Favourable ethical opinion not obtained

This study aims to explore whether feedback from a physical monitoring device (electronic watch) to prompt patients to use an anxiety management strategy can help prevent challenging behaviour.

A multiple baseline "A-B-A" single case experimental design will be adopted with baseline, intervention, and return to baseline phases.

Participants will wear the portable monitoring device with heart rate monitor, pedometer and global positioning system (GPS) functions. A pre-calculated algorithm, developed using machine learning software will be used to predict challenging behaviour episodes 2-4 hours in advance of the event. Recent research has developed a best predictor model with a sensitivity of 82% and specificity of 42%, when averaged across two participants (in press). The device will communicate with a smart phone, via Bluetooth connectivity, when an event has been predicted.

Following enrolment in the study, participants will be randomly assigned to a treatment start time, which will follow from an initial baseline phase. During the baseline phase, participants will wear a commercially available smart watch with physiological monitoring capability, for at least one month in order to gather behaviour frequency data, and associated physiological data; no prompts will be provided. During this period, behavioural and physiological data will be sent to a secure server where a machine learning algorithm will examine the relationships between data streams and 'self-tune' to improve its prediction of challenging behaviour episodes. In the treatment phase, participants and staff will receive a biofeedback prompt in text, or graphic form (e.g. mood light), on the smart watch/connected mobile phone, when an event has been predicted. This will prompt participants to implement a proactive anxiety management strategy, and will prompt staff to support the participant in completing this, if required. The strategy used will be predetermined in the participant's standard care plan e.g. breathing relaxation, reframing the problem, behavioural redirection. Details will be recorded using an adapted version of the Overt Aggression Scale-Modified for Neurorehabilitation (OAS-MNR), describing the situation in which the biofeedback prompt occurred, participant and staff responses to the biofeedback prompt, and any resulting behaviours, including the occurrence of challenging behaviour. Following the intervention phase, participants will return to a baseline phase for one month, in which no prompts will be provided. Again, details of any challenging behaviour will be recorded via the OAS-MNR.

Inclusion Criteria:

• participants who have sustained a severe injury to the brain, as determined by length of unconsciousness (greater than 24 hours), a Glasgow Coma Scale of 3 to 8 and a Post Traumatic Amnesia period greater than 1 day
• participants who are resident in the Graham Anderson House, Brain Injury Rehabilitation (BIRT) Unit
• participants presenting with challenging behaviour, during the recruitment period
• participants will NOT be excluded on the basis of comorbid psychiatric disorders or drug/alcohol use, in order to provide conclusions representative of the clinical population
• participants will NOT be excluded on the basis of being unable to provide informed consent, in order to provide conclusions representative of the clinical population

Exclusion Criteria:

• a level of cognitive impairment that would prevent individuals from understanding the feedback obtained via digital or verbal prompts
• participants without the capacity to consent, that do not provide verbal assent to participate in the study