophysiological insights from real-life. METHODS: Five volunteers used RevitalMe's prototype in natural environments for eight days each. Ecological Momentary Assessment was used to collect individuals' stressful states, and to label real-life data. The Wilcoxon Signed-Rank Test was performed to verify a significant difference between the labeled states. Then, RevitalMe classified psychological states based on physiological measurements through machine learning, associating them with the behavior of the individual. After that, visual insights were generated through contexts processing and presented to psychotherapists as evidence of an individual's daily behavior and psychological state. Twelve psychotherapists evaluated the clinical acceptability of RevitalMe, answering six quantitative statements and two qualitative questions. Furthermore, a t-Test was performed to investigate clinical acceptability given therapy field and clinical years. RESULTS: The Wilcoxon Signed-Rank Test succeeds in proving that labeled states were statistically significant, and RevitalMe achieved an F1-Score of 75% in the binary classification of stressed states in natural environments. The evaluation showed clinical acceptability of 90%, composed by partial agreement of 62% and a total agreement of 28%. In this regard, the t-Test provided that the level of interest from cognitive-behavior therapists in psychophysiological insight was higher than that from psychodynamic therapists. CONCLUSIONS: The psychophysiological insights approximate cognitive-behavior psychotherapy to individual's behavior and daily events, focusing on assistance in mental healthcare.