Automated Detection of Activity Transitions for Prompting

Abstract

Individuals with cognitive impairment can benefit from intervention strategies like recording important information in a memory notebook. However, training individuals to use the notebook on a regular basis requires a constant delivery of reminders. In this work, we design and evaluate machine learning-based methods for providing automated reminders using a digital memory notebook interface. Specifically, we identify transition periods between activities as times to issue prompts. We consider the problem of detecting activity transitions using supervised and unsupervised machine learning techniques, and find that both techniques show promising results for detecting transition periods. We test the techniques in a scripted setting with 15 individuals. Motion sensors data is recorded and annotated as participants perform a fixed set of activities. We also test the techniques in an unscripted setting with 8 individuals. Motion sensor data is recorded as participants go about their normal daily routine. In both the scripted and unscripted settings a true positive rate of greater than 80% can be achieved while maintaining a false positive rate of less than 15%. On average, this leads to transitions being detected within 1 minute of a true transition for the scripted data and within 2 minutes of a true transition on the unscripted data.

A survey of methods for time series change point detection

“…Therefore, any supervised learning algorithm that generates an interpretable model (such as a decision tree or a rule learner) will not only identify a change but also describe the nature of the change. Support vector machines [21][40], naïve Bayes [21], and logistic regression [21] have been tested using this approach. This type of problem will also suffer from extreme class imbalance as there are typically many more within-state sequences than change point sequences.…”

“…In a time series, using sliding window X t as a sample instead of x t , an interval χ t with Hankel matrix { X t , X t +1 , … , X t + n –1 } as shown in Figure 2 will be a set of n retrospective subsequence samples starting at time t [2][21][22]. …”

Prompting technologies: A comparison of time-based and context-aware transition-based prompting

“…For this reason, we developed an activity recognition algorithm that has the ability to detect transitions and deliver prompts automatically during activity transitions, thus eliminating the need for human intervention [36]. In a separate study, we evaluated detection of transition periods in scripted and unscripted environments and found that the recognition algorithm was able to detect transition periods greater than 80% of the time, with a false positive rate of less than 15% [37]. However, our completed prompting system would require users to have infrared motion sensors installed in their homes.…”

Learning Activity Predictors from Sensor Data: Algorithms, Evaluation, and Applications

“…A number of techniques have been proposed for this task. Some approaches are unsupervised and utilize object-use fingerprints [64], [65] or statistical change point detection [66], [67]. …”