Project History

​

Initially, in our instrumentation development efforts we pursued means to detect walking speed (2000), distal limb position (2006), sock use (2012, 2013.2), and user activity (2013.1). These efforts resulted in functional technologies but we lacked a cohesive plan for how to implement them in clinical care. In 2014, Professor Brian Hafner (Rehabilitation Medicine, University of Washington) with us published a paper presenting a strategy for how prosthesis sensors might fit into three important aspects of rehabilitation - training, diagnosis, and prognosis (2014). We focused on creating a sensing system that has the potential to provide important clinical insight during all three phases.

​

In a subsequent study, we effectively used proximity sensors within the socket and accelerometers on the limb to detect user activity (2016.2). This system effectively distinguished doffed, walking (including all types of movement), standing, and sitting. However, the thigh accelerometer, which was attached to the liner over the anterior thigh, was bothersome to some participants. Interested in extending to a more detailed characterization of the limb-socket interface, we placed various sensors within 3D printed inserts that we custom-designed for each participant (2016.1, 2017, 2018.2). This system allowed flexibility in sensor selection and placement. Using different sensing modalities simplified interpretation of field collected data. An inductive sensor that we placed in the distal region of the socket provided promising insight towards monitoring sock use. We believe that the printed insert technology has a place in early fitting assessment since during this time the practitioner changes the socket shape frequently, and the capability to frequently manufacture new instrumented inserts would be a benefit. Our focus shifted to improvement of the promising inductance sensing modality, and we created a novel inductive sensing system for further investigation. Using the sensor on participants with a definitive prosthesis, we taped the sensor antennae to the inside of the socket. A conductive fabric target was adhered to the outside of the liner. We implemented this system to monitor user activity in multi-week take-home testing (2018.1). In a study on 21 participants, we characterized prosthesis day duration, how often people changed their sock thickness, and how long they temporarily doffed their socket during the day. While the sensor was effective, the target, the conductive fabric, was prone to degradation which electron microscopy revealed was due to microcracks within the conductive fibers. The cracks formed as a result of the mechanical stress from the residual limb. Using a modified version of the sensor that eliminated this problem during two weeks of take-home use, Professor Geoff Balkman (Rehabilitation Medicine, University of Washington) in a study on three participants found substantial differences between prosthetists' estimates of their patients' prosthesis use and data measured by the monitor, alluding to the well-established challenges in obtaining accurate information related to patient activity (2019.1). We concluded that a better understanding of patients' true prosthesis use and activity can facilitate an evidence-based approach to clinical practice and mitigate prosthetists' reliance on patient-reported history.

From 2018 to 2020, we reported on several iterations of a new target design, a liner-embedded target, that served as a more seamless and durable sensing element (2018.3, 2018.4, 2019.2, 2020). The target material is a small amount of iron powder that is embedded beneath the surface; it does not contact the user. This design differed from our previous effort in that the target was a magnetically permeable material instead of a conductive material. This change eliminated the microcrack problem and its associated durability issue described above. When used with a socket that had the antennae taped to the surface (see socket figure above), the new liner proved considerably more durable and a reliable signal was obtained. In a study on four participants, we demonstrated solid detection accuracy for identification of sit, seated shift, stand, standing weight-shift, walk, partial diff, and non-use (full doff) (2022.1). It was clear that step count may provide an incomplete picture of prothesis use, and larger studies should be pursued to investigate how bodily position and type of activity may facilitate clinical decision-making. In a subsequent study collecting data from 14 participants in their at-home environment, we were able to distinguish walking bouts from low commotion activity, helping to distinguish participants who used their prosthesis to move around in a confined space from participants who walked long distances (2022.3). This distinction may help practitioners to better identify their patients' componentry needs.

​

We approached prosthetic liner manufacturers to see if they could manufacture the ferrous liners, i.e., place a trace amount of iron powder in their product, for our research projects. One company was able to do so, and this advance has helped simplify our clinical study execution. We continue to make ferrous liners in our lab for participants who cannot wear the company's type of liner. Also, we enhanced the sensor sockets by embedding the antennae within the socket during fabrication (2022.2), an advancement that has essentially eliminated sensor failure problems. Testing the new system in participants with transtibial amputation revealed that as a socket was slowly enlarged, the sensors picked up the early signs it fit deterioration well before an experience research prosthetist was able to visually detect a change, and well before the participant indicated a sock ply change was needed (2021.1). Our current efforts are directed towards use of the limb motion sensing system in clinical prosthetics for training, diagnosis, and prognosis purposes.

​

In parallel with creating a technology to monitor motion at the limb-socket interface, we developed a sensor to detect locking pin vertical motion (2021.2). Similar to the limb-socket interface technology, this sensor uses an inductive sensing modality. We recently improved the versatility of the pin motion sensor and used it in a study on three participants during week-long take-home use. Results are forthcoming (2022.4).