Data collection

Sociodemographic data about the patients and caregivers and data relating to patients’ cognitive impairment, depression symptoms and experience in the use of technology were collected at the baseline. The Mini Mental State Exam (MMSE) was used to evaluate the level of patients’ cognitive impairment [22] and the Geriatric Depression Scale was used to quantify the intensity of any symptoms of depression [23]. Following Rubin & Chisnell [24], the patients were categorized as having low, medium or high levels of experience in the use of technology. This was done through specific questions that were presented to them and related to the previous use of similar appliances.

Patient acceptance of technology was assessed using the integrative framework from the Wearable Technology Acceptance in HealthCare (WTAH) survey developed by Gao [25]. This is an instrument designed to evaluate the factors associated with a user’s willingness to adopt wearable technology in healthcare. It is based on the unified theory of acceptance and the use of technology 2 (UTAUT2) [26], the protection motivation theory (PMT) [27], and the privacy calculus model [28]. In this research, we slightly adapted Gao’s original construct in order to fit it into the context of using a wearable lifelogging camera to improve memory. Each construct contained 3 items and each item was measured on a five-point Likert scale, with scores ranging from 1 (“strongly disagree”) to 5 (“strongly agree”). The last construct used, called the Behavioural Intention (BI), was the dependent variable. The different constructs are described in Table 1. A composite score was obtained for the different constructs by calculating average scores for the 3 items used in each construct. To ensure that the patients had correctly understood the content of the survey items, a member of the research team was present when each patient completed the survey and helped them to resolve any questions and to clarify the content. The WTAH survey has been made available as Additional file 1.

Constructs included in the Wearable Technology Acceptance in Health Care (WTAH) survey

Adapted from Gao [22]

Another quantitative analysis consisted of computing the rate of obtaining useful images with respect to the total number of images recorded and downloaded by each subject on a daily basis. This measurement was used to assess the extent to which the patients had used the camera correctly and to check whether an adequate number of images had been obtained to represent significant daily episodes that could subsequently be used in the cognitive intervention programme. The task of selecting the images was performed by computer vision tools that contained a deep learning algorithm. This enabled them to organize the large-scale collection of images captured by the patients on a daily basis [29]. The deep learning algorithm was specially designed to automatically discriminate between informative and non-informative images. It was also programmed to discard non-informative images, such as dark images, blurred images or images without content, such as when the camera photographed a wall, the ground, a ceiling, or the sky, etc., and also any repeated images. The cognitive intervention of the ReMemory Project requires 60 informative images for each daily episode. Assuming that the camera will collect a large number of non-informative images (we calculate that approximately 75% of the images captured tend to be non-informative given that the camera automatically takes images once every 30 s), it is expected that each patient should be able capture at least 300 relevant images per day.

The qualitative data were obtained from two focus groups (one for each patient centre) in which the patients and their caregiving relatives participated. The duration of each focus group’s activity was approximately 90 min and both focus groups were led by the same person. This person, who was a member of the research team and had relevant experience in organizing this type of task, acted as moderator for these groups. The group leader followed a pre-established script and facilitated discussion. They also had the support of an assistant moderator, who took notes throughout each session. The resulting script was structured around three major topics: a) ease of use of the camera; b) privacy, and c) a general assessment: acceptance, expectations and recommendations for improvement. The proceedings of the focus groups were recorded and transcribed.