Can artificial intelligence bring out the common mechanisms that affect the Quality of Live of anxiety-depressive patients?

TRAIBER, L.; PASCARIELLO, G. O.; AGUSTÍN IBAÑEZ; PEREZ DEL CERRO, J.; DELEGLISE, A.; LUCAS SEDEÑO; YORIS, A.; DONNELLY KEHOE, P.; TORRENTE, F.

Dresden

Congreso; First official European Congress on Clinical Psychology and Psychological Treatment of EACLIPT; 2019

The European Association of Clinical Psychology and Psychological Treatment

A new issue among current mental health challenges is the high comorbidity between anxiety disorders and depression -/considered together as negative affectivity. This complexity is unattainable for most psychological treatments, designed for mutually excluded diagnostic categories. Additionally, it is proposed that high comorbidity could be explained by the existence of mysterious transdiagnostic dimensions at different levels (behavioral, cognitive, neuropsychological, and neurophysiological) that are common to these disorders. High comorbidity negatively impacts the patients? Quality of Life (Q-L) as well as its severity. Q-L is defined as patients? satisfaction with general and specific areas of their own life, comprising work, health, interpersonal relationships, leisure activities, among others (Endicott et al., 1993). Particularly, anxiety disorders (Koran, Thienemann, & Davenport, 1996; Rubin et al, 2000; Wittchen & Belochy, 1996) and depression (Pyne et al, 1997) are associated with a significant deterioration in Q-L and functioning, compared to control groups. Regarding depression, some studies reported greater Q-L deterioration than healthy controls (Sherbourne, Wells & Judds, 1996; Olfson et al., 1997; Schonfeld, Verboncoeur, & Fifer, 1997). Among negative affectivity, depressives? Q-L was similarly to PTSD but more impaired than anxiety disorders (panic, agoraphobia, social anxiety, OCD, GAD) (Rapaport et al., 2005; Demyttenaere et al., 2008). Consequently, it is feasible that Q-L in depression and comorbid anxiety disorders are further impaired than anxiety disorders without comorbidity. However, few is known about how Q-L interacts with symptoms, core cognitions, and specific diagnosis inside this group of patients. In order to reveal that it is important to know what is more relevant in Q-L deterioration: if the comorbidity or severity; ii) if any symptom has more power to?.; or iii)By combining machine learning approaches, we developed a nontraditional statistical analysis to understand how Q-L interacts with different-level dimensions of anxiety-depression psychopathology. Five hundred anxiety-depressive outpatients admitted at the Anxiety and Trauma Clinic between 2014 and 2017 completed 15 clinical questionnaires about symptoms and core cognition: 1) PDSS-SR: Panic Disorder Severity Scale ? Self Report; 2) ASI-3: Anxiety Sensitivity Index; 3) BSQ: Body Sensations Questionnaire; 4) ACQ: Agoraphobic Cognitions Questionnaire; 5) SIAS: Social Interaction Anxiety Scale; 6) OCI-R: Obsessive-Compulsive Inventory-Revised; 7) OBQ-20: Obsessive Beliefs Questionnaire; 8) PSWQ: Penn State Worry Questionnaire; 9) WW: Why Worry Questionnaire; 10) HADS: Hospital Anxiety and Depression Scales; 11) FMPS: Frost Multidimensional Perfectionism Scale; 12) PSS-SR Posttraumatic Stress Disorder Scale- Self-Report Version; 13) PTCI: Posttraumatic Cognitions Inventory; and 14) CDS: Cambridge Depersonalization Scale; and 15) Q-LES-Q-SF: Quality of Life Enjoyment and Satisfaction Questionnaire Short Form. all participants were grouped into 3 subgroups: high, low, and impaired Q-L, based in their scores in Q-LES-Q-SF questionnaire and previous classification reports [Gao, Su, Sweet, Calabrese, 2018; Rapaport et al., 2005]. We obtained three pared groups in sociodemographic variables. Three hundred ninety-three participants integrated the final database used for machine learning analysis. Then we use them to develop a Multi Classifier System (MCS). We compare the MCS against each diagnostic. We repeated this comparison using three state-of-the-art classification algorithms. Throughout different machine learning analyzes, we predicted the associations between the three levels of impairment of the Q-L with the symptoms and cognitions reported by the patients, concluding in the existence of a pattern of poor Q-L and high comorbidity in all participants. These interactions could be 'learned' by machine learning to predict treatment outcomes or to guide new transdiagnostic treatments.