IEEE Trans Neural Syst Rehabil Eng. 2017 Nov;25(11):2009-2017. doi: 10.1109/TNSRE.2017.2698005. Epub 2017 Apr 25.
Patient condition during rehabilitation has been traditionally assessed using clinical scales. These scales typically require the patient and/or the clinician to rate a number of condition-related items to obtain a final score. This is a time-consuming task, specially if a large number of patients are involved. Furthermore, during rehabilitation, user condition is expected to change steadily in time, so assessment may require to run these scales several times to each user. To save time, much effort has been focused on developing clinical scales that require little time to be completed. This is usually achieved by measuring a reduced set of features, i.e., focusing the scales on specific features of a defined target population (Parkinson's disease, Stroke, and so on). However, these scales still require the therapist's intervention and may be tiresome for patients who have to fill them repeatedly. This paper proposes a novel approach to automatically obtain balance scales from the onboard sensors of a robotic rollator. These sensors are used to extract spatiotemporal gait parameters from patients using the rollator for support. These parameters are derived from the user forces on the rollator handles and its odometry. Resulting parameters are used to predict the Tinetti mobility clinical scale on the fly, without therapist intervention. Our approach has been validated with 19 rollator volunteers with a variety of physical and neurological disabilities at Hospital Civil (Malaga) and Fondazione Santa Lucia (Rome). Clinicians provided traditionally obtained Tinetti scores and the proposed system was used to estimate them on the fly. Results show a small root mean squared prediction error. This method can be used for any rollator user anywhere in everyday walking conditions to obtain the Tinetti scores as often as desired and, hence evaluate their progress.
患者在康复过程中的状况传统上是通过临床量表进行评估的。这些量表通常需要患者和/或临床医生对多项与病情相关的项目进行评分，以获得最终分数。这是一项耗时的任务，特别是如果涉及大量患者。此外，在康复过程中，用户的状况预计会随着时间的推移而稳步变化，因此评估可能需要对每个用户多次进行这些量表的评估。为了节省时间，人们已经在开发需要很少时间完成的临床量表方面付出了很多努力。这通常通过测量较少的特征集来实现，即通过关注量表来测量特定的特征集（如帕金森病、中风等）。然而，这些量表仍然需要治疗师的干预，而且对于必须反复填写量表的患者来说可能会很麻烦。本文提出了一种从机器人助行器的板载传感器自动获取平衡量表的新方法。这些传感器用于从使用助行器进行支撑的患者身上提取时空步态参数。这些参数是从用户对助行器手柄的力和其里程计中得出的。生成的参数用于在无需治疗师干预的情况下实时预测 Tinetti 移动性临床量表。我们的方法已经在位于 Malaga 的 Hospital Civil 和位于 Rome 的 Fondazione Santa Lucia 的 19 名助行器志愿者身上进行了验证，这些志愿者具有各种身体和神经残疾。临床医生提供了传统的 Tinetti 得分，然后使用提出的系统实时对其进行估计。结果表明，预测误差的均方根很小。这种方法可以用于任何在日常步行条件下使用助行器的用户，以便根据需要经常获得 Tinetti 得分，并评估他们的进展情况。
