Live Z-Score Linearity

The linearity for both BrainDX Live Z DLL and also the ANI Live Z DLL are shown below.  What is evident is that there are differences in the intercept and the slope, which must be accounted for when viewing and interpreting the live Z-scores.  Since these results emerge directly from the corresponding DLL's, BrainMaster does not alter them.  However, this demonstrates the integrity and accuracy of the implementation, as is achieves what the providers of the DLL's would expect.  These tests were conducted on 7/12/2013 on the then-current hardware and software.  At the bottom of the page, we show the exact correlation between the two databases.


-          Krohn-Hite 5920 S/N 132

-          BrainMaster Mighty-Mho model 371002 S/N 0137

-          BrainMaster Discovery 24E S/N 60054

-          BrainAvatar software

BrainAvatar was configured to control the HK5920 and the frequency was set to the center of the band being measured. For example, for ANI alpha this was 10.0Hz and for BrainDx alpha it was 10.2Hz. Age was set to 10 and ZScores set for eyes closed. The HK5920 was then stepped from 0.15V to 5V in 0.05V steps, held for 20 seconds for each step and the ZScores recorded. After each voltage sweep the next band was selected and the voltage sweep repeated.  After each band sweep the age was increase by 10 and the voltage/band sweep repeated.


All results are log linear with a fit of at least 0.999

ANI results for P3 P4 T3 T4 theta:



BrainDx results for P3 P4 T3 T4 theta:



The following graph shows these values plotted against each other, producing the correlation coefficient.  For this graph, we show all 19 channels and all 4 components (delta, theta, alpha, beta).  The correlation coefficients are 0.95, 0.99, 0.99, and 0.98.  Note that not all components have the same intercepts for zero.  That is due to the databases themselves, and cannot be changed by BrainMaster.  Nonetheless, the correlations are very strong.


Attached Files


      Related Entries