Portal de Periódicos da CAPES

Use of Telepresence Robots for Home Safety Evaluations: A Feasibility Study

2020; Volume: 3; Issue: 3 Linguagem: Inglês

10.3928/24761222-20200116-01

2476-1230

Mi Jung Lee, Erica Poulin, Gail Castañeda, Sergio Romero,

Social Robot Interaction and HRI

Original Research freeUse of Telepresence Robots for Home Safety Evaluations: A Feasibility Study Mi Jung Lee, PhD, BSOT, ; , PhD, BSOT Erica Poulin, OTR/L, ; , OTR/L Gail Castañeda, PhD, ; and , PhD Sergio Romero, PhD, , PhD Mi Jung Lee, PhD, BSOT , Erica Poulin, OTR/L , Gail Castañeda, PhD , and Sergio Romero, PhD Published Online:January 22, 2020https://doi.org/10.3928/24761222-20200116-01Cited by:1PDFAbstract ToolsAdd to favoritesDownload CitationsTrack CitationsCopy LTI LinkHTMLAbstractPDF ShareShare onFacebookTwitterLinkedInRedditEmail SectionsMoreAbstractObjective:Home safety evaluations are necessary for home modifications to increase the functioning and safety of patients in their homes. However, in some cases, home safety evaluations may not be feasible because necessary resources or local experts are not available. The goal of this study was to examine the feasibility of conducting home safety evaluations remotely with a telepresence robot.Methods:Five occupational therapists received instruction in driving a telepresence robot, completed a practice driving course in a controlled environment, and conducted a remote home safety evaluation.Results:On average, therapists needed 40 minutes to complete the training (including instruction and completion of the practice driving course) and 48 minutes to perform the remote home safety evaluation. Participants identified a number of barriers to conducting the remote evaluation: (a) poor driving technique, (b) limitations of the equipment, (c) unstable network connections, and (d) use of an unfamiliar evaluation tool. The mock caregivers provided minimal assistance to complete the remote evaluation, including repositioning the robot and providing navigation guidance.Conclusion:Although several limitations to the use of telepresence robot technology were identified, overall, therapists indicated that using the robot was a viable approach for conducting remote home safety evaluations with adequate assistance from caregivers. This study provides preliminary results on the use of telepresence technology to conduct remote home safety evaluations. [Annals of International Occupational Therapy. 2020;3(3):119–126.]IntroductionAccording to an American Association of Retired Persons survey of Americans in 2014, 87% of survey participants 65 years or older preferred to age in their homes and communities (American Association of Retired Persons, n.d.). Through aging in place, older adults enjoy independent control over their lives and maintain social connections to their friends and communities (Iecovich, 2014). However, aging in place can entail significant challenges.As a person's competency or health declines with age, an unchanged environment can result in negative health outcomes, such as functional difficulty at home (Gitlin et al., 2006; Petersson, Lilja, Hammel, & Kottorp, 2008; Szanton et al., 2011), depression (Gitlin, 2003), and a reduced sense of well-being (Oswald et al., 2007). An unsupportive environment can restrict older adults from performing daily activities, endanger home safety, increase the burden on caregivers, and increase the risk of hospitalization (Pynoos, Nishita, Cicero, & Caraviello, 2008).The American Occupational Therapy Association emphasized that different environmental supports are needed for individuals as they age, and home modifications have been identified as a critical component of productive aging (American Occupational Therapy Association, n.d.). The occupational therapy literature indicates that home modification provides the right person-environment fit to support daily functioning and improve home safety (e.g., reduced number of fall-related injuries) (Gitlin, 2015; Lawton & Nahemow, 1973; Wahl & Gitlin, 2007).In 2008, Petersson et al. concluded that older adults who received home modifications had increased functional independence in activities of daily living compared with those who did not receive home modifications. Another study showed a 26% decrease in the rate of fall-related injuries at home for participants in the home modification intervention group, after controlling for participant demographics and history of falls (Keall et al., 2015). By enhancing functional independence and home safety, home modification is a promising method for supporting successful aging in place (Damme & Ray-Degges, 2016; Marquardt et al., 2011; Semeah et al., 2017).Although home modifications are encouraged to reduce fall-related injuries and improve functional independence, these modifications require a labor-intensive assessment of the home environment by a qualified professional (typically, an occupational therapist visiting a client's home) to identify barriers and risks. This evaluation is challenging in areas where expertise is not available (Bayer & Harper, 2000; Pynoos et al., 2008) and where distance limits the feasibility of sending a qualified professional from the closest medical facility. This problem is particularly prominent in rural areas.Remote technology has been used to improve home safety for people in remote areas by allowing them to be connected to home assessment experts (Sanford & Butterfield, 2005). Sanford, Jones, Daviou, Grogg, and Butterfield (2004) used real-time teleconferencing to conduct home safety evaluations with the assistance of a technician. This remote evaluation accurately detected 51 of the 59 problems (86.4%) identified in a previous on-site evaluation. The authors concluded that remote technology could provide meaningful information about the safety of a home environment (Sanford et al., 2004). This study provided evidence supporting the use of real-time teleconferencing to conduct home evaluations; however, it still required sending personnel (i.e., a technician) to the remote location, which does not resolve challenges with travel distance.Other technologies permit clinicians to navigate a distant environment autonomously and communicate with clients at the remote site. Telepresence robots can achieve this goal by connecting through the Internet with a clinician who can control the robot and communicate through it, providing a "virtual" clinician presence at the remote site. With a telepresence robot, clinicians can virtually navigate clients' homes to conduct a safety assessment and communicate with a caregiver or patient without needing a technician to be present in the home. Telepresence robots have been used successfully in medical intensive care units to improve patient care, patient satisfaction, and overall patient communication (Becevic et al., 2015). Thus, telepresence robots are well suited to conduct remote evaluations with minimal assistance from the client. The goal of this study was to investigate the feasibility of using a telepresence robot to conduct home safety evaluations. We examined both the amount and the type of assistance needed to conduct the home safety assessment to determine whether a "typical" caregiver could provide this assistance.MethodsParticipantsWe recruited five licensed occupational therapists from local rehabilitation facilities in Gainesville, Florida. Inclusion criteria for therapists were as follows: (a) experience conducting home safety evaluations and (b) no previous experience operating a telepresence robot. In addition, four occupational therapy students served as mock caregivers to assist with the remote evaluations if needed. Before the start of training, all participating occupational therapists provided demographic information and indicated their confidence level (with the use of a 3-point Likert scale) in operating a robot. This study was approved by the appropriate institutional review board at the University of Florida.Telepresence Robot and Related DevicesDouble Robotics features a "double" robot, wherein "double" means a "body double"—a technological twin of a real-life remote user whose face is displayed by an iPad at the "head" of the robot. The remote user's "body" is a stick pole connected to a wheeled base. A Double Robot uses a gyroscope and accelerometers to balance, similar to the popular Segway (Double Robotics, n.d.). The Double Robot has a "raise-lower" feature that allows users to adjust the height of the robot to the viewer's eye level. An iPad is mounted on top of the robot for videoconferencing, and a mirror is attached to the bottom of the mount to provide an extended view of the ground below it. The approximate weight of the Double Robot, including an iPad, is 15 lb. The robot is designed to be used in indoor settings.A clinician can use a computer or an iPad that is connected to the Internet to navigate the Double Robot remotely and interact with others through videoconferencing, recreating the sense of being physically present. Thus, a Double Robot requires a high-speed Internet connection. Because the goal of our study was to test the feasibility of using the robot autonomously, we used a 4G cellular connection for Internet access. The connection was encrypted with the existing Double Robotics application. This study tested the first generation of the Double Robot (Figure A, available in the online version of the article) and the two compatible iPad models (iPad 2 and iPad Air 2).Figure A: The first generation of the Double Robot.Development of the Training ProtocolA step-by-step educational protocol was developed for occupational therapists operating the robot with the following protocol sequence: (a) the occupational therapists were assigned to watch two Double Robotics instructional videos; (b) they read written instructional handouts; and (c) then they completed a practice driving course. Instructional videos provided general guidelines for operating the robot, descriptions of the robot's functions, and driving tips to overcome certain barriers, such as going over objects on the floor, climbing ramps, and others. Written instructions included images of the Double Robot, directions for establishing connectivity (iPad to base) and achieving a virtual connection to the robot from remote computer users, computer screenshots with detailed explanations of each icon on the navigation screen, keyboard shortcuts, and driving tips. The Double Robotics user manuals and instructional videos were used to create these written instructional handouts. The final phase of user training included robot navigation via a controlled training course that required the therapist to complete 13 driving maneuvers, including going straight, turning, circling, overcoming thresholds (i.e., electrical cords), going through doorways, navigating around objects, backing, and parking. A map with directional arrows for each maneuver was provided to help drivers to complete the training course (Figure B, available in the online version of the article).Figure B: A map with directional arrows for each maneuver was provided to help drivers to complete the training course.Training of Occupational TherapistsThe occupational therapists completed all steps of the training protocol at the offices of the Center of Innovation on Disability and Rehabilitation Research (CINDRR), North Florida/South Georgia VA Medical Center, Gainesville, Florida. They were advised to spend as much time as necessary to complete the training. To examine whether the training protocol alone was sufficient to train the occupational therapists to conduct remote home safety evaluations, no questions about the protocol tasks were permitted. At the end of the training, occupational therapists provided feedback on the difficulty of completing the training.Remote Home Safety EvaluationGator Tech Smart House. After the completion of training, occupational therapists conducted a remote home safety evaluation of the Gator Tech Smart House from our CINDRR office. The Gator Tech Smart House, located in the Oaks Hammock retirement community in Gainesville, Florida, is a home designed for research projects on aging in place. The house provides a standardized environment that is suitable for this type of research study.Westmead Home Safety Assessment. To standardize the process, we used the Westmead Home Safety Assessment (WeHSA) to conduct the home safety evaluation because it is widely used and has been cited in several research studies (Barstow, Bennett, & Vogtle, 2011; Lannin et al., 2007; Romero, Lee, Simic, Levy, & Sanford, 2018). The WeHSA checklist was designed to allow occupational therapists to evaluate potential safety hazards in the home, and it has shown acceptable psychometric properties (e.g., interrater reliability and content validity) (Asher, 2014).Home assessments. Occupational therapists at CINDRR evaluated the Gator Tech Smart House by observing the home features with the robot, which had been transported to this location by mock caregivers beforehand. Occupational therapists were allowed to take as much time as they needed to complete the evaluation, including time for conversation with the mock caregivers to ascertain details about the environment or request assistance (i.e., remove a barrier). The mock caregivers performed additional tasks, including demonstrating features of the home (i.e., opening windows, showing how they use the kitchen area, reaching shelves), providing cues for directional navigation, and physically relocating the robot.Data collection and analysis. During the remote home safety evaluations, we observed the amount of time the occupational therapists spent conducting the home safety evaluation. We also observed the general interaction between the occupational therapists and mock caregivers. We asked the participants to provide comments, describe difficulties encountered, and evaluate their overall performance. We also obtained qualitative data through subjective feedback from occupational therapists and mock caregivers about the difficulties encountered and the feasibility, usability, and limitations of using the Double Robot for home safety evaluation. We analyzed open-ended survey questions with thematic coding and theme frequency counts. We performed thematic coding by reading open-ended responses in their entirety and identifying key words and phrases in subsequent readings. We grouped the key words and phrases into emergent themes and analyzed each theme separately for greater internal consistency. We then calculated theme frequency counts based on all responses provided for each question. We used an iterative approach to determine whether the feedback could be incorporated into the next evaluation. When an issue occurred with multiple participants and it was feasible to minimize the problem, we incorporated a possible solution and tested it with the next participant. For example, we used a wide-angle lens on the iPad mount for two of the five trials to resolve the recurring issue of limited camera view angle.ResultsParticipantsWe had two groups of study participants: occupational therapists and mock caregivers. Occupational therapists included four women and one man. Ages were 20 to 30 years (n = 1), 30 to 40 years (n = 1), and 40 to 50 years (n = 3). Confidence levels for operating robotic devices were as follows: not confident (n = 3), somewhat confident (n = 1), and confident (n = 1). Demographic features are shown in Table 1. Mock caregivers included three women and one man (one mock caregiver participated in two trials), and all were 20 to 30 years old.Table 1 Demographic Features of the Occupational TherapistsCharacteristicSubject 1Subject 2Subject 3Subject 4Subject 5GenderFemaleFemaleFemaleFemaleMaleAge range, years20–3040–5040–5040–5030–40Occupational therapy practice, years32661912Confidence in operating robotSomewhat confidentNot confidentNot confidentNot confidentConfidentQualitative AnalysisWe performed thematic analysis and theme frequency counts on occupational therapists' open-ended responses to the following prompt: "Please describe the barriers you encountered during the telepresence robot training and remote home safety evaluation." We counted the responses of the therapists to given emergent themes for corresponding theme frequencies (Table 2).Table 2 Theme Frequency CountsPhasesThemesCountsPossible countsPercentageTrainingPoor robot driving technique91560Lack of clear instructions82040EvaluationEquipment limitations102050Unstable network connections92045Poor robot driving technique72035Evaluation tool unfamiliarity1520Training of Occupational TherapistsOn average, therapists spent a total of 40 minutes completing the training, which included 18 minutes viewing two instructional videos and reading the written instructions and 22 minutes completing the training course. We identified the following two themes and corresponding frequencies in response to the open-ended prompt on barriers encountered during training: (a) poor robot driving technique (60%) and (b) unclear instructions (40%).Poor robot driving technique was the most frequent (60%) theme for barriers during training. Therapists reported various aspects of driving technique that affected maneuvering of the telepresence robot, such as the ability to make turns, judge distance between the robot and objects in its path, and switch camera views. For instance, one therapist noted that they had experienced "two slight collisions with chairs due to difficulty judging distances in front."Another theme expressed as a barrier for therapists during training was unclear instructions. The therapists asserted that the instructional material needed to be simplified, with supplementary graphic material. For instance, one therapist stated: It was a little bit frustrating because it looks like you can drive the robot fast and easily in the video, but I could not go fast or easily control the robot. I think instructions can be designed much better, with graphics to help the users understand the instructions and go through the instructions completely. I think video animation might offer those features to the users.Several therapists reported challenges with technological troubleshooting (e.g., managing the robot's account setup and log-in procedures, using iPad navigation, and finding the device's power button) because the instructions were unclear. One therapist reported difficulty turning the unit on because they "did not know to hold the button in to turn [the robot] on." The details of identified barriers for each occupational therapist are provided in Table A (available in the online version of the article).Table A Identified barriers for each occupational therapistThemesDetailsOT1OT2OT3OT4OT5TrainingPoor robot driving techniqueSwitching camera views (front and bottom)√√√√Judging distance√√√Maneuver (turn)√√Lack of clear instructionsAccount set up and connection (procedure)√Finding power button√√√√iPad unfamiliarity√Unclear written instruction√√Home safety assessmentEquipment LimitationsBlind spot√Camera angle√√√√Video quality (Unable to determine the texture, shape of small objects)√√√√Poor sound quality√Unstable network connectionsAudio√√√Video√√Computer and robot connection√√iPad and robot connection√√Poor robot driving techniqueBalancing on uneven surface√√Maneuver (speed)√Proprioception in space√√√Overcoming threshold√Evaluation tool unfamiliarityWestmead unfamiliarity√Remote Home Safety EvaluationAll occupational therapists reported that our training protocol was helpful and that using this technology to conduct home safety evaluations was feasible and seemingly comparable to on-site evaluations. On average, participants needed 48 minutes to complete the remote home safety evaluation. We identified the following themes and corresponding theme frequencies in the open-ended responses to barriers encountered during the remote evaluation: equipment limitations (50%), unstable network connections (45%), poor robot driving technique (35%), and unfamiliarity with the evaluation tool (20%).When participants were asked to describe barriers encountered during the remote home safety evaluation, the most frequently expressed theme was equipment limitations (50%). Therapists reported various limitations of the robot for conducting home safety assessments, such as blind spots, narrow camera angle, and low video quality. Blind spots, a limited camera angle, and poor video quality impeded the ability to navigate narrow pathways and observe home features (e.g., floor mat textures, floor surfaces, and the shapes of objects). For instance, one therapist stated that "objects that are shorter than the [robot] are not easily recognized, [creating a] blind spot."Unstable network connections were the second most frequently (45%) noted barrier to home safety evaluation. This theme included various forms of unstable network connections experienced by therapists, such as lagging audio and video, unstable computer-to-robot connections, and unstable iPad-to-robot connections. These issues delayed the remote evaluation process. When the issues were not resolved quickly (e.g., occasional Internet network interruptions), the occupational therapist and mock caregiver used another mode of communication (i.e., cell phone) to solve the problem. The following quote from an occupational therapist describes an unstable connection between the computer and the robot: "When the video was lagging, it was hard to maneuver the robot. I could not hold the arrow button. I had to click it and click it so that I could guess better where the robot would be."Frequency counts for unstable network connections (45%) were closely followed by those for poor robot driving technique (35%). Categories underlying poor robot driving technique included balancing on uneven surfaces, managing driving speed, lacking proprioception with the robot, and overcoming entryway thresholds or small objects on the floor. One therapist stated, "I had a hard time making a turn. The robot went around too fast. I often had to readjust directions." Nonetheless, all occupational therapists were confident that they could resolve these issues with additional practice.Finally, the least frequently expressed theme was unfamiliarity with the WeHSA evaluation tool (20%). One occupational therapist reported that the home safety assessment was delayed because extra time was needed to navigate the WeHSA items. The following quote illustrates how unfamiliarity with the evaluation tool can serve as a barrier to home safety assessment: "I am not familiar with the sequence of the Westmead evaluation form. Sometimes I had to visit the same place twice to check the checklist."The last two occupational therapists had a wide-angle lens for their trials because of the recurring issue of a limited camera angle. Of these two, one still reported that the camera angle was too narrow and made navigation with the robot difficult. Additionally, although the iPad Air 2 was expected to provide faster network connections and higher-quality images, use of an iPad Air 2 provided no observed benefit compared with an iPad 2, possibly indicating that the Internet connection and not the equipment was the limiting factor for maintaining a stable, high-quality connection.DiscussionWe investigated the technical and clinical feasibility of using a telepresence robot to evaluate home safety remotely. All therapists successfully completed the training course and remote home safety evaluations. This study showed that conducting home safety evaluations using a telepresence robot was feasible for occupational therapists who had no previous experience with the technology. One mock caregiver mentioned, "I felt like I was with the real person. The robot moved around the house as the real person is in the house." Our study results suggest the possibility of substituting on-site home safety evaluations with telepresence robot home safety evaluations. However, thematic analysis of the open-ended responses to barriers encountered during training and home safety evaluation identified several limitations, classified into three major themes: (a) poor robot driving technique, (b) equipment limitations, and (c) unstable network connections. These themes underline areas for targeted improvement to achieve effective implementation of home safety evaluations with telepresence robots.First, the participating occupational therapists had different levels of skill and confidence in operating the robot. They showed widely diverse driving ability during the remote home safety evaluations, especially among therapists of different age groups. For example, one therapist (40–50 years) reported difficulty switching camera views, judging distances, and turning the robot, whereas another therapist (20–30 years) had no difficulty maneuvering the robot. This may suggest generational differences in the ability to control a device remotely. Younger individuals who have experience using technology to control devices, such as remote-controlled cars, boats, or drones, and/or driving in virtual environments, such as with gaming, may have greater skills and confidence in operating the robot. Thus, the training protocol may need to be modified to address a variety of levels of skill in operating the robot. Interactive training may be a way to reduce the technical gap. Instead of providing a universal training protocol, tailoring training protocols to the individual's robot driving skill level may be more effective and efficient. Future studies are needed to explore a targeted training protocol tailored to various skill levels.Second, equipment limitations occurred. These included limited camera angles, blind spots, and nonoptimal image quality that obstructed thorough assessments of certain areas, such as narrow or small pathways, bathtubs, and corners. However, although certain features of the house were challenging to observe with the robot, the mock caregivers compensated for these problems. This indicates that the effectiveness and feasibility of this technology for a home safety evaluation depends on the presence of a caregiver during the home safety evaluation. Further, advances in robot equipment, such as enhanced image quality on the screen and improved maneuverability around furniture, could reduce the need for a caregiver. In fact, a new model of the Double Robot with new features, such as a wider-angle lens, better lateral stability, and high-definition capability, has been released since this study. A new feature, called the "always-on-floor view," provides users a dual-camera view that eliminates the need for frequent modification of the camera angle, which could solve some of the recurring issues that therapists encountered.Third, occasional network interruptions delayed the evaluation process. Four occupational therapists reported that audio or video lagged intermittently because of an unstable network. In one case, cell phone communication was needed for a few minutes because the occupational therapist could not talk to the mock caregiver to resolve the situation. Although these occasional network disturbances did not impede the completion of the remote home safety evaluations, stable and high-quality network connections, such as high-speed Wi-Fi, are highly preferred to provide a better experience for both the therapist and the client. Assessing current network speed and stability before conducting remote home safety evaluations may be necessary to minimize network interruptions. Also, providing an alternate mode of communication, such as a cell phone, is strongly advised in case of possible network interruptions. However, given recent technological advances in Internet speed, it is anticipated that the rapid expansion of area coverage will improve Internet connectivity and stability.Participants required minimal assistance from mock caregivers to complete the remote assessment. Mock caregivers mostly provided directions and described features of the home that were not easily captured with the robot. For example, image resolution was not adequate to allow participants to read information on the home thermostat. Also, in one case, the robot became stranded and needed to be lifted and relocated. Our findings suggest that a caregiver must be cognitively competent to maintain active communication with the remote therapist and provide accurate descriptions of the home. In addition, the caregiver must be physically capable of lifting and moving a 15-lb object. However, because the caregiver's levels of cognition, physical strength, and balance could affect the degree and quality of assistance provided, detailed guidelines to determine caregiver eligibility to assist in a remote home evaluation are needed.In general, the quality of the remote interaction via a 4G connection was sufficient for the assessments. This finding indicates that potential clients would not always be required to have a Wi-Fi high-speed Internet connection in the home to be eligible for remote home safety evaluations. This is particularly critical when trying to improve access to this intervention for people who live in remote areas where high-speed Internet may not be available or where there are no qualified professionals available to visit the home.Overall, the Double Robot provided accessible technology for home safety evaluations, and all of the therapists successfully completed remote home safety evaluations of the Gator Tech Smart House. This study showed that remote evaluations can provide an effective way to reach people, regardless of their geographic location. To conduct a successful remote home safety evaluation, three major conditions are recommended: (a) a stable network connection, (b) customized training in the use of the robot, and (c) the presence of an able-bodied caregiver. We used the Double Robot for our study. Other robot technologies are available, and different robots could result in different limitations, barriers, and degrees of feasibility. Future studies are needed to compare the feasibility of different types of robo