See Events

PhD Defence Juliet Haarman

TIBAR 'Therapist inspired balance assisting robot' 

Juliet Haarman is a PhD Student in the research group Biomedical Engineering, her Supervisors are professor Hans Rietman and professor Herman van der Kooij from the Engineering Technology Faculty (ET)

Stroke survivors who are classified with a Functional Ambulation Category (FAC) of 2, typically are able to ambulate on level surfaces, but intermittently require manual contact of a physical therapist to assist their balance during walking. In order to execute Activities of Daily Living (ADL’s), patients require a level of walking independence where no physical balance assistance is needed. Physical therapists use training to improve gait. During these training sessions, patients are able to move freely and only receive balance assistance (support) when they are not able to keep themselves upright. Providing balance assistance not only allows patients to continue their training safely, it also lets them experience the boundaries of their abilities without actually falling.

Even though gait training was found to be effective for stroke survivors in terms of regaining functional independence, the one-on-one contact with the patient and the constant need for supervision limits the training volume of patients. Patients benefit, among other aspects, from a training environment where sufficient training hours can be made at a suitable training intensity. Specifically, as our population is aging and consequently the number of patients is increasing, an even higher burden is put on the training volume of patients in the near future, thereby limiting their rehabilitation effectiveness. A solution that is expected to positively contribute to such a situation is the use of a robotic training device that supports balance, with which patients can undertake additional training hours in a self-administered training environment and that does not ask for the direct presence of a therapist during the training.

The development of such a device was described in this thesis: The TIBAR (Therapist Inspired Balance Assisting Robot). The TIBAR is a robotic device that provides balance assistance to patients at times when they are not able to keep themselves balanced. The device could be used in a self-administered way and must be used on a treadmill. An augmented reality environment could additionally be used on the treadmill to adapt the level of training exercises to the abilities of the patient. Research questions have been set in chapter 1, prior to the development of the TIBAR.

  • What is the best method of implementing optimal training conditions into a robotic training device?
  • How should stroke related consequences such as hemiparesis be accounted for in a robotic training device?
  • How can the behavior of physical therapists in terms of providing balance assistance best be implemented into a robotic training device?

The research that has been performed in this thesis elaborates on these research questions and their implementation in the TIBAR. Research question 1 aims to implement optimal training conditions that are important in (re)learning motor tasks into the design of a robotic training device. Aspects of motor learning that are believed to be of importance are error-based training, active participation, functional/context specific training, feedback, motivation and training volume.

Chapter 2 focused on the implementation of the aspect of active participation and error-based learning in robotic gait training, two aspects in motor learning that are believed to be applied by physical therapists and are important in functional recovery. A systematic review was performed in this chapter that compared robotic gait training studies with devices that did incorporate these aspects into their controller design (‘patient-in-charge’ devices) to studies with robotic devices that did not incorporate these aspects (‘device-in-charge’ devices). Results of the comparison between both categories showed no preference for one category over another in terms of functional recovery. An important confounder in these results is the effect of training volume. The training duration of the included studies was often too short to expect any functional recovery at the side of the patient.

Chapter 3 additionally focused on the effects of motivation and feedback in a training environment. This chapter focused on the ability of subjects to learn a visuo-motor walking task in an augmented reality training set-up. An augmented reality training environment is often used to increase the motivational component of a training, by adding explicit goals (targets/games) to the training environment. Results identified the intuitive character of such a training environment and the ability of such an environment to adapt the level of the training to patient specific needs.

Research question 2 is related to stroke related characteristics such as hemiparesis. It was set in the requirements that training with the TIBAR should not be limited by effects of hemiplegia. Therefore, the capabilities of these patients to deal with perturbations and their asymmetrical movements were taken into account in the development of the TIBAR. In chapter 4, several perturbation magnitudes and perturbation directions were applied to both the paretic and the non-paretic leg of stroke survivors and stepping responses were mapped. The results implicate that the TIBAR does not need to take different modulation strategies with both legs into account. However, the controller design of the TIBAR should account for the asymmetrical medio-lateral COM movement-trajectory of patients.

The behavior of therapists during conventional therapy in terms of manual physical balance assistance appears to be effective for the functional recovery of stroke patients. Therefore, research question 3 focuses on the implementation of this behavior into the design of the TIBAR. Chapter 5 elaborates on this aspect as it focused on mapping the intuitive and effective training method of therapists, in terms of physical manual contact between patient and therapist. Characteristics of therapeutic balance assistance were captured in terms of amount of force, duration and impulse. Predicting the intuitive and complex behavior of therapists in providing balance assistance logically depends on many aspects such as patient specific muscle strength, fall history and coordination characteristics. Still, it was attempted in this chapter to develop an algorithm that was able to predict the behavior of physical therapist in providing balance assistance with one sensor, such that this information could be transposed into the controller design of the TIBAR.

The information from these chapter has been combined to develop a first prototype of the TIBAR. A description of this device has been presented in chapter 6. The device has been tested with stroke survivors in a clinical setting and results have additionally been presented in this chapter. Results showed the ability of the system to provide balance assistance to patients with limited balance control. Technical validation of the TIBAR indicated the similarities between the behavior of therapists and the device, in terms of force characteristics and timing of the balance assisting events. User acceptance evaluations have indicated the good change for acceptance of such a device in a clinical setting.