Figure 9. First Pre-Processing Then, we only keep the information necessary for the analysis (stamp, areaid, events). On the other hand, to make correct and fast calculations, data filtering becomes highly beneficial when working with Unix time to represent time intervals. Employing Unix time as the timestamp format streamlines the task of comparing and filtering events, facilitating smooth operations based on their temporal attributes as shown in Figure 10. 101

IADIS International Journal on Computer Science and Information Systems Figure 10. Filtering of all events by areaid Given the inter-individual variability in the physical activity intensity and duration, total energy expenditure can be estimated by multiplying resting energy expenditure by a factor reflecting the intensity of an individuals physical activity. This physical activity level factor (PAL) has been determined for many activities of daily living, including sedentary, occupational and sports activities (Westerterp, 2013). = The Basel Metabolism Rate (BMR) is obtained by the following basal metabolic rate formula: Men: BMR=88.362+ (13,397weight in kg) + (4,799height in cm)(5,677age in years) Women: BMR=447.593+ (9,247weight in kg) + (3,098height in cm)(4,330age in years)

Total energy expenditure=PALBMR Establishing predefined rules for food-related activities is an essential step in understanding and analyzing the nutritional behaviors of individuals. These rules make it possible to categorize and identify specific activities and behaviors associated with food consumption and preparation. A sequence of events or conditions that corresponds to a food-related activity is established: It begins when a person enters the kitchen or opens the refrigerator or cupboard. If the person opens the fridge or cupboard after entering the kitchen, we consider this as a part of the eating activity. The activity ends when the individual is no longer in the kitchen or dining room. Once these rules have been applied to our dataset, we can calculate correlations and analyze how different rooms and mobility patterns are related to eating-related activities (Figure 11). Steps to create our correlation matrix:

Create a New Variable: Start by adding a new category in the dataset to represent activities related to nutrition. This variable will be assigned values based on specific criteria, using a systematic approach. Organize the Data: Once the nutrition-related variable is established, we organize the data by selecting the essential columns needed for calculating presence, occupancy rates, and mobility. These columns typically include "areaid," "event," "timestep," and the newly created variable for nutrition-related activities 102 (1) (2) EMPOWERING HEALTH IN AGING: INNOVATION IN UNDERNUTRITION DETECTION AND PREVENTION THROUGH COMPREHENSIVE MONITORING Calculate Presence and Occupancy: we Determine how often different rooms (such as the kitchen, dining pantry, fridge, and dining room) are used for nutrition-related activities. This is done by grouping the data based on the room's identification ("areaid") and finding the average frequency of events in each room.