6  Limitations and Future Recommendations

Keywords

travel behavior, mental health, motivation, suicidality, activity types, DBSCAN-TE

There are some important limitations and future considerations following the analysis of the BYU CAPS data. These limitations are due to potential issues with data collection, lack of activities duration from the DBSCAN-TE algorithm, and the inability to confirm activity engagement for the participants.

6.1 Data Collection

The primary limitation of this study pertains to the quality of the data of the participants. Since participants participated for varying lengths of time and their phones were not always on to collect LBS data, the data was sometimes sparse. Even though a substantial amount of data had to be discarded due to poor quality, influenced by factors such as participants turning off their phones or the app failing to record data accurately, we did our best to accurately account for the well-being and activity patterns of the individuals. The inconsistency in data collection may have made it difficult for the DBSCAN-TE algorithm to perfectly identify activities. While it performed with 91.5% accuracy, some of the activity days that were manually checked appeared to be missing identified activities. Additionally, since the userID-activity days needed corresponding mental health data and activity data, we tried to account for the gaps by imputing the number of activities for a seven-day rolling average number of activities and by calculating the convex hull area and the distance traveled.

To address these limitations in future studies, improving data quality would be paramount. Strategies could include implementing measures to encourage consistent phone usage among participants or enhancing the app’s reliability in recording data accurately. Additionally, incorporating redundancy measures within the data collection process, such as cross-referencing data from multiple sources or employing complementary data collection methods, could help mitigate the impact of sporadic data collection. Moreover, refining the DBSCAN-TE algorithm to improve accuracy in identifying activities, potentially through machine learning techniques or incorporating additional contextual information, could enhance the reliability of activity data. By prioritizing efforts to enhance data quality and implementing more robust data collection and processing procedures, future studies can better capture and analyze individuals’ mental health and activity patterns, thereby yielding additional findings.

6.2 Activity Duration

The study also lacks detailed information on the duration of participants’ activities. While the dataset indicates that certain activities occurred, it does not specify how long participants spent engaging in these activities. For example, we know whether participants visited parks, but not the duration of their stay at the park. This limitation means we cannot accurately assess the impact of time spent in specific environments on mental well-being. Additionally, since the algorithm determined activities based on relatively stationary LBS data, we did not account for instances where participants might have merely passed through green-spaces or parks without spending significant time there. This omission further complicates our understanding of the relationship between activities and mental health. These limitations highlight the challenges in using mobile-based data collection for mental health research.

Future studies could focus on improving the accuracy of location-tracking algorithms, ensuring consistent data collection, and capturing detailed activity duration to provide a more comprehensive understanding of the interaction between travel behavior and mental well-being.

6.3 Activity Diaries

One other potential limitation of our study is the inability to confirm which specific activities individuals participated in throughout the day. While we have identified activities using the DBSCAN-TE algorithm, there is a possibility that certain activities were not accurately identified by the algorithm, leading to their omission from our analysis. This lack of precision could potentially result in some activities going undetected, thereby limiting the comprehensiveness of our activity analysis.

To address this limitation in future research, integrating activity diaries into survey methodology could prove beneficial. By incorporating activity diaries, participants would have the opportunity to provide detailed accounts of their daily activities, including specific locations where these activities took place. This additional information could enhance the completeness and accuracy of our activity analysis, as it would provide valuable insights into the types and locations of activities individuals engage in throughout the day.

6.4 Overall Recommendations

To build on the findings of this study and address its limitations, future research should prioritize the following:

  • Enhance Data Collection Methods: Consider using multiple data sources or complementary methods to ensure comprehensive data capture. Invest in refining activity identification algorithms and explore advanced machine learning techniques to improve accuracy and reduce gaps in activity data.

  • Capture Detailed Activity Duration: Focus on capturing both activity types and durations by implementing time-tracking features to better understand the impact of activities on mental well-being.

  • Implement Participant Activity Diaries: Integrate activity diaries to confirm activity engagement and duration. Enhance the completeness and accuracy of daily activity engagement.

By addressing these recommendations, future studies can improve the quality of data, enhance the accuracy of activity analysis, and provide a more comprehensive understanding of the interplay between activity patterns and mental health.