Abstract
Phantom limb pain (PLP) after amputation is commonly experienced and can impact quality of life.[1] Management of PLP is challenging with few effective treatment options. One non-drug intervention is graded motor imagery (GMI) that includes 3-phases (i.e., limb laterality, motor imagery, and mirror therapy). Although clinicians report patient benefit from use of GMI, barriers exist to supporting at-home use of the intervention. This report focuses on refining a mobile app (VA-GMI) through serial cognitive interviews with 12 Veterans with transtibial amputation and PLP. Following each round of interviews, we summarized the major points to guide iterative changes in the mobile app. Our first round of interviews suggested that the Veterans find the VA-GMI app acceptable, but Veterans endorsed the need for added instructions. Concurrent with these enhancements, we worked to optimize the motor imagery and mirror therapy phases. We then invited Veterans to try the new version and repeated the process. Our second round of interviews suggested the need for additional education to support Veteran understanding of pain rehabilitation. This report describes the development and testing work to date using human centered design considerations. We anticipate we will reach consensus on the VA-GMI mobile app within the next iteration.
Keywords: transtibial, amputation, rehabilitation, pain, graded motor imagery
1. INTRODUCTION
Pain, including phantom limb pain (PLP), is commonly experienced following limb amputation. PLP is pain perceived from the limb that no longer exists and poses unique rehabilitation challenges. Graded Motor Imagery (GMI) is a non-drug, rehabilitation intervention for pain. GMI uses graded stimuli and therapeutic activities to gradually engage the nervous system while minimizing the pain experience. The goal is to “retrain the brain” and impact pain outside of treatment sessions to increase function long-term. The structured progression of GMI prepares the individual to tolerate and participate in more traditional pain therapies (e.g., manual therapy, gradual return to physical activity). GMI is proposed to mitigate activity-dependent changes following amputation.[2] GMI includes 3-phases: 1) limb laterality training, 2) motor imagery (e.g., visualizing a desired activity such as golfing by listening to a guided script), and 3) mirror therapy (i.e., observing movement using the visual illusion of an intact limb in a mirror) (Figure 1). GMI is introduced in the clinic and then intended to be one approach of an independent home program. Although GMI has been shown to be effective for post-amputation pain, clinicians report challenges to implementing a home program.[3] One potential solution is to develop a mobile app that could be used as a telehealth tool to support at-home use of GMI.
FIGURE 1:
GRADED MOTOR IMAGERY (GMI) PHASES OF INTERVENTION. (STOCK PHOTOS OR USED WITH PERMISSION).
To reach Veterans who may depend more on virtual care and develop tools that could support at-home practices, the purpose of this project was to develop and refine a GMI mobile app using an iterative design informed by end user pilot testing. Human centered-design is a cross-discipline approach that leverages stakeholder feedback to guide iterative design.[4] This paper represents progress to date on this project.
2. MATERIALS AND METHODS
2.1. Preparatory work to inform initial design
To understand the current state of the literature, we conducted a scoping review on the clinical evidence for GMI. That work is previously published and identified underutilization of limb laterality and motor imagery phases of the GMI intervention.[5] We then surveyed Department of Veterans Affairs (VA) amputation clinicians about their desires for technology using an electronic REDCap survey. Those results are previously published and indicated a desire for innovation in GMI.[6] We continued to explore the clinician stakeholder perspective by working with the VA Amputation System of Care that includes 26 Amputation Rehabilitation Coordinators nationally. The role of the Amputation Rehabilitation Coordinators is to focus on care coordination and clinical improvement initiatives that align with national VA Amputation System of Care standards. Prior to the inception of the GMI mobile app prototype, we held virtual meetings with the Amputation Rehabilitation Coordinators to learn more about their priorities for technology and the common barriers their patient population experiences when using GMI. The partnership with the Amputation Rehabilitation Coordinators was an iterative process with sharing initial ideas and feedback on the GMI mobile app and the different components. One example of the ongoing collaboration was writing the motor imagery scripts based on clinical experience and prior training programs. The motor imagery scripts were initially developed by the study team and the rehabilitation clinicians, who also have additional subject matter expertise as a certified Therapeutic Pain Specialists. The clinicians provided feedback and suggested changes for these recordings. This preparatory work was used to develop the initial prototype. These partnerships across the VA Amputation System of Care will serve as future beta testing sites and inform future large-scale implementation.
2.2. Design Considerations for Mobile App
The custom mobile application was developed using Unity 2022 (Unity Technologies, “Unity”, Version 2022 San Francisco, CA, 2022). The application offers a suite of features designed to enhance user interaction with GMI. The application consists of 3 main features. The first component, Laterality Identification, presents users with a dynamic series of limb images, the user must identify using buttons or swipe gestures. The second feature, Guided Audio, provides an auditory experience delivering calming exercises with customizable background soundscapes, including music or nature. The third feature, the Smart Mirror, leverages Unity’s AR Foundations framework to generate real-time, limb mirroring using the camera of the device. The user’s limb is mirrored but the background is not distorted, creating an immersive interaction platform. By combining these features, we were able to create a powerful platform to support user engagement and interaction with GMI.
2.3. Participant Recruitment for Testing GMI Mobile App
This study was reviewed and approved by the VA Institutional Review Board (#1775526). Our team used a Data Access Request Tracker and clinician referrals to aid in recruiting a diverse sample of 12 outpatient Veterans with respect to gender and race. A total of 254 Veterans were screened for eligibility and of those Veterans, 50 were eligible for recruitment. Inclusion criteria included the Veteran must be at least 18 years old, have a unilateral transtibial amputation, be at least one year since their amputation, have moderate or severe phantom limb pain (numeric rating scale ≥ 4 on a 0–10 pain intensity scale), stable medications over the past two weeks, be willing and able to give informed consent, and be able to participate in the telehealth study activities with a personal device. Exclusion criteria included unstable medical conditions (e.g., uncontrolled diabetes, heart failure exacerbation), unstable mental illness or substance use disorder (e.g., active suicidality or psychosis), or unable to give informed consent due to a cognitive impairment. For all eligible Veterans, the study team mailed recruitment letters and then proceeded with a follow-up phone-call if they had not heard back from the Veteran. Primary reason for not enrolling a Veteran in the study was due to no positive contact with the Veteran, followed by the Veteran reporting no phantom limb pain. Our target sample size for this user interface testing was 12 Veterans. We enrolled 12 Veterans to participate in repeated, individual video-recorded study sessions.
2.4. Participant Testing Procedures
The study design involved three sessions of cognitive interviews with the Veteran participants and the major themes from testing to guide iterative changes to VA-GMI (Figure 2). We developed a semi structured interview guide for the cognitive interviews using open-ended questions to understand ease of use, feedback on user interface, willingness to try, and appeal of each phase of the intervention. All study sessions were designed to be conducted remotely using video, allowing us to recruit Veterans who are unable to attend in-person visits. Participants were provided an iPad for trialing the mobile app. This work is the first aim of a larger project, and the final prototype will be tested in a small clinical trial that includes use of the mobile app as an intervention (clinicaltrials.gov# NCT06106984). Outcomes will include usage patterns, task success, and acceptability.
FIGURE 2:
CROSS SECTIONAL STUDY DESIGN.
We completed the development work for the initial mobile app prototype (Figure 3). This work included: 1) developing a library of limb pictures from staff volunteers with a wide range of skin tones, 2) adding custom drawn icons to ease home screen navigation, 3) simplifying the user interface for each phase of intervention, and 4) recording motor imagery scripts.
FIGURE 3:
PROTOTYPE FOR VETERAN TESTING. A) HOME SCREEN, B) LIMB LATERALITY, C) MOTOR IMAGERY, AND D) MIRROR THERAPY USING AUGMENTED REALITY. (PHOTOS WITH PERMISSION).
All Veteran participants completed baseline demographic information and patient-reported outcome measures (PROMs) to help describe the cohort. The selected PROMs reflect clinically relevant domains that impact pain including the Patient Reported Outcome Measurement Information System (PROMIS) Short Forms including Depression 4a [7], PROMIS Anxiety 4a [7], and Pain Self-Efficacy Questionnaire – 4[8], [9]. The PROMIS measures raw scores range from 4 to 20 points. The T-score indicates how the score relates to the normative sample with a standardized score with a mean of 50 and a standard deviation of 10. On the PSEQ, total scores range from 0 to 24 with a higher score indicating a higher confidence to cope and engage in activities despite pain. Prior research in chronic pain suggests that a score in the upper range is predictive of success with self-management approaches for pain.[10] Additionally, post-amputation pain (i.e., residual limb pain and phantom limb pain) and sensation (i.e., phantom limb sensation) were measured using the Amputation-Related Pain and Experiences.[11] The score for each post-amputation pain or sensation represents the average 0–10 numeric rating scale for 1 question of pain or sensation intensity and 2 questions on how pain interferes with enjoyment of life and daily activities.
The primary outcome measure was the Acceptability of Intervention Measure (AIM), secondary outcome measure was Feasibility of Intervention Measure (FIM).[12] Each of these measures is 4 questions with rating responses from 1 (Completely Disagree) to 5 (Completely Agree) with questions inquiring on perceived approval, appeal, and ease of use, among others. Total scores range from 4 to 20 points. To supplement these outcomes, we asked participants to complete ratings on the Mobile Application Rating Scale (MARS) that includes sections on engagement, functionality, aesthetics, subjective quality scale, and app-specific.[13] We omitted the MARS scale on information quality (i.e., use in clinical trials, description on the app store) due to this early development phase. Each item on the MARS was rated from a 1 (Inadequate) to a 5 (Excellent). The score for each section is totaled and averaged.
2.5. Data Analysis
All sessions were conducted virtually, recorded, and transcribed. Analysis of the transcripts was guided using rapid turnaround methods that includes identifying neutral domains derived from the interview guide.[14] A summary template is then applied to each transcript. The team then reviews the themes across participants to identify key improvements. The responses for the AIM and FIM are averaged with a high average indicating a positive response to the intervention. The responses for each section of the MARS are averaged with a high average indicating a positive response to the intervention.
3. RESULTS AND DISCUSSION
Twelve Veterans were recruited for the study with a wide range of self-reported technology experience (Table 1).
TABLE 1:
BASELINE DEMOGRAPHIC AND PATIENT REPORTED OUTCOME MEASURES.
| Sample (% of overall sample) |
|
|---|---|
| Age | |
| 40–65 years old | 9 (75%) |
| >65 years old | 3 (25%) |
| Gender | |
| Female | 2 (17%) |
| Male | 10 (83%) |
| Race | |
| White | 9 (75%) |
| Asian | 1 (8%) |
| Black | 1 (8%) |
| Native American | 1 (8%) |
| Other | 0 (0%) |
| Ethnicity | |
| Non-Hispanic | 12 (100%) |
| Hispanic | 0 (0%) |
| Side of amputation | |
| Left transtibial | 8 (67%) |
| Right transtibial | 4 (33%) |
| Time since amputation | |
| 1–5 years | 3 (25%) |
| >5 years | 9 (75%) |
| Cause of amputation | |
| Trauma | 10 (83%) |
| Dysvascular | 2 (17%) |
| Self-reported diagnosis of diabetes prior to amputation | |
| Yes | 5 (42%) |
| No | 7 (58%) |
| Prosthesis User | |
| Yes | 12 (100%) |
| No | 0 (0%) |
| PROMIS Depression - 4a (Average total score, T-Score, Standard error) | 7 (53.9)(2.4) |
| PROMIS Anxiety – 4a (Average total score, T-Score, Standard error) | 7 (53.7) (2.8) |
| PSEQ-4 (Total Score, standard deviation) | 18 (5.2) |
| ARPE: Residual Limb Pain (Average score, standard deviation) | 4.1 (3.1) |
| ARPE: Phantom Limb Pain (Average score, standard deviation) | 4.6 (2.8) |
| ARPE: Phantom Limb Sensation (Average Score, standard deviation) | 4.9 (2.0) |
Note: ARPE: Amputation-Related Pain and Experiences; PROMIS: Patient Reported Outcome measurement Information System; PSEQ-4: Pain Self-Efficacy Questionnaire – 4. The mean baseline score for each patient-reported outcome measure is reported.
The analysis of the cognitive interviews guided our development work. Most Veterans found the VA-GMI mobile app acceptable, easy to use, and had high expectations for the possibility of the app. There was strong interest in the augmented reality feature of the mirror therapy phase with several Veterans reporting the perceived instant gratification of this phase and a positive interaction with their phantom limb moving in a healthy way. One participant stated, “It felt like a little victory like I’m moving it [limb on amputated side] again”.
We found that participants had some expectation that the app would be effective in managing PLP and that even small improvements were meaningful. As one participant stated, “If it gives me even 1–2% less phantom pain, I’d be happy with that because anytime you take a little bit of phantom pain away from what that person is dealing with, it helps their daily activities, their daily life. So, if you can take 1–2% away, that’s a win-win.” Although instructions were provided in the form of a user manual and verbally during testing sessions with the study team, Veterans requested further instructions to set expectations and prepare for each phase. Cognitive interviews were supplemented with the primary and secondary outcome measures reported in Table 2.
TABLE 2:
CHANGE IN ACCEPTABILITY OF INTERVENTION MEASURE (AIM), FEASIBILITY OF INTERVENTION MEASURE (FIM), AND MOBILE APPLICATION RATING SCALE (MARS) ACROSS ITERATIONS.
| Score | |
|---|---|
| AIM | |
| Prototype 1 Mean | 4.4 |
| Prototype 2 Mean+ | 4.4 |
| Change Score | 0.0 |
| FIM | |
| Prototype 1 Mean | 4.2 |
| Prototype 2 Mean+ | 4.5 |
| Change Score | 0.3 |
| MARS | |
| Section A: Engagement | |
| Prototype 1 Mean | 4.0 |
| Prototype 2 Mean+ | 4.0 |
| Change Score | 0.0 |
| Section B: Functionality | |
| Prototype 1 Mean | 4.2 |
| Prototype 2 Mean+ | 4.3 |
| Change Score | 0.1 |
| Section C: Aesthetics | |
| Prototype 1 Mean | 4.2 |
| Prototype 2 Mean+ | 4.2 |
| Change Score | 0.0 |
| Section E: App Subjective Quality | |
| Prototype 1 Mean | 4.0 |
| Prototype 2 Mean+ | 4.2 |
| Change Score | 0.2 |
| Section F: App-Specific | |
| Prototype 1 Mean | 4.2 |
| Prototype 2 Mean+ | 4.1 |
| Change Score | −0.1 |
Note: AIM: Acceptibility of Intervention Measure, FIM: Feasiblity of Intervention Measure, MARS: Mobile Application Rating Scale. MARS Section D (Information, e.g., app store description, use in published scientific research) was omitted due to the development phase of this work.
11 of 12 participant responses to date.
Following the first round of testing, usability improvements included: 1) adding instructions both written and recorded to guide the user experience, 2) increasing sensitivity for swiping in laterality selection and adding UI elements to alert user of not swiping far enough (Figure 4), 3) enhancing motor imagery scripts with professional recordings (Figure 5), 4) allowing for customization features (e.g., selecting background pictures, music, or nature sounds) and 5) optimizing the mirror augmented reality to reduce previous white outline on the mirrored phantom limb.
FIGURE 4:
PROTOTYPE 2 LIMB LATERALITY ENHANCEMENTS INCLUDING A) AUDIO INSTRUCTIONS, B) CUSTOMIZEABLE BUTTON OR SWIPING SELECTION, C) VISUAL FEEDBACK TO SWIPE FURTHER AND D) ADVANCED LATERALITY TRAINING WITH ADDED CONTEXT.
FIGURE 5:
PROTOTYPE 2 MOTOR IMAGERY ENHANCEMENTS INCLUDING A) INSTRUCTIONS, B) EXPANDED SCRIPT OPTIONS, C) CHANGE IN BACKGROUND IMAGES, AND D) CUSTOMIZEABLE BACKGROUND IMAGE BASED ON PREFERENCE.
Optimizing the augmented reality was achieved by editing the overlay shader used by the application to mirror the detected body parts. The overlay shader converts the detected human body segments in the camera view into a usable texture for the mirroring processing. The processing effect uses features exposed from the AR Foundations and AR Subsystem framework made available in the Unity development platform. The framework exposes device specific AR toolkits such as ARKit (Apple) and ARCore (Google).[15] The AR framework implementation interfaces with a camera and depth manager as well as an occlusion manager to detect and mirror the human overlay segment. Once the human occlusion segment is detected the application creates a mirrored texture to overlay onto the camera feed with each frame. With the original implementation of this process there were some artifacts as the mirror texture can have a padding around the detection human segment. Through testing and debugging the padding around the texture was reduced making a more visual appealing and immersive experience for the user.
To date 11 of 12 Veteran participants have completed qualitative interviews with prototype 2. One Veteran did not respond to outreach to schedule the study visit for prototype 2 for unknown reasons. Anticipated major changes include education, additional script development for motor imagery, and minor user interface enhancements. For education, Veterans stated not understanding the “why” behind the phases might contribute to getting frustrated or apprehensive without knowing the intent of the intervention. Veterans preferred a mix of patient-education formats including short videos, audio recordings, or a handout (if used in a clinical environment). These educational components will focus on the conveying the purpose behind the exercises to build the Veteran’s understanding of GMI as a pain treatment technique.
Overall, the Veterans responded positively to the professionally recorded motor imagery scripts with the expansion from one script to seven scripts. Several Veterans experience PLP in the evening while trying to fall asleep or waking in the night. Most Veterans in this sample stated interest in additional sleep scripts or shorter scripts focused solely on breath work. Collectively, these scripts work to influence the nervous system and are a technique used clinically to treat pain.
Throughout Veteran testing, the study team observed user errors. For example, one customization feature in motor imagery allows for picture selection and audio control. If both customization menus are launched at the same time, closure of these screens must be done sequentially. This potential for user error will be addressed in prototype 3. We have received minor user interface enhancements and anticipate adding those (e.g., including more music and nature sound options for the motor imagery, consistency in button naming, and simplifying the audio controls. Several options within motor imagery were identified that could be “hidden” from the user’s screen to further simplify the user interface. For example, the progress bar for the duration of the motor imagery recording and the ability to skip ahead or behind 10 seconds are features that are not needed.
We anticipate we will reach agreement with the final planned prototype once these enhancements are completed. This prototype will be used for a small clinical trial. Despite initially having a literature review and clinician subject matter experts to contribute to the design of VA-GMI mobile app, the human centered design approach proved to be critical in achieving a Veteran-centric intervention.
4. CONCLUSION
We used a human centered design approach with iterative prototyping to develop a mobile app for GMI. Repeated meetings with Veteran stakeholders were useful to identify improvements for this app. This work not only informed potential future applications but also informed future clinical trial design for acceptability and feasibility. Future clinical trials will include objective measures of app performance (e.g., usage patterns, completion rates, and task success).
ACKNOWLEDGEMENTS
Dr. Rich’s work is supported in part by the U.S. Department of Veterans Affairs Rehabilitation Research, Development, and Translation Career Development Award IK2RX004805-01A1. The views, opinions, and interpretations expressed in this article are those of the authors and do not represent the views of the U.S. Department of Veterans Affairs or the U.S. Government.
NOMENCLATURE
- AIM
Acceptability of Intervention Measure
- FIM
Feasibility of Intervention Measure
- GMI
Graded motor imagery
- MARS
Mobile Application Rating Scale
- PLP
Phantom limb pain
- PROM
Patient reported outcome measure
- PROMIS
Patient Reported Outcome measurement Information System
Contributor Information
Tonya L. Rich, Minneapolis VA Health Care System; University of Minnesota, Minneapolis, MN, USA
Katharyn G. Cristan, Minneapolis VA Health Care System, Minneapolis, MN, USA
Timothy P. Truty, Minneapolis VA Health Care System, Minneapolis, MN, USA
Andrew H. Hansen, Minneapolis VA Health Care System; University of Minnesota, Minneapolis, MN, USA
Princess E. Ackland, Minneapolis VA Health Care System; University of Minnesota, Minneapolis, MN, USA
REFERENCES
- [1].Padovani MT, Martins MRI, Venâncio A, and Forni JEN, “Anxiety, depression and quality of life in individuals with phantom limb pain,” Acta Ortop. Bras, vol. 23, no. 2, pp. 107–110, 2015, doi: 10.1590/1413-78522015230200990. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [2].Moseley GL, “Graded motor imagery for pathologic pain: A randomized controlled trial,” Neurology, vol. 67, no. 12, Art. no. 12, Dec. 2006, doi: 10.1212/01.wnl.0000249112.56935.32. [DOI] [Google Scholar]
- [3].Limakatso K, Madden VJ, Manie S, and Parker R, “The effectiveness of graded motor imagery for reducing phantom limb pain in amputees: a randomised controlled trial,” Physiotherapy, vol. 109, pp. 65–74, Dec. 2020, doi: 10.1016/j.physio.2019.06.009. [DOI] [PubMed] [Google Scholar]
- [4].Göttgens I and Oertelt-Prigione S, “The Application of Human-Centered Design Approaches in Health Research and Innovation: A Narrative Review of Current Practices,” JMIR MHealth UHealth, vol. 9, no. 12, p. e28102, Dec. 2021, doi: 10.2196/28102. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [5].Falbo KJ, Phelan H, Hackman D, Vogsland R, and Rich TL, “Graded motor imagery and its phases for individuals with phantom limb pain following amputation: A scoping review,” Clin. Rehabil, vol. 38, no. 3, Art. no. 3, Mar. 2024, doi: 10.1177/02692155231204185. [DOI] [Google Scholar]
- [6].Rich TL et al. , “Clinician perspectives on postamputation pain assessment and rehabilitation interventions,” Prosthet. Orthot. Int, Oct. 2023, doi: 10.1097/PXR.0000000000000284. [DOI] [Google Scholar]
- [7].Kroenke K, Yu Z, Wu J, Kean J, and Monahan PO, “Operating Characteristics of PROMIS Four-Item Depression and Anxiety Scales in Primary Care Patients with Chronic Pain,” Pain Med, vol. 15, no. 11, pp. 1892–1901, Nov. 2014, doi: 10.1111/pme.12537. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [8].Chiarotto A et al. , “Responsiveness and Minimal Important Change of the Pain Self-Efficacy Questionnaire and Short Forms in Patients With Chronic Low Back Pain,” J. Pain, vol. 17, no. 6, pp. 707–718, Jun. 2016, doi: 10.1016/j.jpain.2016.02.012. [DOI] [PubMed] [Google Scholar]
- [9].Rich T et al. , “Examining patient reported outcome measures for phantom limb pain: measurement use in a sample of Veterans with amputation,” Disabil. Rehabil, pp. 1–9, May 2024, doi: 10.1080/09638288.2024.2356017. [DOI] [Google Scholar]
- [10].Asghari A and Nicholas MK, “Pain self-efficacy beliefs and pain behaviour. A prospective study,” Pain, vol. 94, no. 1, pp. 85–100, Oct. 2001, doi: 10.1016/S0304-3959(01)00344-X. [DOI] [PubMed] [Google Scholar]
- [11].Falbo KJ et al. , “Development and pilot administration of the amputation-related pain and sensation assessment tool,” Disabil. Rehabil, pp. 1–10, Jul. 2024, doi: 10.1080/09638288.2024.2374489. [DOI] [Google Scholar]
- [12].Weiner BJ et al. , “Psychometric assessment of three newly developed implementation outcome measures,” Implement. Sci. IS, vol. 12, no. 1, p. 108, Aug. 2017, doi: 10.1186/s13012-017-0635-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [13].Stoyanov SR, Hides L, Kavanagh DJ, Zelenko O, Tjondronegoro D, and Mani M, “Mobile App Rating Scale: A New Tool for Assessing the Quality of Health Mobile Apps,” JMIR MHealth UHealth, vol. 3, no. 1, p. e27, Mar. 2015, doi: 10.2196/mhealth.3422. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [14].Beebe J, Rapid assessment process: an introduction. Walnut Creek, Calif.: AltaMira Press, 2001. [Google Scholar]
- [15].Unity Technologies, AR Foundation. ((n.d.)). Accessed: Jul. 10, 2024. [Unity]. Available: com.unity,xr.arfoundation@6.0 [Google Scholar]