Skip to main content
. 2021 Aug 30;76(4):527–534. doi: 10.1038/s41430-021-00999-y

Table 2.

Fivefold cross-validation for predictive equations developed from same patients in the acute phase (n = 168) and late phase (n = 168).

A
Acute phase (<5 days)
Test set Developed predictive equations R2 RMSE SEE
1 REE = 1686.9Ht + 30.7VeMax + 8.0Wt - 1975.8 0.401 292.5 344.6
2 REE = 1627.8Ht + 25.3VeMax + 8.4Wt – 1830.3a 0.414 294.0 343.7
3 REE = 2048.8Ht + 115.7TMax + 7.1Wt – 6462.8 0.442 408.6 318.8
4 REE = 1542.8Ht + 23.4VeMax + 7.5Wt – 1629.1 0.395 381.9 322.4
5 REE = 1740.4Ht + 50.2PEEP + 34.6Ve – 1876.1 0.391 426.3 317.1
B
Late phase (≥ 6 Days)
Test set Developed predictive equations R2 RMSE SEE
1 REE = 919.1 + 12.0Wt + 43.1VeMax -7.9Age 0.491 412.4 328.5
2 REE = 1008.3 + 14.4Wt + 26.1VeMax– 7.9Agea 0.498 423.6 335.9
3 REE = 1143.3Ht + 11.2Wt – 8.9Age – 318.4 0.432 380.5 346.0
4 REE = 1244.6Ht + 10.6Wt – 6.9Age – 556.3 0.398 301.7 364.5
5 REE = 1445.3Ht + 9.3Wt + 27.6VeMax -1522.7 0.363 395.3 344.4

Acute phase, ≤5 days; HR heart rate (beats/min); Ht, Height (m); late phase, ≥6 days, PEEP positive end-expiratory pressure; R2, coefficient of determination, REE resting energy expenditure, RMSE root mean square error, SEE standard error of the estimate, TMax maximum body temperature in the previous 24 h (°C), VeMax maximum minute ventilation in the previous 24 h (L/min), Wt weight (kg).

Multiple linear regression analysis was applied.

A stepwise method of variable selection applied. Model assumptions fulfilled.

No interactions between independent variables. No multicollinearity detected.

aThe selected equation.