What is accurate enough? An arbitrary value choosen by the author?
Since I am doubtful you are getting my point, or understanding the basis of that abstract. They have fitted a model (like that equation you quoted previously) to a certain group (how they have done so, I do not know, as I haven't read the whole paper, there are many ways of fitting models to data), doing a certain activity under certain conditions.
That is what I referred to above, you can do it, if you fit a model to some data, just like those seemingly arbitrary values in the equation you linked are likely fitting parameters to fit the equation to some data collected for a group doing a certain activity. BUT the crux of the issue is that one model is not usually very good for more than one scenario, especially empirical models, change the activity and you have to change the fitting parameters or reformulate altogether in some cases. To get accurate data from your HRM, the HRM must use a model that approximates based on a given activity, otherwise it will more than likely be way off.
Where do they show that calorie consumption between various activities can be estimated accurately using the same model?