Four Robin-T (TES Co., Osan, Korea) units were used at Dong-A University Hospital, in Busan, Korea since February 2014. Originally, the robot was developed for transferring patients who had difficulty to move so that their activities of daily living (ADL) could be improved. However, it could not be used because of a long preparation time to put on safety equipment and its slow speed of moving. However, the safety equipment for preventing falls provided good stability. This feature could allow the robot to be used as a sitting device, a standing frame, or a partial body-weight support high-walker. It could be utilized safely within a ward. Such robot can alert patients who want to exercise by themselves in their ward but could not walk independently due to poor balance or hemiplegia after stroke. Therefore, they could have more time for rehabilitation.

The robot weights approximately 120 kg. It has four wheels to make it easier to move. Two straps affix the lower limbs and one strap affixes the waist. Its main frame has a handle that can be grabbed easily (Fig. 1). The two straps that affix the lower limbs become an X shape when patient wears them. It could help partial weight-bearing. The waist strap could prevent backward falling tendency. These straps combined with its heavy weight frame can protect patient from falling backwards or sideways. To get on this robot, the patient needs to sit on the bed. The caregiver places the robot in front of the patient. The patient then leans on the trunk of the robot and grabs the handle with their hands. The caregiver then helps the patient put on the waist and lower limbs straps. The height can be adjusted using a remote control attached to the robot.

According to functional status of the patient, usage level of the robot was set before treatment. Patients who were unable to maintain independent sitting balance could use the robot as a sitting device. Those who were unable to maintain independent standing balance actively could use the robot as a standing frame to improve their antigravity muscle strength, control head and trunk posture, maintain standing ability, and prepare for gait training. Patients who were unable to walk without assistance could use the robot as a high-walker to improve gait mobility and balance. When patient reached desired function during treatment, the usage level of the robot could be changed to the next level. In rehabilitation wards, the robot was used as a sitting device for patients who needed sitting balance training (Fig. 2A). Fixed wheels acted as a safe standing frame for patients who needed standing balance training and lower extremity strengthening (Fig. 2B). For patients who performed gait training with a partial body weight support device in a rehabilitation room and those who conducted gait training using a walker or cane, this robot could be used as a partial body-weight support high-walker (Fig. 2C). Using the robot, patients could ambulate around the ward and go to restroom under their caregiver's supervision or with minimal assistance. To utilize these robots in wards, physicians explained to patients and caregivers on how to use the device and safety equipment. After the demonstration, patients and caregivers verified that they could understand and use the device safely as a measure of exercise by themselves. All patients' caregivers used this device independently without any help from the medical team. Patients watched television while using the device in a sitting position. It enabled safe sitting balance training and core muscle strengthening exercises.

Medical records of patients who had used Robin-T were analyzed retrospectively. Patients who had been diagnosed with subacute stroke were enrolled. Subjects were referred to the Department of Physical Medicine and Rehabilitation, Dong-A University Hospital between June 2014 and December 2016 to receive rehabilitation treatment. During this period, 22 patients used these devices. Each patient used the robot at least 30 minutes per day. Among these patients, those with recurrence of stroke, cognitive impairment, or previous history of musculoskeletal issues resulting in gait problems were excluded. Patients whose caregiver were changed twice or more, those who did not comply with the guidance, and those who could not utilize the robot longer than 30 minutes per day were excluded. Retrospective review was performed for the remaining 15 patients. All patients had hemiplegia or ataxia within 3 months of stroke onset. All patients received conventional rehabilitation programs, including medication, physical and occupational therapy, and neuromuscular electrical stimulation.

Inclusion criteria were: (1) those who understood Korean, (2) those who were still in the subacute rehabilitation phase (1 week to 3 months post stroke), (3) those who had a caregiver willing to participate in the program, (4) those who could follow instructions, (5) those who had mobility problems caused by stroke (Functional Ambulation Category [FAC] score <4), and (6) those who had adequate cognition to participate (Mini-Mental State Examination score >23). Caregivers (partner, family member, or private caregiver) were eligible to participate if they were physically able to perform the exercise with the patient, understood Korean, and agreed to provide support to the patient without significant symptoms of depression.

All data were collected from medical records after treatment. Clinical improvement was assessed by changes between before and after caregiver driven robot-assisted in-ward training using the Korean version of Modified Barthel Index (K-MBI), Functional Independence Measure (FIM), and FAC. The K-MBI measures independence in ADL, with a maximum score of 100. It assesses 10 ADLs: bowel control, bladder control, personal hygiene, toilet transfer, bathtub transfer, feeding, dressing, wheelchair transfer to and from bed, walking (or wheelchair management if the patient is non-ambulatory), and ascending and descending stairs. The FIM has a total of 18 items. It is also used to assess performance during tasks that can be broadly categorized as ADL, mobility, and cognition. Each item is scored from 1 to 7, with 7 signifying complete independence or normal functioning and 1 signifying complete dependence or requiring total assistance. The total maximum score of FIM is 126 which indicates total independence. Its minimum score is 18, indicating that full assistance is required for all 18 items. The study protocol was approved by the Institutional Review Board of Dong-A University Hospital (No. DAUHIRB-17-129).

For statistical analysis, Wilcoxon rank sum test was used. Data were analyzed using SPSS version 22.0 for Windows (IBM, Armonk, NY, USA).

In the present study, Case #2 (bilateral pontine hemorrhage, a 48-year-old man) was transferred to the rehabilitation center at 65 days after the onset of stroke. The patient exhibited dysarthria, dysphagia, and mild double hemiplegia and severe ataxia on both upper and lower limbs, all of which greatly inhibited ADL. He initially presented with severe ataxia and weakness on both sides. He was poor at weight bearing and balance. Thus, maximal assistance by the physical therapist was needed for training. In the ward, sitting balance training at bedside presented limitations. He could not maintain a sitting posture without maximal support. We used the device to assist him with sitting balance training. With the X-figure leg strap and waist support strap, the patient could make a chair sitting posture (Fig. 2A). Using this robot for sitting balance training, the patient's erector muscles of the spine seemed to be getting stronger. He could maintain sitting balance independently without any back support. He was encouraged to get out of the bed during the day. At discharge, he could stand up with moderate assistance.

The most dramatic improvement in K-MBI and FIM scores was observed in Case #5 (right pontine infarction, a 60-year-old woman). She was transferred to the rehabilitation center at 41 days after the onset of stroke. The patient exhibited left hemiplegia and ataxia on the upper and lower left limbs. She had difficulty of gait initiation with backward falling tendency. Thus, supervision or minimal assistance by a physical therapist was needed for standing balance training. In the ward, static standing balance training by a caregiver was dangerous due to her poor balance. To support her body weight, the robot was used. With waist support strap and handle, the patient demonstrated increased stability. She could perform inward standing training under a caregiver's supervision (Fig. 2B). Using this device for therapeutic exercise was effective in improving gait and balance function and ADL (p<0.05) for this patient. She and the caregiver (her husband) could independently use the device without help from the medical team. At discharge, she could walk through support from a walker with moderate assistance.

Case #15 (left cerebellar infarction, a 65-year-old man) was transferred to the rehabilitation center at 58 days after the onset of stroke. Although his muscle powers on the left upper and lower limbs were relatively well maintained, appropriate motor functions could not be performed due to ataxia and abnormal motor responses, all of which inhibited self-gait. In the ward, high-walker gait training with a caregiver was dangerous due to the patient's falling tendency. We used this robot for gait training in the ward. With the waist strap, the patient demonstrated decreased falling tendency. He could walk with minimal assistance from a caregiver in the ward (Fig. 2C). The private caregiver could independently use the robot without any help from the medical team. At discharge, the patient could walk with a walker under a caregiver's supervision.