The three data mining methods analyzed in this study yielded similar results, with standard errors (Syx%) of 24.85% for the first procedure (one nearest neighbor), 24.10% for the 2nd procedure (inverse Euclidean distance weighted with 3 neighbors) and 24.32% for the 3rd procedure (inverse of the squared Euclidean distance weighted with 3 neighbors).
With regard to the automated Bluetooth feature, the usage rate was almost the same before the ASHA's 10th procedure (50% usage rate) and after her 10th procedure (44% usage rate).
During the 15th procedure the patient briefly felt unwell and complained of having a metallic taste in his mouth.
Before the 10th procedure, Bluetooth usage rate was 40% for npB1, 50% for npM1, and 10% for npL1 while after the 10th procedure (midway between their 'assigned' target of 20 procedures as previously described in Section 'Field testing'), the usage rates were 65%, 44%, and 80% respectively.
However, with regard to Bluetooth usage, we observe that even though only a single attempt (or 10%) was made before her 10th procedure, her usage rate increased to 80% thereafter.
For example, if we are to estimate the mean completion time and confidence interval (CI) for an ASHA who has performed 10 risk assessments, her procedure times until and on the 10th procedure would be taken as our sample for bootstrapping.
But assuming these aren't your 29th and 30th procedures, I have no quibble, especially if you are reasonable in your expectations.
Arizona's two-drug procedure is different from Oklahoma's three-drug regimen but similar to the drug combination Ohio used this year in another lengthy execution.
This Report is based on data downloaded on 23rd January 2015 from the master database that gave a shunt procedure dataset from 1st January 1995 to 31st December 2014 of 53,767 procedures in 29,341 patients.
