The extent to which variation in food-related metabolites are attributable to non-dietary factors remains unclear, which may explain inconsistent food-metabolite associations observed in population studies. This study examined the association between non-dietary factors and the serum concentrations of food-related biomarkers and quantified the amount of variability in metabolite concentrations explained by non-dietary factors. Pregnant women (n = 600) from two Canadian birth cohorts completed a validated semi-quantitative food frequency questionnaire, and serum metabolites were measured by multisegment injection-capillary electrophoresis-mass spectrometry. Hierarchical linear modelling and principal component partial R-square (PC-PR2) were used for data analysis. For proline betaine and DHA (mainly exogenous), citrus foods and fish/fish oil intake, respectively, explained the highest proportion of variability relative to non-dietary factors. The unique contribution of dietary factors was similar (15:0, 17:0, hippuric acid, TMAO) or lower (14:0, tryptophan betaine, 3-methylhistidine, carnitine) compared to non-dietary factors (i.e., ethnicity, maternal age, gestational age, pre-pregnancy BMI, physical activity, and smoking) for metabolites that can either be produced endogenously, biotransformed by gut microbiota, and/or derived from multiple food sources. The results emphasize the importance of adjusting for non-dietary factors in future analyses to improve the accuracy and precision of the measures of food intake and their associations with health and disease.

Purpose The McMaster Monitoring My Mobility (MacM3) study aims to understand trajectories of mobility decline in later life using multisensor wearable technology. To our knowledge, MacM3 is the first major cohort to combine accelerometry and a Global Positioning System (GPS) to track real-world mobility in community-dwelling older adults. Participants Between May 2022 and May 2024, MacM3 recruited 1555 community-dwelling older adults (mean age 73.9 years, SD=5.5) from Hamilton and Toronto, Ontario. Of the cohort, 68.4% were female, 62.4% married/partnered, 75.3% had post-secondary education and 62.9% had≥3 comorbidities. Most were Canadian born (69.4%) and white/Caucasian (88.0%), with greater ethnocultural diversity observed at the Toronto site. Findings to date At baseline, 56.7% of participants reported no mobility limitations, 15.9% had preclinical limitations and 27.4% had minor mobility limitations. Mean gait speed for the total sample was 1.23 m/s, with a mean Timed Up and Go time of 9.4 s and a 5× sit-to-stand time of 13.0 s. A total of 1301 participants had valid wrist-worn device data, and 1008 participants who agreed to wear the thigh-worn device had valid data (≥7 days with ≥10 hours of wear per day). Step count data (n=1008) revealed a mean of 8437 steps per day (SD=2943), with 5073 steps in the lowest quartile and 12 303 steps in the highest. Future plans Ongoing work aims to develop predictive models of mobility decline by integrating wearable, clinical and environmental data. Pipeline enhancements will enable GPS/inertial measurement unit fusion to explore mobility-environment interactions and support ageing-in-place tools.