Article Preview
TopIntroduction
The percentage of elders’ population over and above the age of 60 is increasing a lot in the last 10 years and this will continue to increase for another 20 years (WHO, 2015). In the year 2015, the elders’ population was around 900 million in the world which is expected to increase triple by 2050 (WHO, 2015). As the age of elders increases, they face more challenges and issues to live independently. This makes the family members to take up the role of informal caregivers to provide them care and help to do their basic daily activities. These challenges attract the researchers of Ambient Assisted Living (AAL) over the last decade to provide some creative solutions for ensuring the safety and health quality of the senior citizens in various care needs such as Human Activity Recognition, Fall detection, Alzheimer’s disease detection, blind support etc.,
Human Activity Recognition (HAR) now plays a vital role in the field of AAL for the welfare of elders (Zdravevski et al, 2017) as most of the elders live independently. AAL comprises pervasive and ubiquitous hardware components termed as HAR devices that provide services to the elder with disabilities or for adults who cannot or who choose not to live independently. In general, HAR components gather signals from pervasive and ubiquitous devices such as ambient and wearable sensors to be processed through any Machine Learning (ML) algorithms for activity recognition (Ramanujam and Padmavathi, 2019). The HAR system recognizes multiple activities in various domains of applications such as intelligent transportation, education and Healthcare (Wang et al, 2019). In the field of Healthcare, HAR has the greatest impact on recognizing Activities of Daily Living (ADLs) such as basic, complex and postural transitions. These activities are categorized based on the duration of seconds. For instance, basic activities are characterized by a longer duration and can be either dynamic or static (e.g running and reading). Complex activities are the extension of basic activities where the subjects interact with any other physical objects may be like playing sports, etc. Finally, the Postural transitions comprise of transition from one gesture to another gesture such as sit-to-stand, stand-to-sit, etc. More number of research works has been proposed in the last decade for the HAR (Hernandez et al, 2020) with elders.
In the last decade, more research works have been proposed for HAR as reviewed in a survey done by (Hernandez et al, 2020). Mubashir et al, (2013) have categorized HAR systems into three such as Ambient, Wearable, and Vision systems. Many research works (Seneviratne et al, 2017) has suggested that wearable devices are appropriate in most of the emergency situation. The success of this device purely depends on user involvement such as pressing the button of the device, etc. To be specific, the wearable devices have to be worn throughout the day by the elder and need to be charged regularly. The elders may forget to do those due to their age factor. On the other side, ambient sensors (Wang et al, 2019) are costlier and not feasible to install all over the actuation area of the elder. Moreover, the maintenance cost seems to be very high and it makes the elder live only in the controlled environment. In the case of vision systems (Jegham et al, 2020), the surveillance cameras highly interrupt the seniors from their personal privacy and security.