ITOKI and Matsuo Institute used AI models to estimate the reality of office break areas by analyzing behavioral data from 854 employees. The findings suggested that the number of meetings and the volume of incoming email affect mental and physical condition and interpersonal relationships, potentially influencing productivity.
ITOKI, a major office design company, and Matsuo Institute built an AI model using anonymized data from 854 employees at the ITOKI Nihombashi office to estimate how everyday behavior relates to mental and physical condition, interpersonal relationships and, ultimately, productivity. The analysis was based on data from 1 location over 1 month.
The model has three layers. The first extracts 70 features, including location data, calendar information, and email and chat interactions. The second estimates mental, physical and interpersonal condition, while the third calculates the degree to which individuals demonstrate their capabilities based on self-assessments. Causal direction was estimated using VAR-LiNGAM, a time-series method.
The number of meetings was found to have a greater effect on mental condition than the length of meetings. Mental condition tended to be lower the day after a day with many meetings, while it tended to be higher the day after a day on which time had been secured for work. The findings suggested that the source of busyness was related to an increase in the number of messages received.
Changes in interpersonal relationships within the organization were observed on days when many messages were received, and there were also cases in which improvement was seen the day after a day with a high volume of outgoing messages. Because the analysis was based on 1 month of data, the report said that separate models would be needed to apply the findings to other locations or companies. Validation for commercialization is also reportedly under way.