Data Analysis

The primary method for statistical analysis includes statistical substantiation, blind testing, and bias elimination. Although the bias and subjectiveness of the images have been constrained (within the researchers’ capacities), there is inherent bias from the perspective of the researcher. Several 3rd party sources will be referenced to decide particular Amyloid and Tau images to be analyzed. 3rd party data decision will provide variety and strengthen the neural network, as well as eliminating flawed bias. The sample of accuracy and precision is primarily sampled through the threshold score. The threshold score of ~0.15 will be used to determine broad positive and negative classification accuracy (<0.15 indicating negative and 0.15< indicating positive). Accuracy and precision will be quantified on bar graph analysis. Moreover, training data will be further indicated on a line graph, highlighting iterations and the learning pace.

Broad Classifier Diagnosis: Positive and Negative Recognition

Broad classification is comprised of simple input and output recognition. Negative and positive classification distinguishes and diagnoses positive and negative cases of progressive Alzheimer’s disease (in both early and latent stages). Initial data responses were recorded on API confidence score ratings. However, confidence scores do not provide a viable benchmark for accuracy recordings. Following nearly 50 positive and 50 negative test trials, standard averages were calculated in both classification groups to determine a reliable threshold-accuracy rating.