< Back to previous page


Assessment of Cognitive Performance in Elderly Life via Meaningful Play

Cognitive decline is the deterioration of one or more cognitive functions such as attention, memory, or processing speed. Inherent to aging, most people will encounter some form of cognitive decline during their lifetime while retaining the ability to perform instrumental activities of daily living. However, beyond this age-related cognitive decline due to aging, some people will experience pathological cognitive decline: an abnormal rate of cognitive impairment due to neurological diseases such as dementia or due to physical trauma. In contrast to age-related cognitive decline, this pathological cognitive decline hinders activities of daily living. In between the stages of age-related and pathological cognitive decline, is the stage of Mild Cognitive Impairment (MCI), which is characterized by a slight yet noticeable decline in cognition. Nevertheless, instrumental activities of daily living remain mostly intact in this stage. While the diagnosis of MCI is not always followed by a diagnosis of dementia, people diagnosed with MCI have a markedly higher chance of progressing to dementia. As such, early case-finding of MCI and timely adjusting the diagnosis is crucial to ensure apt medical support. To aid this cause and to better understand the dwindling of cognition, this dissertation sets out to explore the possibilities of using digital card games to assess differences in cognitive performance due to cognitive aging and MCI.

In particular, the use of digital biomarkers, i.e., user-generated physiological and behavioral data collected through digital devices, is investigated. Embedded into day-to-day interactions, these digital biomarkers can be used to support diagnosis without interfering with the person's daily routine. In addition, as they are high-resolution in nature, they allow for making informed inferences of neuropsychological processes previously unavailable to psychologists.

In this doctoral dissertation, two different aspects of cognitive decline are measured using different digital card games. First, digital biomarker caused by cognitive aging are assessed using the card game FreeCell. To this end, a generic image processing toolkit was built to extract digital biomarkers from the Microsoft Solitaire Collection. Using this toolkit, data was captured from three different age categories. Machine learning models trained on this data showed promise in classifying the younger and older age categories but lacked in classifying games played by the middle-aged category.

Second, digital biomarker differences caused by MCI are assessed using the card game Klondike Solitaire. For this part, an Android application was custom-built to capture digital biomarkers while leaving gameplay untouched. Candidate digital biomarkers were identified in collaboration with 11 experts in cognitive decline. Next, gameplay data was captured from both healthy older adults and older adults diagnosed with MCI. A generalized linear mixed model analysis was conducted to investigate differences between healthy older adults and older adults living with MCI. The results of this analysis suggest it is possible to discriminate healthy participants from participants diagnosed with MCI at a group level. In addition, machine learning models were trained to discern games played by older adults with MCI. These models show promise on an individual level and are successful in discerning healthy older adults from adults living with MCI. While exploratory in nature, the results indicate similar psychometric properties as commonly used screening tests.

In sum, these findings suggest that commercial off-the-shelf card games, not built for the purpose of measuring cognition, can be used to capture digital biomarkers of cognitive performance sensitive to the cognitive decline due to aging and MCI.

Date:12 Sep 2016 →  8 Jun 2021
Keywords:Cognitive performance, eHealth, Datamining
Disciplines:Applied mathematics in specific fields, Computer architecture and networks, Distributed computing, Information sciences, Information systems, Programming languages, Scientific computing, Theoretical computer science, Visual computing, Other information and computing sciences
Project type:PhD project