Personalised ambient monitoring for people with bipolar disorder

Abstract

Objective: The main objective of the work was to design and develop a support platform for people with bipolar disorder. Currently, monitoring the state of such patients is achieved mainly via the use of interviews and structured questionnaires. The proposed plat- form would utilise sensors and pervasive technology to provide patients and their caregivers with unbiased information about their behavioural patterns. The central hypothesis is that clearly identifiable changes in such patterns might be indicative of an upcoming episode and so could be used to issue an early warning signal. This could also trigger actions that may help prevent the patient undertaking actions with serious consequences during the occurrence of the episode. Methods: The system uses a mobile phone as a medium for both collecting sensory data and providing a means of interacting with the user. The collected information is then processed to extract features considered likely to be affectedbythecourseofbipolardisorder.Thedesign includes a set of wearable sensors monitoring the user's position, social interactions (via sensing the Bluetooth context) and psychomotor movements (via accelerometers) as well as sensing ambient environmental features like the noise and light levels and usage of household appliances. To verify the capability of the system to detect behavioural changes, it was trialled in several configurations on users with no history of mental conditions as well as a small group of bipolar disorder patients. Results: It was found that the proposed system provided sufficient information about behaviour and habits to detect major changes. In addition the trials were valuable in giving insight into the potential users' views on the usability and acceptability of the system. The most significant finding was the likelihood that a system requiring a large a number of sensors and devices incorporated into the everyday routine may cause usability issues and hence result in the user rejecting the system altogether. Therefore, the possibility of realising a solutionbasedsolelyuponamobile phone was also explored considering that most people now carry such a device at all times. Applying appropriate processing techniques to phone-harvested data renders it possible to identify the number and significance of visited locations and (to some extent) the intensity of interactions with other individuals, both of these being likely to be highlyaffectedbythecourseofbipolardisorder. Conclusion: Pervasive mobile computing technologies provide tools with the potential to enhance the process of managing bipolar disorder. However, establishing the final shape of such a device must take the patient's and caregiver's perspectives into consideration. The research to date suggests that use of the patient's own phone for data harvesting (for example of location, Bluetooth encounters, and motion via accelerometers) coupled with unobtrusive sensors in the home (monitoring activity, sleep patterns and the chosen environment in terms of noise and light levels) could provide.