LAB 3 WRITE-UP
Descriptive Stats and Graph
- Gold Standard Average = 96.6
Spree Average = 95.5
- Gold Standard Standard Deviation = 1.92
Spree Standard Deviation = 0.869
- Gold Standard Average = 98.1
Spree Average = 98.95
- Gold Standard Standard Deviation = 22.99
Spree Standard Deviation = 24.83
Heart Rate p-value = 0.427
When testing heart rate, the results of the spree tests were very significantly related to that of the gold standard, as we can see from both the graph (the bars and error bars are very similar) and the fact that the p value exceeds .05.
Temperature p-value = 1.906 E-21
When testing temperature, the results from the gold standard measurements covered a much broader area than the spree test, showing greater accuracy. The p-value reasserts this, as the correlation coefficient is incredibly small.
(P-values were determined through a two-tailed, paired t-test)
Design Flaws and Recommendations
The Spree device does a good job collecting the necessary data on heart rate but not so much on temperature values. The results of the gold standard and Spree were incredibly similar. Both p values on heart rate exceed .05. With the temperature results, the gold standard measurements covered more than the Spree. There is more accuracy in the gold standard results than the Spree. Overall the Spree results showed really good accuracy compared to the gold standard. The design of the Spree was anesthetically unappealing, but served it's purpose in collecting the data on heart rate and temperature.
Experimental Design of Own Device
Current methods for measuring blood glucose levels are painful and confusing.
Other forms of body fluid can be used to obtain an equally accurate blood glucose level reading while being less invasive. Our product will use saliva to make this measurement and will have a chart with suggestions to make results seem less confusing for new diabetes patients.
We will set up ten small labs across the country in different regions with different lifestyle environments for the inhabitants there. Then we will put out an advertisement for any individual, whether diabetic or not, to spend five paid hours in our lab as we monitor their blood glucose levels each hour using both our device and the current measurement device on each patient. Patients will be instructed to fast before coming into our lab, starting at midnight the night before (diabetic patients should still keep an eye on blood glucose levels and responsibly treat those while still within the parameters of a fast, ei. Drink juice or something to level out blood glucose and insulin, instead of eating something to fix the problem). We will take an initial blood glucose reading with both devices. After that, food will be provided for test subjects, after which another blood glucose reading with both devices will be taken. For the rest of the time remaining, every hour a blood glucose reading will be taken with both devices.
The fast is necessary for our research, because we want to measure different levels of blood glucose. After eating, blood glucose levels spike and then level out after that, so if we have all of our subjects come in with having not eaten since midnight the night before, then have them eat, we will be able to see how well our device measures the spike in blood glucose levels after eating, as well as being able to expose our device to a wider range of blood glucose values to make sure it is consistently accurate throughout the range. However, we do not wish to endanger the lives of any diabetic patients who wish to participate in our study, therefore it is important that they still treat their insulin and blood glucose levels accordingly while still trying to keep in line with the fast.
The time frame we have selected is to be able to monitor how blood glucose levels change over time. The average time it takes for blood glucose levels to return to normal after a spike is about five hours, therefore it is a fitting time span for our experiment.