This study is a 2026 study which used machine learning to comb through variables to help answer the question- what determines your baseline calorie needs? Resting metabolic rate is an important metric to know, so you can adjust your calorie intake and maintain, lose, or gain weight. We know 70% of it is determined by fat free mass (muscle) and major organs. But what leads to the 30% variation?
Study:
This study developed and validated machine learning models to predict resting metabolic rate (RMR) in 454 healthy adults from the Enable cohort.
- The groups were placed into 3 age groups: young adults (18-25), adults (40-65), and older adults (75-85)
- They compared 94 potential predictors to identify which factors most accurately estimate RMR. This included body temperature, outdoor temperature, microbiome of the gut, short chain fatty acids, lab tests (T3, GFR for kidney function, MCHC mean corpuscular hemoglobin concentration),
- They were all healthy, nonsmoking, Caucasian. They were excluded if they were pregnant, had chronic infections, diabetes, history of heart attack, or untreated blood pressure, cancer within 3 years, or severe disease of a major organ)
- Testing was done by indirect calorimetry, done in the morning between 7:30-9, after an overnight fast. Temperature and humidity were controlled. The test was done for 45 minutes, where they discarded the first 10 minutes to improve accuracy. Note: This is what we do at Biohackr Health, though we have you acclimate for around 5 minutes and the test is usually 10-15 minutes. We agree with early morning, same time when you do repeated tests, and fasting. See our RMR prep INFO
What did this study find?
The Lasso regression model performed best. The Lasso model is valuable in medical research, as it excels in “big p, small n” situations where there are many potential predictors (hundreds or thousands of variables) relative to the number of patients or observations. This test helped increase RMR accuracy, explaining 76.8% of RMR variance, when trained on the full Enable cohort feature set. The most important predictive features were fat-free mass, body weight, and mean outdoor temperature.
What variables had importance for RMR?
The study systematically evaluated which types of variables contribute to RMR prediction.
- Fat-free mass and body weight emerged as the dominant predictors, consistent with established physiology.
- Mean outdoor temperature was identified as a novel contributor to RMR prediction.
- Blood-based clinical parameters provided marginal improvements to model- GFR and MCHC performance.
- Gut microbiome composition and fecal short-chain fatty acids did not contribute to explaining RMR variance, suggesting these factors are not useful for RMR prediction in healthy adults.
What do we at Biohackr Health think?
We love doing resting metabolic rate. Interesting that this study adds outdoor temperature to the mix. It makes sense- if it is cold out, your body needs to “burn more calories” to keep warm.
KNOW YOUR MUSCLE PERCENTAGE.
Do an InBody scan so you can see real numbers. Remember men and women differ in what is considered a healthy amount. Middle aged men ideal rate is 15-18%, middle aged women 21-33%.
BUILD MUSCLE
We think everyone should be on creatine, particularly if vegan/vegetarian, particularly as you get older. Lean muscle mass is the biggest predictor for resting metabolic rate. It also helps prevent diabetes, lowers cardiovascular disease risk, protects your joints and bones, and helps prevent dementia. Win. Win. Creatine SHOP Creatine blogs
LOSE WEIGHT
Fat mass is not good for burning calories. It is weight without a positive benefit. If your BMI is out of the healthy range, look at our medical weight loss options. Medical Weight Loss
TEST YOUR RMR
Don’t know what your resting metabolic rate is? Test it! The test is super easy, and it is one of the most variable metrics when you are actively losing weight or changing your exercise routine. Test RMR
Medical Citation
Predicting resting metabolic rate in healthy adults: a comparative analysis using the enable cohort
American Journal of Physiology-Endocrinology and Metabolism