Brain-computer interfaces patient preferences: a systematic review.

Background; Brain-computer interfaces (BCIs) have the potential to restore motor capabilities and functional independence in individuals with motor impairments. Despite accelerating advances in the performance of various implanted devices, few studies have identified patient preferences underlying device design, and moreover, each study has typically captured a single aetiology of motor impairment. We aimed to characterise BCI patient preferences in a large patient cohort across multiple aetiologies. Methods; We performed a systematic review of all published studies reporting patient preferences for BCI devices. We searched MEDLINE, Embase, and CINAHL from inception to April 18th, 2023. We included any study reporting either qualitative or quantitative preferences concerning BCI devices. Article screening and data extraction were performed by two reviewers in duplicate. Extracted information included demographic information, current digital device use, device invasiveness preference, device design preferences, and device functional preferences. Findings; Our search identified 1316 articles, of which 28 studies were eligible for inclusion. Preference information was captured from 1701 patients (mean age = 42.1-64.3 years). Amyotrophic lateral sclerosis was the most represented clinical condition (n = 15 studies, 53.6%), followed by spinal cord injury (n = 13 studies, 46.4%). We found that individuals with motor impairment prioritise device accuracy over other device design characteristics. We also found that the speed and accuracy of BCI systems in recent publications exceeds reported patient preferences, however this performance has been achieved with a level of training and setup burden that would not be tolerated by most patients. When comparing populations across studies, we found that patient preferences vary according to both disease aetiology and the severity of motor impairment. Interpretation; Our findings support a greater research emphasis on minimising BCI setup and training burden, and they suggest future BCI devices may require bespoke configuration and training for specific patient groups.