How INSIGHT wearable unlocks the full power of a pressure map.
Back when the first INSIGHT™ prototypes started producing pressure maps, we and our partner prosthetists were faced with the challenge of understanding exactly what was upon us. Mario, one of our co-founders, used to say that we were in this curious situation somewhat analogue to having all the technology required for radiology, but no radiologists to make sense of it.
So, we were faced with the pressing need of providing context to the pressure data being gathered. Our first take on it was using video. The first user-facing version of INSIGHT™, pictured below, required the prosthetist to simultaneously capture video while pressure data was being acquired.
The first versions of INSIGHT™ collected pressure data and acquired video simultaneously. It was neither the prettiest nor the most intuitive interface, but hey!, every butterfly was once a caterpillar
Like most first takes, it ended being both sub-optimal and a great learning opportunity:
if the recording device (at the time, a computer) was placed in a fixed position, logistical challenges in the capture of the patient’s entire pathway were introduced.
if the prosthetist was required to actively engage in video recording, then he would be taking his attention off the patient - something totally against our core product design values.
after the acquisition was done, no summary results were presented: the only way to extract knowledge was to review it in its entirety.
At that point, we knew we had to do better.
INTRODUCING AUTOMATED GAIT ANALYSIS
One of our first insights (eheh) regarding socket pressures was that they are not inherently bad. In fact, they are expected in certain socket locations during certain (load-bearing or not) events of the gait cycle. Their absence (orpresence in unexpected moments/locations) is often a marker of an improperly fit socket. With this information, it became clear to us that the best possible context for pressure information was the gait cycle itself. More concretely, the phases of the gait cycle.
Thanks to some clever foreshadowing during our hardware design stages, INSIGHT™ Wearable was already equipped with an IMU, an inertial sensor, capable of measuring linear acceleration and rotational velocity. To achieve the gait analysis performance we were striving for, we required additional biomechanical information - and so INSIGHT Wearable™ IMU, our smaller device, came to be. The information collected by these two units is the cornerstone of our gait characterization system which, in a fully automated and passive way, detects the patients’ turns, strides and, inside each stride, divides it in the 5 key gait phases: Loading Response, Midstance, Terminal Stance, Pre-Swing and Swing.
KNOWLEDGE IS A MATTER OF PERSPECTIVE
The development of this gait analysis system involved the collection of considerable amounts of patient data and the development of a custom pipeline mixing traditional signal processing and state-of-the-art Machine Learning modules. After successfully building it, challenges still remained: we had the pressure data and the gait phases data, but how to best combine and present it?
With video recording out of the equation (well in line with our value of solidarity towards clinical time), while acquiring only an animation of the pressure over a fully interactive 3D model of the socket is shown. The prosthetist is invited to manipulate and explore it whenever intrigued with a particular pressure pattern or spot, but he can at anytime take his eyes off the iPad and refocus on the patient. In the background, INSIGHT will keep collecting data to guarantee nothing is missed. Less than 5 seconds after a dynamic acquisition is finished, three different perspectives over the collected data, designed to support all kinds of analyses, will be available:
Average Pressure per Gait Phase
The so-caled Processed Data perspective grabs the most relevant patient strides (purposedly excluding outlier ones, as well as starting, stopping and turning steps) and averages the pressures in each gait phase of those strides - summarizing each gait phase through a single view.
The end result, the average pressure inside the socket for each gait phase, captures the overall trends of pressure distribution in the most relevant moments of the patient’s gait cycle.
Gait, however, is not merely a set of static moments. It’s a continuum during which socket and residual limb interact to accomplish functional tasks related to weight acceptance, weight shifting, limb support and limb advancement. To visualize how socket pressure relates to these dynamics, the Average Stride perspective exists.
Again agglomerating data from all relevant strides, it allows to smoothly navigate through a patient’s average stride, seeing how the pressure distributes itself as gait phases change and the different gait’s functional tasks are tackled. Additionally, it is also possible to get a glimpse at the relative duration of each gait phase, often correlatable with the patient’s confidence in the prosthetic device (from our experience, for instance, less confident amputees often shorten their Mid and Terminal Stance phases and opt for an extended Pre-Swing, where double support is regained).
This perspective is the result of some very insightful days spent in beautiful Prague, at the 27th Annual Meeting of the European Society for Movement Analysis in Adults and Children (ESMAC 2018).
The final perspective, Raw Data, directly follows from our mantra of keeping the focus on the patient. It allows to review, in its entirety, the dynamic acquisition. The same pressure data collected while the patient was walking is always available for review, giving the prosthetist the confidence to take his eyes off the iPad at any time and interact with the patient.
Unlike the other two perspectives, here the data is shown completely unprocessed, with no calculation whatsoever of aggregated metrics. In this way, if required, a specific instant of the acquisition can be easily analyzed. In the absence of video and to provide some context, we have our very own small 3D human (lower right corner), which mimics the patient’s actions: walking, turning or stopping.
These three perspective over our pressure maps serve totally different purposes. However, all three share the very particular property of transforming the numeric values from our pressure and inertial sensors into clear visuals that will hopefully lead to actionable knowledge and, ultimately, a better socket.
As INSIGHT™ evolved, so did the importance of our Wearable devices: from a mere relay of pressure data, they evolved into fully-featured biodata gathering devices, central to all the knowledge the system provides. In a follow-up blog post, we will explore how we applied the same user-centered design principles used in the Scanner’s design to come up with a myriad of small usability features that, when taken together, make using the Wearable simple and seamless.
