Frame Work 1.

Classification Angle (ClassA) Framework.

Create a scatter plot of first differences within HRV by plotting [(4x(n+2)−3x(n+1)−x(n+3))/2] against [(4x(n+1)−3x(n)−x(n+2))/2] for a signal, x. Compute, in the anti-clockwise direction, the angle, αn, that each point in the scatter plot makes with the abscissa, such that data-points in the first quadrant of the scatter plot will make acute angles with the abscissa, those in the second quadrant will make obtuse angles, and those in the third and fourth quadrants will make reflex angles. Sum the angles, αn, computed in Step 2, and divide the total by the number of data-points, N, in the HRV signal. Designate this total as the Real Angle Sum (RAS), $RAS=\frac{\sum {\alpha }_n}{N}$. Count the number of points in the scatter plot which fall within the first quadrant, and denote this by $Q^{{}_1}N$. Divide $Q^{{}_1}N$ by N to find the proportion of the total number of points that lie in the first quadrant, $P^{{}_{Q1}}$, that is $P^{{}_{Q1}}=\frac{Q^{{}_1}N}{N}$. Repeat Steps 4 and 5 to compute the proportion of points in the second and fourth quadrants together, $P^{{}_{Q2,4}}$. Course-grain the HRV signal to access the signal at a higher temporal scale, and repeat Steps 4 and 5 to compute the proportion of points in the third quadrant, $P^{{}_{Q3}}$. The metrics RAS, $P^{{}_{Q1}}$, $P^{{}_{Q2,4}}$ and $P^{{}_{Q3}}$ are the four outputs of ClassA. Create a three-dimensional plot of $P^{{}_{Q2,4}}$ against $P^{{}_{Q3}}$ against $P^{{}_{Q1}}$ to identify different states of stress. The overall trend in the data is indicated by RAS.