Lab Reader
You can find papers:
Marquette Lab (Thursday) - Computer Science/Mental Health/Veterans
Masters Theses
PhD Dissertations
Cognitive Task Analysis & Cognitive Walthroughs
It was a good discussion today, it took us a while to get on the same page, but I think it was worth it. As I mentioned today, I think your focus groups should really involve peer mentors, not so much veterans. You are looking for Subject Matter Experts (SMEs) input, and the peer mentors because of their experience across multiple veterans experiencing crisis will have more insight about patterns than focus groups with veterans.
Second, I think because we are basically trying to compare machine learning early warning signs (and alerts) to peer mentors' early warning signs (and intervention), having a couple of focus groups with the peer mentors sets you up to do the next task, which is increasingly sounding like a set of simulations where you present EMA data as a set of snapshots to the peer mentors and ask where they think there are early warning sign indicators (that would predict later long term crisis as you've defined it with the PCL5 cut score for acutely symptomatic, and thus at highest risk). This type of approach, which simulates an actual cognitive task (i.e. in the real world, the peer mentor would go through the available prior data they have in their memory about a veteran, and then take in new information [the weekly check in] and make a determination about the need for higher intensity intervention) is usually called a Cognitive Task Analysis. There is a substantial literature on this approach, one good example is here.
The Sensemaking Process and Leverage Points for Analyst Technology as Identified Through Cognitive Task Analysis
Also, strategies from a related idea in the useability field may be important in this case, called Cognitive Walk Through
https://www.interaction-design.org/literature/article/how-to-conduct-a-cognitive-walkthrough
Between these two ideas, you should be able to operationally define a session where you are walking the peer mentors through several sets of data and asking them to mark where (really when - or which week) would be a point at which an early warning sign is strong enough to merit additional intervention. There are some issues with this, among which the peer mentors may not care much about false positives, and suggest intervening at the slightest hint of trouble. We may need an approach to constrain that response so that they are focused on identifying the best single point in time for intervention, etc.
All of this goes to implicit theories about mental health crisis prediction in the minds of the peer mentors. You will want to do some reading on implict or "lay persons" theories, and how those are tapped into - a point we can discuss more in the next couple of weeks.
Dr. Franco's Papers can be accessed through any of these portals:
Google Scholar
Academia.edu
ResearchGate
Marquette Lab (Thursday) - Computer Science/Mental Health/Veterans
Masters Theses
- Casey's Thesis: mPeer: A Mobile Health Approach to Monitoring PTSD in Veterans
- Tanvir's Thesis: Towards Predicting Risky Behavior Among Veterans With PTSD by Analyzing Gesture Patterns
- Kowser's Thesis: mCAT: Development of a Generic mHealth Tool for Continuous Assessment, Automatic Intervention, and Analysis
PhD Dissertations
- Rizwana's Dissertation: A Mobile Health Approach to Assist Veterans Reintegrating into Civilian Life"
- Priyanka's Dissertation: Predicting Mental Health Crisis in Veterans: Early Warning Signs,Precursors and Protective Factors
Published Papers
- ISCRAM2018 conference paper: Detecting & Visualizing Crisis Events in Human Systems: an mHealth Approach with High Risk Veterans
- ISCRAM2016 conference paper: Crisis Warning Signs in mHealth for Military Veterans: A Collaborative Design Approach
Veteran Engagement Grants
- mPeer (Advancing a Healthier Wisconsin)
- iPeer (Advancing a Healthier Wisconsin)
- Milwaukee PROMPT (Robert Wood Johnson Foundation - Clinical Scholars Program)
- PCORI
- QRF+ (Advancing a Healthier Wisconsin)
Cognitive Task Analysis & Cognitive Walthroughs
It was a good discussion today, it took us a while to get on the same page, but I think it was worth it. As I mentioned today, I think your focus groups should really involve peer mentors, not so much veterans. You are looking for Subject Matter Experts (SMEs) input, and the peer mentors because of their experience across multiple veterans experiencing crisis will have more insight about patterns than focus groups with veterans.
Second, I think because we are basically trying to compare machine learning early warning signs (and alerts) to peer mentors' early warning signs (and intervention), having a couple of focus groups with the peer mentors sets you up to do the next task, which is increasingly sounding like a set of simulations where you present EMA data as a set of snapshots to the peer mentors and ask where they think there are early warning sign indicators (that would predict later long term crisis as you've defined it with the PCL5 cut score for acutely symptomatic, and thus at highest risk). This type of approach, which simulates an actual cognitive task (i.e. in the real world, the peer mentor would go through the available prior data they have in their memory about a veteran, and then take in new information [the weekly check in] and make a determination about the need for higher intensity intervention) is usually called a Cognitive Task Analysis. There is a substantial literature on this approach, one good example is here.
The Sensemaking Process and Leverage Points for Analyst Technology as Identified Through Cognitive Task Analysis
Also, strategies from a related idea in the useability field may be important in this case, called Cognitive Walk Through
https://www.interaction-design.org/literature/article/how-to-conduct-a-cognitive-walkthrough
Between these two ideas, you should be able to operationally define a session where you are walking the peer mentors through several sets of data and asking them to mark where (really when - or which week) would be a point at which an early warning sign is strong enough to merit additional intervention. There are some issues with this, among which the peer mentors may not care much about false positives, and suggest intervening at the slightest hint of trouble. We may need an approach to constrain that response so that they are focused on identifying the best single point in time for intervention, etc.
All of this goes to implicit theories about mental health crisis prediction in the minds of the peer mentors. You will want to do some reading on implict or "lay persons" theories, and how those are tapped into - a point we can discuss more in the next couple of weeks.
Dr. Franco's Papers can be accessed through any of these portals:
Google Scholar
Academia.edu
ResearchGate