Conceptual |
Use spatial energetics data to isolate specific geographic contexts driving behaviors in certain individuals; algorithms can identify places individuals spend time and trips taken, use to generate hypotheses about their relative importance for energetics
Develop theory and experimental designs that utilize bouts of activity and sedentary behavior rather than day or week aggregations
Combine spatial energetics data with electronic mobility surveys to understand motivations for visiting certain locations, or use of specific routes/transit modes to account for selective daily mobility bias
Aggregate GPS data into known participant anchor points and examine behavior around these anchor points to account for selective daily mobility bias
|
Technical |
Validate and update GIS layers to maximize accuracy
Identify factors related to compliance with GPS procedures to increase compliance in these populations or to statistically adjust for lack of compliance
Use software that automatically cleans large GPS datasets by imputing and reducing scatter, such as UCSD’s Personal Activity and Location Measurement System (PALMS)
Apply machine learning techniques derived from training data collected during free living to support classification of key behaviors e.g., driving, walking, and cycling
|
Analytical |
Create tools for efficiently linking, processing, and analyzing diverse and complex data streams
Apply approaches to rapidly visualize GPS points, processed accelerometry data, and travel diaries
Develop cross-classified multilevel modeling approaches when analyzing data to account for correlations of measures within individuals
Borrow statistical and computing methods from other fields dealing with “big data,” such as genetics, economics, and machine learning
|
Ethical |
Develop standardized protocols for data sharing and data security that ensure that the privacy of study participants is protected
Consider Certificates of Confidentiality to ensure participant data privacy
Remove or mask spatial data in sensitive locations and develop common standards for data masking to begin working towards open data
Communicate clearly with participants so that they have a full understanding of the data they are contributing and how those data will be used, as well as the ability to opt out of studies at any time
|