Energy Expenditure as a Digital Biomarker

Total energy expenditure is the energy a person uses over a period, made up of resting metabolic rate, the energy cost of digesting food, and the energy cost of physical activity. Wearables estimate the activity component from movement and heart rate, add a resting estimate derived from age, sex, height and weight, and report a daily calorie total.

The presentation of that number invites more confidence than it deserves. A wearable reports a specific figure, often to the calorie, produced by a proprietary model whose inputs are an accelerometer signal and a set of demographic assumptions. Systematic review and meta analysis of activity monitors against reference methods has found error that is substantial and inconsistent between devices, which is a different situation from the step count on the same screen.

The reference standard is doubly labelled water, in which a participant drinks water containing stable isotopes and their elimination rates over one to two weeks reveal carbon dioxide production and therefore energy expenditure. It is accurate, expensive and gives one number for the whole period rather than a time course.

Wearables take a different route. Movement is converted into an activity intensity estimate, often expressed in metabolic equivalents drawn from published compendium values, and heart rate is used to refine the estimate during exercise. Both inputs are indirect. The compendium assigns an average energy cost to an activity type, which is a population value applied to an individual, and heart rate reflects many things besides metabolic demand.

The defensible use is relative. Comparing a person to themselves before and after an intervention, on the same device, gives a usable index of whether activity related energy expenditure changed. Comparing the absolute figure between people, between devices, or against a dietary intake estimate does not.

In research it is used in obesity and weight management studies, in exercise physiology, in cardiac and pulmonary rehabilitation, and as a way of connecting activity measures to metabolic outcomes within a single recording period. It is reported alongside activity questionnaires such as the SQUASH, the Physical Activity Scale for the Elderly or the Duke Activity Status Index, which describe the same behaviour through self report and provide a check on whether the device figure is plausible.

Absolute accuracy is the core problem. Meta analytic evidence shows activity monitors estimate energy expenditure with error large enough to matter for any application that depends on the number itself, such as balancing intake against expenditure, and error differs between devices and between activity types. Non ambulatory activity including cycling, resistance training and carrying loads is systematically underestimated by movement based estimation.

The resting component, which is the largest part of daily expenditure, is not measured at all but predicted from demographic equations that carry their own error. Presenting the result as a precise calorie total gives an impression of accuracy that the method cannot support, and studies should treat it as an index rather than a measurement.

• Ainsworth BE, et al. 2011 Compendium of Physical Activities: a second update of codes and MET values. Med Sci Sports Exerc. 2011. pubmed.ncbi.nlm.nih.gov
• O'Driscoll R, et al. How well do activity monitors estimate energy expenditure? A systematic review and meta-analysis of the validity of current technologies. Br J Sports Med. 2020. pubmed.ncbi.nlm.nih.gov
• Speakman JR. The history and theory of the doubly labeled water technique. Am J Clin Nutr. 1998. pubmed.ncbi.nlm.nih.gov
• Westerterp KR. Doubly labelled water assessment of energy expenditure: principle, practice, and promise. Eur J Appl Physiol. 2017. pubmed.ncbi.nlm.nih.gov