gms | German Medical Science

Background: As to the older persons discharging from hospital, they are often weak, lacking in mobility, suffering from complications and at risk of deterioration [1]. The traditional approaches to measure their rehabilitation and health status are clinic visits and medical assessments, which is costly and can only provide snapshots of patient’s health status rather than on-going measurement. In the context of ambient assisted living, sensor-enhanced information system [2] is expected to offer continuous and automatic assistance. Because of the acceptance and privacy issues, unobtrusive measurements have to be relied on for practical usage in real-life scenarios. It is still challenging to extract useful information from unobtrusive measurements.

Objectives: The objective of this research is to propose methods to discover human lifestyle based on sensor-enhanced living environments, so as to provide complementary information to clinicians or nurses for health assessments.

Material and Methods: In sensor-enhanced living environments, corresponding sensors can be triggered when the residents perform movements in specific areas or operations on specific objects. For events extraction, the abstraction of in-home living environment scenarios is conducted, where the topology was represented by a graph. In this research, apriori algorithm [3] is opted to discover frequent sensor events sets that occur with higher correlation. Assuming that people are experiencing periodic lifestyle, the periodic analysis is carried out in this research. Firstly the analysing period is split into a number of sections based on normal person’s habit; secondly, a mechanism to limit searching range was designed to construct input data. The itemsets with higher support value are selected to denote the resident’s lifestyle. Aiming to validate the effect, these methods are applied on the datasets with unobtrusive sensors that are gathered in GAL-NATARS study, which is conducted in the setting of real-life scenarios within the framework of the Lower Saxony Research Network Design of Environments for Ageing (GAL) [4].

Results: The first two unobtrusive sensors datasets, which are gathered from two subjects’ apartments for over 80 days, are analysed in the current work. The results show that the sensors typical events patterns relating to residents’ lifestyle are discovered within specific time intervals, including morning, midday, afternoon, evening, night, and late night. The pattern changing over time, i.e., the evolution of residents’ lifestyle, is represented as well. Moreover, both variance and stableness are shown in resident’s lifestyle.

Conclusions: In order to discover human lifestyle patterns, the periodic analysis is carried out. Through the methods proposed in this work, the residents’ typical lifestyle can be successfully discovered through sensor-enhanced living environments. This research also proves that the human behaviour patterns can be extracted merely using unobtrusive measurement.

Acknowledgements: This work would not have been possible without the data collected in the GAL-NATARS study. The study is conducted as part of the Lower Saxony Research Network Design of Environments for Ageing. Clinics in the multi-center study are Klinikum Oldenburg, Klinikum Braunschweig and St. Bonifatius Hospital Lingen. Study center is the PLRI at Medical School Hannover. We thank our colleagues for their work in GAL-NATARS.