Back

Evaluating the clinical utility of hand performance information from at-home egocentric video in outpatient neurorehabilitation

Kadambi, A.; Manzone, D. M.; Zariffa, J.

2024-11-03 health systems and quality improvement
10.1101/2024.09.27.24314512 medRxiv
Show abstract

BackgroundRestoring hand function is a primary focus of rehabilitation after neurological injuries, such as stroke and spinal cord injury. However, monitoring hand use outside the clinic remains challenging. This study aims to evaluate how therapists perceive and would utilise information from a clinical decision support system (CDSS) that uses egocentric video to monitor patients hand use at home. MethodsFive patient-therapist dyads were recruited. Patients recorded daily activities using head-mounted cameras. Therapists reviewed dashboards of processed video data from their patients and completed semi-structured interviews and structured questionnaires. A multi-methods approach with thematic analysis was used to evaluate the CDSSs clinical usefulness. ResultsQualitative analyses revealed four main themes: therapists strongly preferred observational video data over quantitative metrics, valued insights into home environments that cannot be captured in-clinic, identified ways the information could tailor therapy to real-world activities, but noted practical implementation challenges including time constraints and patient burden. The system achieved high usability scores (mean SUS: 80.0, 85-89th percentile). ConclusionsThis study demonstrates the potential of using egocentric video to inform clinical decision-making in neurorehabilitation, particularly for hand function. The strong preference for video over metrics suggests clinical decision support systems should prioritize interpretable, observation-based information aligned with clinical reasoning processes. Despite implementation challenges, therapists across technical familiarity levels expressed trust in the system and willingness to use it regularly. These findings indicate that egocentric video systems can bridge the clinic-home divide when designed to match interest-holder priorities.

Published in BMC Medical Informatics and Decision Making · training set

Matching journals

The top 4 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.