Outcomes

The multidimensional data were then synthesized and visualized to convey the everyday struggles that older adults face at home.  The goal of our work was to combine our various data sources together so that we could better understand these sources in context.  While graphing biomarkers can show when events occur, traditional graphs do not provide information on where these events occur.  By combining our various data sources together we can analyze the ways in which participants utilize their environments and how their resulting biomarkers are affected.

To achieve this, an interactive visualization system was built inside of the Unity3D game engine. This visualization was able to show how participants interacted with their environment alongside their biomarkers during the observation period.  The visualization used simplified 3D characters to showcase the participant’s position in space and sliders demonstrated different biomarkers.  The analyst was able to see the coded timeline of the different spaces the participant had traveled to and was able to play or scrub the visualization to be at any point during this timeline.  Finally, to add a greater context to the environment, quotes were placed to the spaces which were referenced during the interview process.  An example of the visualization is shown below.

For certain participants, the home modifications provided data that supported positive health outcomes. For example, when tracking the participant in the space before the intervention we see that they spent the majority of their time in bed.  However, after the intervention, the participant was much more active in the space, with the majority of their time being spent in the living room areas.

Despite this increased mobility, the average heart rate of the individual was not increased. Further analysis can be performed by analyzing the rate against movement around the space (shown below). As expected, a trend is shown between the movement and heart rate, with increased movement leading to increases in heart rate. Using this trendline, we can estimate a resting heart rate for the point at which movement goes to zero. We see that while the average resting heart rate was at approximately 92 BPM in the pre-modification condition, the average heart rate dropped to 72 BPM post-modification. This is due in large part to the reduction of high heart rate activities during periods of low activities. This reduction in resting heart rate was found in half of the participants analyzed.

The integrated Virtual Reality (VR) environment provides ready access to highly precise spatial information of older adults’ homes and their behavioral and health outcomes as they occur within their life spaces. Our data synthesis and analysis demonstrated that our data collection and fusion method across multiple platforms is not only feasible but also valid.

 

The dataset may offer a valuable platform for researchers, educators, and clinicians where they can explore older adult’s life spaces, pose important research questions, and answer them. For example, researchers interested in the impact of home modifications on older adults’ functional independence, satisfaction, activity levels, and acute stress levels can examine the data set, formulate research hypotheses, and statistically test them. Housing professionals concerned about accessible housing can explore the life space of our participants and learn about where common problems occur. Health care professionals can learn how older adults use their spaces and how they might deploy home health care strategies in the best way possible.

While inaccessible homes pose a big challenge to older adults, modifying them to their needs is even more challenging in today’s complex healthcare system and housing industry. The home modification industry as a whole is highly fragmented between healthcare professionals, home assessment, and handyman services. Navigating between them while dealing with health conditions is not easy. Mistrust and lack of knowledge prevails.

For the research team, navigating this fragmented system to recruit and contact participants was very challenging. High cost of home modifications also limited the extent with which we could modify their homes to the fullest extent. This in turn reduced the potential health benefit of home modification. The home is intrinsically personal and that has a life’s long accumulated meaning to older adults. Changing to their medical needs was not always compatible with their ideal image of home, making them resist some of key recommendations.

An additional challenge came from trying to track the individual in their home.  While GPS systems have provided a proven way to track location in outdoor spaces, these technologies do not work while indoors. The team explored a variety of different technological solutions, from the use of Bluetooth beacons, to fingerprinting of WIFI routers. While these solutions showed some early promise, it was determined that their accuracy was not yet reliable enough to be used in the field.

In turn the team turned to utilizing a wearable camera that would capture an image on a 30 second interval for 24 hours continuously.  Each image was then referenced against other images of the home to determine the location of the individual.  This method had several challenges as images angles sometimes made locations difficult to decipher.  Furthermore, periods of continuous movement would result in rooms being overlooked due to the low sampling rate. Finally, the participants sometimes felt uncomfortable wearing the camera, leading to additional form of potential data loss.

Despite these challenges, the context provided by the location alongside the biomarkers was seen as providing new insights into participant’s health in their home.  Future work will aim to provide this data in a more reliable, less invasive and more automated fashion.

The biggest challenge of course was the COVID19 pandemic. The yearlong field data collection came to a complete halt in 8.5 months. The research team invested the rest of the project period on a high-level visualization.