Abstract:
Over recent years, more and more attention has been paid to Internet- and Mobile-based Interventions for bipolar disorder. Especially the Mobile Health development has had a great impact on this field. Research projects have developed mobile-based systems that used self-assessment or multi-channel data to benefit mood monitoring or mood state recognition in bipolar disorder. Two potentially important factors, however, have not yet been considered within this context: Research and clinical practice indicate the importance of emotional aspects and third-party anamneses for the disease outcome or diagnostic purposes. Therefore, the objectives for this dissertation were the development and evaluation of an emotion-sensitive, technical assistance system for bipolar disorder that accounts for both: It was supposed to analyze the verbal and facial expressions of patients with respect to emotional cues. It was further supposed to allow for ambulatory third-party assessments of close contacts concerning the patients’ mood states. The current work describes the methodological approach and results of four different stages throughout the developmental process. As for the first stage, the attitudes of patients towards technical systems and potential features were examined during an online survey and semi-structured interviews. Within the second stage, the theoretical and the technological bases were worked out in order to develop the system concept. The third stage was a small pilot study with patients examining a prototype of the system in regard to its acceptability and feasibility. Finally, three cases of patients were selected and analyzed separately to establish the clinical value of either the emotion-sensitive module (Case 1 and 2) or the third-party assessments (Case 3). Patients with bipolar disorder showed positive attitudes towards technical assistance and innovative features such as emotion recognition modules (Stage 1). Based on these findings, the first mobile-based system for bipolar disorder has been developed that analyzes emotional expressions of patients and allows for optional third-party assessments (Stage 2). All examined features were well-accepted. The incorporated self-assessments were feasible with a small sample of patients, the emotion recognition was partly feasible, but the third-party assessments were not feasible. In regard to the technical functionality, the analysis of verbal expressions was more dependable than the recognition of facially expressed emotions (Stage 3). Manic symptoms correlated with behavioral changes within usage-related system parameters. With one patient, mood swings were associated with changes of verbally or facially expressed emotions. There were indications of a certain predictive value of expressed emotions or behavioral patterns with respect to occurring mood state changes. Furthermore, self-assessments and third-party assessments were highly correlated (Stage 4). Except for the online survey, the small samples or single case studies limit the generalizability of the results and a selection bias through a high technical affinity of participants cannot be excluded. Additionally, the exploratory approach of the pilot study and the case studies increases the risk of alpha inflation. More data is needed to generate substantial findings regarding the feasibility and validity of system features. However, the results indicate a good acceptability and a moderate feasibility of the emotion-sensitive module. Individual differences within the perception and usage of the features call for flexible systems, which allow to activate or deactivate modules. The results further support the individuality of bipolar symptoms and ramifications and they suggest a diagnostic and prognostic value of emotion recognition. More research is necessary to confirm these findings and to examine the third-party assessments. Yet, in the future, the long-term monitoring of emotional expressions in bipolar disorder may be an acceptable and feasible feature and might possibly be helpful to diagnostic, prognostic or even therapeutic purposes.