Concept for Fall Risk Interaction

Last modified by Figma System ⋅ 2 months ago

Introduction

Smart devices are widely available and serve a variety of purposes. This research investigates how users engage with these devices when in motion, particularly how and if they monitor their gait and how they utilise this technology to help them improve in real time. This research focuses on the gait for which a measure has been available since 2021 and the introduction of iOS 15.

Falls are believed to contribute to many injuries, particularly among older people (DuPree et al. 2014). This research wants to explore the idea that an improved smartwatch user experience could reduce the chance of injury. The information is relayed in real-time from a user’s mobile device, which can record their movement and suggest a course of action. The research could potentially have a more extensive use outside of older people and include those who may be fatigued or have an illness affecting their movement.

Therefore, the research is to be inclusive to a wide range of ages and abilities as there are potential gait issues which could be temporary, such as fatigue and medical conditions. For instance, research conducted in 2016 found that people with Fibromyalgia had poor Spatiotemporal gait parameters in a six-minute test (Heredia-Jimenez et al. 2016). Further suggesting potential applications for this research to impact a broader audience.

Background and context

Background

It is suggested that falls may be a significant contributor to pedestrian-only injuries and that falls are the second leading cause of unintentional injury death in the older population (Oxley et al. 2018). As indicated in the paper ‘Smartphone-Based Data Mining for Fall Detection: Analysis and Design’, fall detection promises to be important in healthcare (Hakim et al. 2017). Identifying gait improvements and anticipating falls could reduce the risk of injury. Similar research was conducted in 2011. A gait assessment model was used to predict the fall risk of someone walking. The authors later proposed using a wearable device to warn early of potential falls and slips (Jiang et al. 2011). Further research suggests that fall detection has generated much interest in the HCI space (González-Cañete and Casilari 2021). This implies the potential for this research to impact the wider community positively.

Existing interaction

In a 2016 paper titled: Exploring Experience of Runners with Sports Tracking Technology, the author notes user experience and running as an experience. One point of particular interest was how the user feels before and after using the product (Kuru 2016). Different smart device products are on the market, and this research will focus on the interaction between humans and devices. Previous research has explored HCI in digital wearable devices and how these could support different aspects of sports (Mencarini, 2022). These wearable devices vary in form and can include bodywear such as socks, t-shirts and shorts created from smart fabrics to more traditional smartwatches. It is suggested in a paper from 2019 that some devices, such as the Apple Watch, lack some functions related to form improvement (Han 2019). However, this paper was authored in 2019. Although the terms ‘gait’ and steadiness’ were not featured, the ‘lack of functions’ mentioned in the above paper may have since been introduced.

A general theme is that technology's capability in this area is growing and becoming more common. Within a 2021 paper, reference was made to the Apple ‘hard fall’ detection system, and deep learning models have proved ineffective in covering all patterns of falls. The authors found that using data collected from the user and an offline deep-learning model with simulated falls was effective. The authors detail how their fall detection application was run on a mobile phone, presenting problems; however, a watch UI would be less problematic for users. Interestingly, the proposed solution for automating this data was to use a cloud storage system (Couchbase 6) to collect and archive data periodically. This provides personalised fall detection and tracking datasets (Ngu et al. 2021). However, research undertaken in 2018 used an artificial neural network to detect falls, using a Raspberry Pi as an intermediary device to collect and transmit the data (Purushothaman et al. 2018). This relates to a newer paper from 2022 in which the researchers explored the use of the Internet of Things (IoT) in relation to wearable fall detection and highlighted the limitations of standalone detection systems. The authors also outlined a framework for how the detection system could connect to an IoT system and, by extension, the Internet (Qian et al., 2022). A consideration of real-time connectivity is response time (Fu and Wang 2022). The delay between the device collecting the data, transmitting the data to be processed and then the user receiving feedback to act could be limited by factors such as analysis time and even the device data speed.

Considerations when designing an interface for a smartwatch

When considering how the interaction could occur between the human and the computer (in this case, a smartwatch), the practicality of use whilst in movement, different methods have been explored in response to various scenarios such as temporary limitations such as carrying bags, to pushing a pushchair, to more permanent situations like motor impairment. The approaches to these constraints have been moving the forearm, tapping the foot, using the whole body, and even blowing onto the device (Heo et al. 2022). The method of physical interaction would need to simple and unimposing for use, if someone were in motion, they may not wish to halt their progress. This interaction can also be delivered as feedback to their wearable devices. It has been suggested that whilst real-time feedback is preferred, it can also create a dependency, potentially affecting how someone learns. The paper indicates that feedback should be specific to the measurement and that the frequency decreases over time (Van Hooren et al. 2020). Real-time feedback for the user to make corrections to their gait is an important consideration for the output of this research.

Methods

A simple survey was distributed to gather user insights and understanding of their activity level and device use. Observations will also be made to understand how people use their devices before, during, and after an activity. It is thought that there may be scope to develop a debriefing phase to the experience to inform future sessions. The tools will be Google Forms and Google Sheets, and data will be collected with pen and paper. The initial survey will be distributed online using the social media channels of LinkedIn, Facebook, and Instagram. The survey will target those over 18, and their activities will not be restricted. The survey will be distributed for ten days or until a sample of 25 people has been gathered. The data will then be analysed, and the results will be published. The survey on LinkedIn achieved 454 impressions in total; however, this did not convert to the desired number of participants. The movement was broken down into three categories for this research. Exercise consists of those who walk/jog/run for leisure/daily activities or fitness. Training is when you walk/jog/run as you are training for an event, or consider training for an event, and competitive is when you walk/jog/run as you regularly participate in competitive events.

Results

The research survey was live for ten days but did not yield the number of responses desired. However, this still has provided results that can be useful when developing a user interface. The survey aimed to understand the level of the movement undertaken. Two participants responded with others, which could be categorised into exercises which applied to 60% of the respondents. (Fig 1).

Fig 1. Do you consider yourself in the category of? Most participants were aware of their gait and said they would find a method of gait performance beneficial Fig 2. Similarly, the same proportion found a method of monitoring gait as beneficial (Fig 3). However, these figures may not reflect a larger portion due to the sample size.

Fig 2. Gait awareness

Fig 3. Would this interaction be beneficial Given the nature of the task being completed, feedback during the movement would best be received with the primary method of haptic vibration supported by a visual cue on the screen, as indicated below (Fig 4).

Fig 4. Interaction method The method of interaction and time for receiving notifications was as expected. Participants preferred the notifications occurred during or after the event (Fig 5)

Fig 5. When would notifications be most helpful The rationale behind the age ranges was to understand the device usage and preference. The limited sample size, whilst giving a variety of ages, was less diverse than desired for this research (Fig 6).

Fig 6. Age range This research is aimed specifically at a smartwatch; however, understanding which devices were being used by the audience could affect how the findings contribute to the final artefact. Most people carried mobile phones, and only a few wore smartwatches. (Fig 7).

Fig 7. Device usage Eight out of the ten participants provided commentary on their views on what they would like to see in a device which could help improve and monitor their performance. The results of this question generated some insightful answers (Fig 8).

Fig 8. Commentary

Discussion

The data gathered has presented insights into how devices are used and the nature of the activity undertaken. There was seemingly a correlation between those who were aware of their gait and finding a monitoring method beneficial to those who were unaware. However, this result could also show those who were knowledgeable and would not find a technique helpful. This could be investigated with further research. With that consideration, the artefact produced from this research could benefit from a brief initial interaction informing the user of the purpose and intent of the interaction.

The preference for receiving feedback whilst in motion was strongly indicated to be haptic and visual, with the notifications being delivered during and after the event rather than an interaction in anticipation of a movement event. How the haptic feedback offered is implemented would need to be well considered. For example, if it is continuous, it could become annoying. Also, users could become accustomed to and reliant on the feedback, making it less effective. (Van Hooren et al. 2020). Therefore, finding just the appropriate amount is vital in defining the process for triggering the feedback. The topic of this research is a growth area with advances being made, the IoT and Edge-Cloud computing being an area to utilise.

People were more likely to use a mobile phone rather than a smartwatch when undertaking an activity. The preference of the device could be affected by the type of activity the user is performing. There is scope within the artefact to include some information about gait and the impact it can have. This could be included in response to those unaware of gait, as it may help inform them of the benefits of such functionality.

For people who are not exercising but potentially have a higher chance of falls caused by poor mobility through age or illness, having their mobile phone as a support device could provide a better experience than using a device on the wrist.

Further Research

This research would have benefitted from a larger sample of participants in the survey, which was distributed through social media channels. Older people were identified as a critical demographic; however, those people may not have access to social media. Future research would consider alternate exploration routes, such as Doctors’ surgeries and charities such as Age UK. This would require further investigation. Future research would seek to use a more appropriate survey system to recruit people, including a longer survey window. Contextual inquiry using an ethnographic approach (Nunnally and Farkas 2016) (Kellogg n.d.) would have provided a richer and deeper understanding of current interactions. In hindsight, the survey target number of participants could have been increased, and the one featured in this paper could have been used to screen through to richer questioning, such as finding out more specifically about falls and balance. This could have also led to exploring the optimum frequency and how this process could be automated for a more immersive experience.

Conclusion

Falls are believed to contribute to injuries, particularly amongst older people significantly, but are not limited to this group alone. This research investigates if the interaction of a regular smartwatch could assist people by providing warnings by utilising the onboard sensors and haptic feedback to alert the user once a series of events have been triggered. Evidence was found that feedback would need to be appropriate to prevent the user from becoming annoyed with the volume of notifications or becoming accustomed to them that they go unnoticed (Van Hooren et al. 2020). This could be supported by building profiles to discover deeper insights into when a fall could occur, therefore increasing the accuracy of the notification timing. Some people may be unaware of what gait is and how technology could support them without impeding their daily activities. While this research focused on how a smartwatch interaction could meet this goal, incorporating the smartphone into this process could enhance the experience by using features such as a larger screen and the ability to handle datasets. There is evidence of expansion in this area of research area with technology such as the Internet of Things and edge-cloud computing.

References

FU, Dong and Jian WANG. 2022. ‘Design and Application of Yoga Intelligent Teaching Platform Based on Internet of Things’. Mathematical problems in engineering 2022, [online], 1–8. Available at: http://dx.doi.org/10.1155/2022/8407408. GONZÁLEZ-CAÑETE, Francisco Javier and Eduardo CASILARI. 2021. ‘A Feasibility Study of the Use of Smartwatches in Wearable Fall Detection Systems’. Sensors (Basel, Switzerland) 21(6), [online], 2254. Available at: https://www.mdpi.com/1424-8220/21/6/2254 [accessed 8 Nov 2022]. HAKIM, Abdul, M. Saiful HUQ, Shahnoor SHANTA and B. S. K. K. IBRAHIM. 2017. ‘Smartphone Based Data Mining for Fall Detection: Analysis and Design’. Procedia computer science 105, [online], 46–51. Available at: https://www.sciencedirect.com/science/article/pii/S1877050917302065. HAN, Barron. 2019. ‘Design and Implementation of Wearable Devices to Enhance Aquatics Physical Education’. [online]. Available at: https://iasdr2019.org/uploads/files/Proceedings/te-f-1332-Han-B.pdf [accessed 6 Oct 2022]. HEO, Seongkook et al. 2022. ‘No Need to Stop What You’re Doing: Exploring No-Handed Smartwatch Interaction’. [online]. Available at: https://www.benlafreniere.ca/assets/papers/nohand_GI17.pdf [accessed 29 Oct 2022]. JIANG, Susu, Bofeng ZHANG and Daming WEI. 2011. ‘The Elderly Fall Risk Assessment and Prediction Based on Gait Analysis’. In 2011 IEEE 11th International Conference on Computer and Information Technology. 176–80. Available at: http://dx.doi.org/10.1109/CIT.2011.82. KELLOGG, Judith S. Olson Wendy (ed.). n.d. Ways of Knowing in HCI. Springer New York. Available at: https://link.springer.com/book/10.1007/978-1-4939-0378-8 [accessed 20 Nov 2022]. KURU, Armağan. 2016. ‘Exploring Experience of Runners with Sports Tracking Technology’. International journal of human-computer interaction 32(11), [online]. Available at: https://www.researchgate.net/publication/304069245_Exploring_Experience_of_Runners_with_Sport s_Tracking_Technology [accessed 6 Oct 2022]. MENCARINI, Eleonora. 2022. ‘New Trends in HCI and Sports’. In Adjunct Publication of the 24th International Conference on Human-Computer Interaction with Mobile Devices and Services. 1–5. NGU, Anne H., Shaun COYNE, Priyanka SRINIVAS and Vangelis METSIS. 2021. ‘Collaborative Edge-Cloud Computing for Personalized Fall Detection’. In Artificial Intelligence Applications and Innovations. 323–36. Available at: http://dx.doi.org/10.1007/978-3-030-79150-6_26. NUNNALLY, Brad and David FARKAS. 2016. UX Research: Practical Techniques for Designing Better Products. ‘O’Reilly Media, Inc.’ Available at: https://play.google.com/store/books/details?id=sfOeDQAAQBAJ. OXLEY, Jennifer, Steve O’HERN, Duane BURTT and Ben ROSSITER. 2018. ‘Falling While Walking: A Hidden Contributor to Pedestrian Injury’. Accident; analysis and prevention 114, [online], 77–82. Available at: http://dx.doi.org/10.1016/j.aap.2017.01.010. PURUSHOTHAMAN, Anagha, K. V. VINEETHA and Dhanesh G. KURUP. 2018. ‘Fall Detection System Using Artificial Neural Network’. In 2018 Second International Conference on Inventive

Communication and Computational Technologies (ICICCT). 2018 Second International Conference on Inventive Communication and Computational Technologies (ICICCT), Coimbatore, April 2018. Available at: http://dx.doi.org/10.1109/icicct.2018.8473219. QIAN, Zhiqin et al. 2022. ‘Development of a Real-Time Wearable Fall Detection System in the Context of Internet of Things’. IEEE Internet of Things Journal 9(21), [online], 21999–2007. Available at: http://dx.doi.org/10.1109/JIOT.2022.3181701. VAN HOOREN, Bas, Jos GOUDSMIT, Juan RESTREPO and Steven VOS. 2020. ‘Real-Time Feedback by Wearables in Running: Current Approaches, Challenges and Suggestions for Improvements’. Journal of sports sciences 38(2), [online], 214–30. Available at: http://dx.doi.org/10.1080/02640414.2019.1690960.