The model
wdm models one person’s body weight as a six-state, continuous-time system driven by a daily food log, observed through a scale and an optional tape measure. Units: kilograms, kilocalories, days; symbols are in Notation.
The six states
| Symbol | Code name | Units | Meaning |
|---|---|---|---|
| \(T\) | tissue_kg |
kg | Tissue: fat plus fat-free solids and their structural water. What a diet changes. |
| \(F\) | fat_kg |
kg | Fat mass, part of \(T\); lean mass is \(M = T - F\). |
| \(G_c\) | gly_water_kg |
kg | Glycogen and its bound water, relative to habitual intake. |
| \(G_n\) | na_water_kg |
kg | Sodium-associated water, relative to habitual intake. |
| \(A\) | adapt_kcal |
kcal/day | Adaptive thermogenesis, negative in a deficit. |
| \(G_{\mathrm{gut}}\) | gut_kg |
kg | Gut content: food and stool in transit (absolute, not a deviation). |
The scale weighs \(W = T + G_c + G_n + G_{\mathrm{gut}}\).
Energy balance
Stored energy changes by metabolizable intake \(I\) minus expenditure \(E\). Self-reported intake is biased (Dhurandhar et al., 2015), so it enters as a noisy input (log bias). Energy is rebuilt from Atwater factors (4, 4, 9, 7 kcal/g) when only macronutrients are logged, and 2 kcal per gram of fiber is subtracted, since labels count fiber as carbohydrate.
Fat and lean: where the mass comes from
Adipose tissue stores \(\rho_F = 9441\) kcal/kg and fat-free tissue \(\rho_L = 1816\) kcal/kg (Hall, 2008).
The same deficit moves the scale five times as far if the mass comes off as lean tissue rather than fat. A fixed 3500 kcal per pound ignores this and falling expenditure, and so predicts linear loss where real trajectories flatten (Hall et al., 2011).
Forbes’ law. Lean mass is approximately logarithmic in fat mass, \(M = 10.4 \ln F + \text{const}\) (Forbes, 1987), so the fat share of a mass change and the energy cost of a kilogram of tissue are
\[ p_{\mathrm{Forbes}} = \frac{F}{F + 10.4}, \qquad \rho_{\mathrm{eff}} = p\,\rho_F + (1 - p)\,\rho_L . \]
| \(F\) (kg) | 50 | 30 | 20 | 15 |
|---|---|---|---|---|
| \(p_{\mathrm{Forbes}}\) | 0.83 | 0.74 | 0.66 | 0.59 |
| \(\rho_{\mathrm{eff}}\) (kcal/kg) | 8128 | 7478 | 6832 | 6319 |
The leaner the body, the more lean tissue each kilogram lost contains, so the last stretch of a diet is harder than the first.
Protein. Higher protein intake during a deficit spares lean mass (Helms et al., 2014). With \(P_g\) the protein eaten (g/day),
\[ p = \operatorname{clip}\!\left( \frac{F}{F + 10.4} + \kappa_P \left( \frac{P_g}{W} - 1.6 \right),\; 0.50,\; 0.98 \right), \]
on days with a logged macronutrient split. \(\kappa_P\) is not recoverable from scale data (identifiability); it stays because omitting it would assert that protein has no effect.
Expenditure
\[ E = \underbrace{c_0 + c_1 T}_{\text{resting + baseline activity}} + \mathrm{TEF} + X + k_{St}\,T\,\Delta\mathrm{St} + A , \]
with \(X\) the logged exercise (kcal/day). The mass term uses tissue, not scale weight: glycogen water, sodium water and gut content cost nothing to maintain, and scaling expenditure with them would let a carbohydrate or salt swing change the predicted burn.
Steps. \(\Delta\mathrm{St} = \mathrm{St} - \overline{\mathrm{St}}\) is the day’s step count minus the episode’s mean, zero on a day without one, and \(k_{St} = 0.0005\) kcal per step per kg of tissue: walking costs roughly 0.5 kcal per kg per km above rest, at 1,300 to 1,400 steps a kilometer, plus the upright posture walking adds over sitting. A 4,000-step swing at 100 kg is 200 kcal/day. Measured from the mean, habitual walking stays inside \(c_0\) and \(c_1\), and steps explain only day-to-day changes in activity. \(k_{St}\) is fixed: a few weeks of step counts cannot separate it from the noise in daily intake. \(X\) covers workouts; steps cover everyday walking.
Resting expenditure. The Mifflin-St Jeor equation (Mifflin et al., 1990), times an activity level of 1.45, centers the priors on \(c_0\) and \(c_1\); both are estimated, since population equations miss an individual by hundreds of kilocalories. \(c_0\) also absorbs baseline activity and any constant log bias.
Thermic effect of food, with macronutrients in g/day:
\[ \mathrm{TEF} = \alpha_P \cdot 4 P_g + \alpha_C \cdot 4 C + \alpha_F \cdot 9 F_g + \alpha_A \cdot 7 A_g , \]
with \(\alpha_P = 0.25\), \(\alpha_C = 0.075\), \(\alpha_F = 0.02\), \(\alpha_A = 0.15\) (Westerterp, 2004), or 10% of intake when no split is logged.
Adaptive thermogenesis. In a deficit, expenditure falls by more than mass loss explains (Leibel et al., 1995), sometimes for years (Fothergill et al., 2016). Adaptation responds to the gap against expenditure without it:
\[ E_{\mathrm{base}} = c_0 + c_1 T + \mathrm{TEF} + X + k_{St}\,T\,\Delta\mathrm{St}, \qquad D = E_{\mathrm{base}} - I, \qquad \dot A = \frac{-\beta_A D - A}{\tau_A}, \]
with \(\beta_A \approx 0.15\) (Hall, 2010) and \(\tau_A \approx 14\) days.
Under a sustained gap \(D\), adaptation settles at \(-\beta_A D\): a 500 kcal/day deficit eventually loses 75 kcal/day to it, and loss slows at constant intake. The plateau is a property of the system, not a failure of the person in it.
The intake that holds weight steady is \((c_0 + c_1 T)/(1 - \phi)\) for a thermic fraction \(\phi\), about 11% above \(c_0 + c_1 T\) at \(\phi = 0.10\).
Water and gut content
Three first-order lags move the scale within days with no change in tissue.
Glycogen water (about 3 g of water per gram of glycogen; Olsson and Saltin, 1970):
\[ \dot G_c = \frac{k_G\,\Delta C - G_c}{\tau_G}, \qquad \Delta C = C - \bar C . \]
A sustained +250 g/day swing puts about 0.8 kg on the scale. This compartment explains the dramatic first week of a low-carbohydrate diet and the “relapse” after one large meal, and keeps that water from masquerading as fat.
Sodium water tracks sodium load, as the kidneys see it:
\[ \dot G_n = \frac{k_{Na}\,\Delta\mathrm{Na}_{\mathrm{eff}} - G_n}{\tau_{Na}}, \qquad \Delta\mathrm{Na}_{\mathrm{eff}} = \Delta\mathrm{Na} + \xi_C \operatorname{clip}\!\left(\frac{\Delta C}{\bar C},\, -1,\, 1\right), \qquad \Delta\mathrm{Na} = \mathrm{Na} - \overline{\mathrm{Na}} , \]
with \(\mathrm{Na}\) in mg/day and \(\Delta\mathrm{Na} = 0\) when sodium is not logged. A salty restaurant meal raises the next morning’s weight by roughly 0.4 to 0.7 kg, gone within two or three days. Carbohydrate below habitual intake makes the kidneys excrete more sodium, and the water with it: removing all dietary carbohydrate acts like eating \(\xi_C = 4000\) mg/day less sodium, and a partial cut acts in proportion (Hall, 2010). \(\xi_C\) is that literature value, held fixed, since a diet’s early weeks are the only time it is visible and fitting it from them would confound it with the energy balance. The term is zero on days without a logged macronutrient split.
Gut content follows how much is eaten, with a lag of a few days:
\[ \dot G_{\mathrm{gut}} = \frac{k_{\mathrm{gut}}\,I - G_{\mathrm{gut}}}{\tau_{\mathrm{gut}}}, \qquad k_{\mathrm{gut}} = 0.0006\ \text{kg per kcal/day}, \quad \tau_{\mathrm{gut}} = 3\ \text{days}, \]
with \(I\) the day’s intake (expenditure on an unlogged day). A 3,000 kcal/day diet keeps about 1.8 kg in the gut and a 1,300 kcal/day diet about 0.8 kg, so starting a diet sheds about a kilogram over the first week. Without this state that drop is counted as tissue lost, which inflates the maintenance estimate. Both constants are fixed, chosen by real-data likelihood and backtest from a small grid around literature-scale values (0.0003 to 0.0008 kg per kcal/day; whole-gut transit takes one to three days); the data barely distinguish neighboring values. A fiber term was tried and did not help.
The habitual-intake reference
\(\bar C\) and \(\overline{\mathrm{Na}}\) are the subject’s habitual, pre-diet intakes: by default 260 g and 4000 mg a day for men, 200 g and 3000 mg for women (The subject file). A diet starts with both water compartments believed to be at their habitual fill, \(G_c(0) = G_n(0) = 0\) (give or take 1 kg each, since pre-diet eating can leave them above it), so the drop to diet intake drains them in the first days: 80 g/day of carbohydrate against a habitual 260 g sends \(G_c\) toward \(-0.54\) kg and, through the kidneys, \(G_n\) toward \(k_{Na} \times 4000 \times (-180/260) \approx -0.55\) kg at the prior gain. The reference is fixed rather than each episode’s mean intake, because a mean hides that drop and moves as days are added. Gut content needs no reference: it starts at \(k_{\mathrm{gut}}\) times the weight-stable intake of the expenditure prior (initial belief).
The assembled model
With state \(x = (T, F, G_c, G_n, A, G_{\mathrm{gut}})\) and daily input \(u = (I, P_g, \Delta C, F_g, A_g, \Delta\mathrm{Na}, X, \Delta\mathrm{St})\):
\[ \begin{aligned} \rho_{\mathrm{eff}}\,\dot T &= I - E - \rho_G\,\dot G_c, & \dot F &= p\,\dot T, \\ \dot G_c &= (k_G\,\Delta C - G_c)/\tau_G, & \dot G_n &= (k_{Na}\,\Delta\mathrm{Na}_{\mathrm{eff}} - G_n)/\tau_{Na}, \\ \dot A &= \bigl(-\beta_A (E_{\mathrm{base}} - I) - A\bigr)/\tau_A , & \dot G_{\mathrm{gut}} &= (k_{\mathrm{gut}}\,I - G_{\mathrm{gut}})/\tau_{\mathrm{gut}} . \end{aligned} \]
Sodium water and gut content carry no energy, so neither appears in the tissue equation.
\(\rho_G \approx 1025\) kcal/kg is the energy density of the glycogen-water complex (4.1 kcal/g of glycogen, diluted by 3 g of water).
Parameters and priors
Nine parameters are estimated, each through a transform of a standard-normal coordinate \(\eta\) (reparameterization): “log, \(s\)” means \(\theta = c\,e^{s\eta}\), “logit, \(s\)” means \(\theta = \sigma(\operatorname{logit} c + s\eta)\), and “normal, \(s\)” means \(\theta = c + s\eta\).
| Code name | Symbol | Meaning | Prior center \(c\) | Spread | Units | Source |
|---|---|---|---|---|---|---|
rmr_intercept |
\(c_0\) | Mass-independent expenditure, plus constant log bias | \(1.45\,(6.25 \times \text{height} - 5 \times \text{age} + s)\) | log, 0.15 | kcal/day | Mifflin et al. (1990) |
rmr_slope |
\(c_1\) | Mass dependence of expenditure | 14.5 | log, 0.25 | kcal/day per kg | Mifflin et al. (1990) |
at_gain |
\(\beta_A\) | Fraction of a sustained deficit offset by adaptation | 0.15 | logit, 0.70 | dimensionless | Hall (2010) |
at_tau |
\(\tau_A\) | Adaptation time constant | 14 | log, 0.50 | days | Order of magnitude |
gly_gain |
\(k_G\) | Glycogen-water gain | 0.0030 | log, 0.50 | kg per g/day | Olsson and Saltin (1970) |
gly_tau |
\(\tau_G\) | Glycogen-water time constant | 1.5 | log, 0.40 | days | Assumed fast lag |
na_gain |
\(k_{Na}\) | Sodium-water gain | \(2 \times 10^{-4}\) | log, 1.0 | kg per mg/day | Extracellular fluid at 140 mmol/L |
na_tau |
\(\tau_{Na}\) | Sodium-water time constant | 1.0 | log, 0.40 | days | Standard physiology |
partition_kappa |
\(\kappa_P\) | Fat-share shift per g/kg/day of protein above 1.6 | 0.05 | normal, 0.05 | per g/kg/day | Helms et al. (2014) |
Held fixed rather than estimated: the thermic-effect coefficients, the step cost \(k_{St}\), the carbohydrate effect on sodium \(\xi_C\), and the gut-content gain \(k_{\mathrm{gut}}\) and lag \(\tau_{\mathrm{gut}}\) (reasons above).
Unlogged days and timescales
An unlogged day assumes intake equal to expenditure, habitual carbohydrate and sodium, and inflated process noise (process noise); a day without a step count assumes the episode’s mean.
The water compartments move in 1 to 2 days, gut content in about 3, adaptation in about 14, tissue over months. Instantaneous adaptation would be confounded with an error in \(c_0\); the transient after a change in intake is the whole basis for estimating \(\beta_A\).
The observation model
The scale reads total mass plus noise on the morning of day \(k\):
\[ y_k = H x_k + v_k, \qquad H = \begin{pmatrix} 1 & 0 & 1 & 1 & 0 & 1 \end{pmatrix}, \qquad v_k \sim \mathcal{N}(0, R). \]
\(R\) is mostly fluid variation and the day-to-day scatter of gut content around its modeled level, not instrument error: an sd near 0.4 to 0.6 kg for consistent morning weights (how to weigh), estimated in phase 1 and gated against typing errors (noise model).
The waist tape
The scale sees fat only through \(\rho_{\mathrm{eff}}\); a tape measure responds to fat directly. The optional waist reading (cm) is modeled as
\[ y^w_k = a_w + b_w\,F_k + v^w_k, \qquad v^w_k \sim \mathcal{N}(0, \sigma_w^2), \qquad \sigma_w = 1.5\ \text{cm}. \]
The offset \(a_w\) and slope \(b_w\) are personal constants, estimated with the metabolic parameters. Their priors come from the Lean, Han and Deurenberg (1996) waist equation for body fat at the starting composition: about 1.2 cm per kilogram of fat for a 100 kg man with 30 kg of fat, offset sd 7.2 cm (11.2 cm for women). The tape anchors the fat level but not the fat share (what the tape identifies).
How firm each piece is
Energy balance is certain and the tissue energy densities and the existence of adaptation are strong. Forbes partitioning, the water couplings and the thermic-effect magnitudes are moderate; protein sparing and the adaptation parameters are weak; Mifflin-St Jeor, Lean and the habitual intakes only center priors. A wide posterior on a weakly founded parameter is the estimator correctly reporting that neither the literature nor the data has much to say.
References
- Dhurandhar, N. V., Schoeller, D., Brown, A. W., Heymsfield, S. B., Thomas, D., Sørensen, T. I. A., Speakman, J. R., Jeansonne, M., and Allison, D. B. (2015). Energy balance measurement: when something is not better than nothing. International Journal of Obesity.
- Forbes, G. B. (1987). Lean body mass-body fat interrelationships in humans. Nutrition Reviews.
- Fothergill, E., Guo, J., Howard, L., Kerns, J. C., Knuth, N. D., Brychta, R., Chen, K. Y., Skarulis, M. C., Walter, M., Walter, P. J., and Hall, K. D. (2016). Persistent metabolic adaptation 6 years after “The Biggest Loser” competition. Obesity.
- Hall, K. D. (2008). What is the required energy deficit per unit weight loss? International Journal of Obesity.
- Hall, K. D. (2010). Predicting metabolic adaptation, body weight change, and energy intake in humans. American Journal of Physiology: Endocrinology and Metabolism.
- Hall, K. D., Sacks, G., Chandramohan, D., Chow, C. C., Wang, Y. C., Gortmaker, S. L., and Swinburn, B. A. (2011). Quantification of the effect of energy imbalance on bodyweight. The Lancet.
- Helms, E. R., Zinn, C., Rowlands, D. S., and Brown, S. R. (2014). A systematic review of dietary protein during caloric restriction in resistance trained lean athletes: a case for higher intakes. International Journal of Sport Nutrition and Exercise Metabolism.
- Lean, M. E. J., Han, T. S., and Deurenberg, P. (1996). Predicting body composition by densitometry from simple anthropometric measurements. The American Journal of Clinical Nutrition, 63(1), 4-14.
- Leibel, R. L., Rosenbaum, M., and Hirsch, J. (1995). Changes in energy expenditure resulting from altered body weight. New England Journal of Medicine, 332, 621-628.
- Mifflin, M. D., St Jeor, S. T., Hill, L. A., Scott, B. J., Daugherty, S. A., and Koh, Y. O. (1990). A new predictive equation for resting energy expenditure in healthy individuals. American Journal of Clinical Nutrition, 51, 241-247.
- Olsson, K.-E., and Saltin, B. (1970). Variation in total body water with muscle glycogen changes in man. Acta Physiologica Scandinavica.
- Westerterp, K. R. (2004). Diet induced thermogenesis. Nutrition & Metabolism.