...
Given the positional differences between the invasive sensors (inside the ventricle) and the non-invasive sensors (outside the skull), strong agreement between the signals was not anticipated. However, the study observed a relatively large region of agreement and a significant correlation between the parameters, as confirmed by additional statistical tests.
Correlation analysis:
Spearman correlation and normalized mutual information were used to assess the statistical dependence of ICP waveform parameters between the Braincare sensor and the invasive sensor.
...
The joint distributions between ICP and B4C parameters demonstrated statistical dependence, confirming a statistically significant correlation between the ICP pulse morphology (waveform) detected by the gold standard invasive methods and the B4C technology, specifically in terms of the P2/P1 ratio and Time to Peak.
Agreement analysis:
Bland-Altman plots and Deming regression models were used to quantify the agreement between the invasive and Braincare-measured ICP waveform parameters—specifically the estimated P2/P1 ratio and normalized time to peak. These analyses estimated the differences between the respective minute-by-minute averages. Given the differing sensor placements (inside the ventricle for invasive sensors and outside the skull for non-invasive sensors), a strong agreement between the signals was not anticipated. However, a relatively large region of agreement between the parameters was observed, demonstrating statistical significance as confirmed by additional tests.
...