Week 5: More Research

With half of the summer immersion already completed, I decided to dedicate my time in research to make some significant progress in my work. Dr. Kuceyeski was able to provide me with the data of subjects who are healthy and who had traumatic brain injury. The amount of information to process is astounding. The functional and structural connectivity of each patient have been obtained using magnetic resonance imaging. These information is used to predict 46 measures of learning, attention and memory. In order to handle the sheer amount of information, I began developing an algorithm to sort and eliminate subjects who do not have all the data collected. On the other hand, an a special parameter must be determined in order to implement the general regression neural network (GRNN). This constant is known as the smoothing parameter. The most straightforward methods to accomplish this task is known as the holdout method. In addition to many disadvantages with this technique, I do not know if it would work with my dataset until I finish implementing it. Finally, due to the high dimensionality of my dataset and the number of subjects, the time it takes for the algorithm to run could be significantly long. As a result, in the following days, I might implement a principal component analysis (PCA) in order to use the important features of my data.