Applying Serious Games and Machine Learning for Cognitive Training and Screening: the COGNIPLAT Approach

The application of serious games as a means of cognitive rehabilitation has become particularly popular in the field of medicine in recent years. Contemporary literature appears to be encouraging on applying cognitive intervention programs to elderly people with the utility of innovative gaming platforms, as research has shown that serious games can strengthen cognitive functions. In this work the COGNIPLAT game platform is presented that consists of serious games that were designed to enhance cognitive functions through different training exercises. A human-centered design approach for the design of the game screens was followed targeting a high usability and acceptability. In addition, machine learning algorithms were applied to train models based on performance data collected from the gaming platform. These models serve as a cognitive screening tool to discriminate healthy from cognitive impaired individuals. An evaluation pilot study took place in order to assess the COGNIPLAT platform in terms of technology acceptance and cognitive efficacy. In the latter case both the cognitive training results and the accuracy of cognitive screening are reported. The sample of the study consisted of 10 seniors living in the community, aged 65-90 (mean 76.1 +/- 7.9). The intervention consisted of 12 30-minute sessions, which took place over a period of 12 weeks. Measurements of cognitive functions (abstraction, attention, concentration, delayed recall, executive functions, language, memory, orientation and visuospatial) were carried out before and immediately after the intervention for comparison.