This week, I am working to get the dynamic time warping functionality into my program. The process of doing so includes re-processing the features to include the time series, putting each series back together when we construct sequences, and then performing the DTW to generate a number that will be used to compute the kNN of each sequence which can then be used for predictions with the models. The processing time of these activities has gone up significantly since we have been using five different metrics with each of the F phase datasets. I am returning to school next week, and once I've completed the DTW processing all that will remain before we put together our second paper (The date for the reach journal we would like to submit it to is October 1), I am hoping I will have time to look again into the Agglomerative Hierarchical Clustering concept, which I did not successfully complete when we explored it earlier in the summer and then changed focus to the paper. We heard back
Now that we have found a solid number of metrics by which we may classify sleep, we want to begin looking for ways to more effectively classify daytime activity. After reviewing more literature on the matter, it appears that a common way of describing daytime activity is to measure the amount of time that the subject is sedentary. The word 'sedentary' has a strict definition based on metabolic energy expenditure and body position, which is beyond what we will be able to easily determine from our activity counts. However, we want to attempt to find a threshold at which we can determine if a patient is moving or inactive. Since Dr. Skornyakov has been our resource for writing rules about when a patient is asleep or awake, we are going to request her assistance on how to determine what this threshold may be.
I found earlier this year that our subjects' longest sleep bouts were very low - below 100 minutes of sleep per night, for many patients. This is not entirely surprising, considering that we expect some degree of disruption in the subjects' sleep patterns due to their injuries. I was concerned, however, that I may have been recognizing the end of a bout when the patient had merely moved in their sleep. I want to find some number of wake minutes that, when surrounded by two sleep bouts, can be considered sleep in order to merge the two bouts. I am hoping that this will give more value to the longest bout calculations. We are also reaching out to Dr. Skornyakov on this matter to see if she thinks this would be a good idea.
Following my last meeting with Dr. Sprint and Alexa, I realized I had made an error in my graphs of previous night's lights off sleep minutes vs following day's lights on wake minutes. Instead of creating scatter plots demonstrating them in relation to one another, it would be more helpful for us to look at the metrics in a way that considered the sequential nature of the relationship. This week, I am going to work on bar charts to compare the sum of lights-on activity counts to the sleep minutes of the night prior and the sleep minutes of the following night.
Comments
Post a Comment