FoodEstNet: Estimating True Food Consumption with Machine Learning

Akogo, D. A.; Danquah, J. B.

doi:10.1101/250506

Home/Archive/bioRxiv

bioRxiv · 10.1101/250506

FoodEstNet: Estimating True Food Consumption with Machine Learning

Akogo, D. A.·Danquah, J. B.

View original source ↗Export metadata ↓

Abstract

We developed a Machine Learning/Artificial Intelligence model that estimates how much of a food type a person truly consumes. People tend to underestimate how much they consume which makes the work of nutritionists and dietitians difficult since they rely on food estimates for food portion size control and nutritional management of diseases. We trained an XGBoost model to estimate how much a patient truly consumes based on Age, Sex, BMI, socioeconomic status and perceived consumption.

Explore related subjects

physiology Preprint

Keep this discovery

Explore connections, maps & timelines

BibTeX RIS

Akogo, D. A., Danquah, J. B.. 2018-01-19. FoodEstNet: Estimating True Food Consumption with Machine Learning. https://doi.org/10.1101/250506

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related preprints

bioRxiv2026

Skeletal muscle aging drives sex- and compartment-specific protein insolubility that is modifiable by proteostasis-directed treatments

Protein aggregation and altered protein solubility indicate proteostatic decline, yet their roles in skeletal muscle aging remain poorly understood. We combined light and electron microscopy, quantitative proteomics, and stable-isotope labeling of skeletal muscle to investigate age-associated changes in protein insolubility, and their responses to exercise and rapamycin. Aged muscle accumulated distinct aggresome and amyloid-positive aggregates despite unchanged bulk insoluble protein content. Instead, aging altered the abundance and soluble-insoluble shifts of lower-abundance proteins across cellular components, particularly mitochondrial, nuclear and extracellular matrix proteins, with greater changes in females. These changes were not broadly explained by intrinsic protein aggregation propensity. Exercise and rapamycin partially reversed age-associated changes in abundance and partitioning without increasing protein synthesis. Together, these findings highlight sex- and cellular component-dependent changes in protein insolubility as important, modifiable aspects of skeletal muscle proteostasis, and indicate that proteostatic maintenance may be a more important target than anabolism for muscle aging.

physiology↗

bioRxiv2026

Trunk muscle activity across various trunk-limb coordination patterns in centipede locomotion

Animals coordinate movements of their limbs and flexible bodies, yet how trunk muscle activity changes across locomotor patterns remains unclear. We simultaneously recorded electromyograms from trunk muscles (deep dorsal oblique muscles) and trunk-limb kinematics in two centipede species exhibiting retrograde and direct locomotor waves during swimming and walking. During swimming, trunk muscle activity followed a similar phase relationship with trunk bending in both species despite opposite wave directions and different involvement of limbs, suggesting common active generation of undulation. During fast walking, trunk muscle activity was strongest around the ipsilateral swing phase in the undulatory retrograde-wave gait of Scolopendra mutilans, but weaker and associated with the ipsilateral stance phase despite the nearly straight trunk in the direct-wave gait of Scolopocryptops rubiginosus. Together, our findings suggest that the same trunk muscles serve two contrasting mechanical roles depending on locomotor mode: either generating trunk undulation or actively straightening the trunk against limb-induced deformation.

physiology↗

bioRxiv2026

Quantifying and Visualizing the Invisible Hazard of Infant Rebreathing

BACKGROUND: Rebreathing of expired gases during prone infant sleeping contributes to hypoxic and hypercapnic exposure, even without airway obstruction. We investigated how anatomical, physiological, and environmental factors influence aerodynamic rebreathing. METHODS: A dynamically scaled in vitro model simulated prone nasal breathing corresponding to infants weighing 2 to 5 kg, with nostril diameters of 3.2 to 6.0 mm and breathing frequencies of 25 to 40 breaths/min. Ambient temperature and sleep-surface characteristics were varied. Rebreathing was quantified using tracer measurements, while flow visualization revealed the underlying aerodynamic mechanisms. RESULTS: Rebreathing exceeded 60% despite unobstructed nasal openings. Increasing ambient temperature from 23 to 30 degrees C increased rebreathing by 68 to 77%, while reducing simulated infant mass from 5 to 2 kg increased rebreathing by 39 to 46%. Fibrous surfaces increased rebreathing, particularly near inclined nasal openings. A novel finding was that increasing nostril diameter from 3.2 to 6.0 mm increased rebreathing by up to 50%. Flow visualization revealed vortical structures that retained expired fluid near the nasal openings, facilitating re-inspiration. CONCLUSIONS: Aerodynamic rebreathing depends strongly on simulated infant characteristics and sleeping environments. These findings are consistent with established safe-sleep recommendations and identify nostril diameter, for the first time, as a potentially important anatomical factor warranting clinical investigation.

physiology↗