Introduction : Cognitive deficits due to a moderate to severe traumatic brain injury (TBI) may greatly interfere with the completion of daily activities, such as money management and meal preparation. In this context, caregivers play a major role in compensating the difficulties and deficits of their loved one (e.g. by providing cognitive assistance or completing the task altogheter), thus increasing their burden overtime. Assistive technologies, such as the Cognitive Orthosis for Cooking (COOK), are a promising way to offer personalized assistance during a task, promoting maximal use of residual abilities following a TBI and reducing caregiver burden. However, better understanding how assistance could be provided according to the person’s difficulties is required to adapt the technology accordingly.
Methodology : Fifty videos of participants living with a moderate to severe TBI and evaluated using an ecological evaluation (IADL Profile) were analysed using a coding grid to identify: (1) moments of difficulty experienced by the participant (e.g. observed behavior and underlying difficulty) and (2) the type of cognitive assistance provided in response by the evaluator. As COOK focuses on meal preparation, only tasks related to this activity (e.g. grocery shopping, preparing a hot meal) were analysed. Descriptive statistics (e.g. frequencies) were completed.
Results : Of the 50 participants, 45 required cognitive assistance to progress in the task. Main difficulties experienced by TBI participants included executing the meal (28 participants, total frequency (n) = 283), formulating the goal of grocery shopping (31 participants, n = 191) and planning their meal (27 participants, n = 145). More precisely, they needed assistance to help them formulate an adequate solution (i.e. goal formulation; 34 participants, n = 235), find alternatives (i.e. planning; 25 participants, n = 110), modify their plan when in action (i.e. execution; 17 participants, n = 102) and find items in the environment (i.e. execution; 16 participants, n = 105). Main behaviors observed when confronted with a difficulty included asking for help (44 participants, n = 280) and task breakdown (24 participants, n = 159). In terms of cognitive assistance, cues (40 participants, n = 297), giving back the control of the task to the participant (40 participants, n = 187) and stimulating the participant’s thought processess (22 participants, n = 86) were the most used types of assistance overall. 
Conclusion : According to these results, cueing, giving back the control of the task to the person and stimulating thought processes are three types of cognitive assistance that could be provided to TBI individuals. Considering that cues are commonly offered by assistive technologies, further steps should focus on integrating implicit assistance such as the latter two in COOK.

Analysing cognitive assistance provided to TBI individuals during an ecological evaluation : A first step toward personalized assistance using assistive technology