[go: nahoru, domu]

Jump to content

Chapel Hill Conference on Depression, Bipolar Disorder, and Suicidality/2019/Day 2

From Wikiversity

This page is under development. More notes and resources will be added soon.

Day 2 is our main conference day! The schedule for the day as well as any applicable notes and resources can be found below.

Welcome And Introduction (8:00 AM)

[edit | edit source]

Mitchell J. Prinstein, Ph.D. & Eric A. Youngstrom, Ph.D.

Discussion Of Flipped Keynote #1 (8:15 AM)

[edit | edit source]

Integrative Data Analysis View before the meeting here. Andrea Hussong, Ph.D.

Notes

[edit | edit source]
Click "Expand" for notes
  • Primer on integrative data analysis (IDA)
  • Different from meta-analysis because meta analysis deals more with summary statistics
  • Advantages of IDA
    • Helps test more novel hypotheses
    • More efficient use of resources- still not fast because you have to learn how someone keeps data
    • Increases statistical power
    • Greater study integration/replication
  • Typical steps in IDA-
    • Explicate theoretical questions of interest (start with a question, not with the data
  • Feasibility analysis
    • Example- intergenerational pathway of depression, but there isn’t a lot of data across generations
    • So, we need to to fit 3 studies of different age periods
    • Map designs together and try to line up based on what’s important to you (comparable age blocks, for instance)
    • IDA is a methodological design more than a statistical technique
    • Might find that each study used a different edited version of the same depression measure
    • Next, develop pool of items including identical items or ones similar enough to reasonably harmonized, and then unique items (which get kept in IRA)
    • Need-
      • Some common items (identical or harmonized)
    • Next, psychometric harmonization
      • Moderated non-linear factor analysis from Dan Bauer- puts together Item response theory and confirmatory factor analysis
        • What items go together and how do items function differently (across studies, age, gender, etc)
      • Not really worried about lining up all items, but want lined up factors
  • Suggestions for conversations-
    • Spend time laying out the big picture areas of data overlap to understand the possible types of questions- sub populations, rare behaviors, or developmental age spans
    • Think carefully about final model and control variables that are important and will need to be pulled and harmonized as well
    • Plan to spend time on data management
  • Q&A
    • What about multiple informants that don’t converge (e.g., CBCL, YSR)? Would you want them in the same thing or use them as controls?
      • It depends- do you want to see how they’re similar and use a trifactor model to try to answer that question. Trifactor model allows you to not have all of the same reporters (if one study has all 3, for instance)
      • Other times, it may just be better to pick a reported
    • Missing data- what do you do if sources differ there?
      • You get to make decisions because you work with the raw data, so team science is important. Missing data comes from design, attrition, etc. and differs by study, and once you harmonize the data it’s business as usual
    • Pros and cons of using a single time point to derive factor scores?
      • During psychometric modeling, you don’t want to have a lot of nested data so there is usually one observation per person
      • Don’t really want to use just one wave, but have to pay attention to nesting
    • Limit of what you can combine? Are there clear boundaries?
      • If completely compounded with study, there’s not much you can do but don’t want to pull study effects, so need to have some studies with multiple methods
    • Advice for someone putting a grant together to help with reviewer concerns?
      • How much of the IDA do I do before submitting?
      • The idea that you don’t have to have identical measures needs to be every 2-3 pages, and have to show it a little in the grant at a simple level (like a pilot study showing that you can put the data together)
      • So, pilot analyses are critical
      • Also helpful to have someone who has done it before, and also make sure you have time and expertise for data management
    • Study effects? Where do differences comes from?
      • IDA is about study effects- can see if study effects go away when you control for them (like gender)
      • Politics as difficult as the analyses because people have opinions about why data turns out one way or another

Table Discussion: Integrating Data: Troubleshooting/Brainstorming (8:45 AM)

[edit | edit source]

Breakouts For Depression, Bipolar, Or Suicide Projects

Notes

[edit | edit source]
Click "Expand" for notes on bipolar
  • Do we have datasets we should be sharing?
    • What’s our wishlist for datasets we hope others have?
  • Challenges in combining across multiple datasets
  • June Gruber: actively doing data collection in college freshmen during transition
    • Lots of attention on college mental health crisis
    • They got $ from university to run this study
    • Learned a lot about online study surveys from masses, checking compliance
    • It would be cool to set things up across multiple labs and share what you get
      • Same measures across sites
      • Google docs was really helpful in collaborating across sites
      • Interested in data on emerging adults
  • Eric asked people to share YMRS data a long time ago but everything was different
    • Some people didn’t even put #s of questions
    • A Google sheet with what our datasets look like would be helpful
      • Variable names for each # to figure out what each person’s looks like
      • Drop in an N
  • Anna: having a question first is helpful; selecting items that are in that interest would be helpful; combining datasets based on question
  • We have at risk and after transition but not a lot collecting transition
    • Combine datasets of different age ranges
    • Clinical datasets that are longitudinal; some people would end up with BD
    • Transition from at risk to disorder is hard to get a grant for; looking at risk factors predating onset is difficult as a result; if we can combine datasets that would be super helpful for looking at people in prodromal phase
  • “Geography is destiny”-- tagline at Dartmouth. Eric is convinced we define things differently across geography
  • Linking several sites’ EMRs would be helpful
  • June has some longitudinal data on the college students but not all
    • Feels underpowered on their own
  • Google doc with one tab for each team
  • There’s a protocol for projects on BD in late life; we could ask them to share and then do the same with childhood-adolescence-early adulthood
  • Create a steering committee
  • A Google doc with meta-data wouldn’t require IRB approval
  • In the CDI, schools eliminate suicide item b/c they’re worried it’ll put it in kids’ heads
  • Organize meta-data in a doc; people can jot notes, have discussions
    • Make a Wiki, build a page on Wikiversity
      • Here’s current version of LAMS dataset; here are the questions; use discuss or talk page for open questions; as those are resolve
Click "Expand" for notes on depression
  • We can use this to identify key components of depression to decide that we target in our treatment
  • Generally being able to pull neuroimaging data as it pertains to depression
    • Task given, region activated
    • Many use different tasks and have small samples. There are lots of card guessing task studies
    • Lots of face reactivity data
  • Moderators of treatment outcomes- increased stability by integrating studies
  • What sample size is good for this? We have a lot of small sample sets that would be good if helpful
  • What about when one of your variables of interest vary systematically across studies?
  • Pooling data can help look at smaller minorities rather than just white vs. non-white, for example
    • Literature is mixed on racial minorities  
    • What about minority stress? How similar could the items be to combine?
    • Community has unique impact on stress, which could yield interesting study effects. What tips off the stress? How do we integrate high level factors like state laws with the individual’s data
  • Data we have?
    • Longitudinal sample of 212 with parents/kids
    • Sample all of Mexican origin ~200 with fMRI with cortisol
    • Alloy- N started around 600 then followed around 8 years spilt in half by male/female African American/Caucasian
    • 800 girls 6th grade to end of high school
    • 500 kids followed from 12-22
  • Different studies have different populations and geography so if we combine those is there a danger of the non-overlapping groups being conflated with study effects
  • Is there funding for this? Maybe add on to existing R01 to address diversity
    • Do they want people to come up with common measures or how to put together existing datasets? Putting together existing ones, and how can we increase the amount of overlapping data
  • Method works better with a small number of large studies rather than a large number of small studies
  • What should we target in treatment of depression? Rates are rising Judy Garber
  • Inform treatment studies with studies looking at mechanisms of change and constructs of depression
  • Cohort effects of documented rise in anxiety?

Break & Poster Viewing (Session 1) (9:15 AM)

[edit | edit source]

To view the posters on OSF, click here.

Click "Expand" for list of session 1 posters

1. Emotional response inhibition interacts with momentary negative affect to predict nonsuicidal self-injury urges

Taylor A. Burke, Kenneth J.D. Allen, Ryan W. Carpenter, David Siegel, Marin Kautz, & Lauren Alloy

2. Predictors of sleep quality: Effects of perceived stress exposure and the additive influence of six serotonergic polymorphisms

Gail Corneau and Suzanne Vrshek-Schallhorn

3. Attentional Biases and Individual Differences in Adolescents’ Mood and Physiological Reactivity to Stress

Cope Feurer, Kiera M. James, Claire E. Foster, & Brandon E. Gibb

4. Detecting suicidal thoughts: The power of ecological momentary assessment

Ilana Gratch, Tse-Hwei Jonathan Choo, Hanga Galfalvy, Liat Itzhaky, & Barbara Stanley

5. Machine learning predicts condom use among men who have sex with men

Hunter Hahn and Woo-Young Ahn

6. Steady-State Visual Evoked Potentials Reveal Deficits in the Ability to Inhibit Attention to Negative Interpersonal Stimuli in Adolescents Who Engage in Nonsuicidal Self-Injury

Kiera M. James and Brandon E. Gibb

7. Influences of Childhood Maltreatment, Emotion Dysregulation, and Recent Life Stress on Pregnant Women's Respiratory Sinus Arrhythmia

Parisa R. Kaliush, Elisabeth Conradt, Sarah Terrell, Robert D. Vlisides-Henry, Dylan Neff, Betty Lin, Nila Shakiba, & Sheila E. Crowell

8. Sex differences in the link between current suicidal ideation and guilt in youth: ecological momentary assessment technology approach.

Anastacia Y. Kudinova, Leslie Brick, & Gracie A. Jenkins

9. Emotion dysregulation and functional connectivity in children with and without a history of depression

Katherine Lopez, Joan Luby, Andrew Belden, & Deanna Barch

10. The longitudinal associations of inflammatory biomarkers and depression revisited: systematic review, meta-analysis and meta-regression

Naoise Mac Giollabhui, Tommy Ng, Lauren M. Ellman, & Lauren B. Alloy

11. Influence of Parental Depressive and Anxiety Disorders on Youth Personality and Psychopathology: Mediational Path Models Via Youth Personality

Daniel M. Mackin, Brady D. Nelson, Megan Finsaas, Greg Perlman, Roman Kotov, & Daniel N. Klein

12. Responses to Affect Subtypes Differentially Associate with Anxious and Depressive Symptom Severity

Rebekah Mennies, Samantha Birk, Julia Case, & Thomas Olino

13. Tell me yet again it's okay: Predicting reassurance seeking from anxiety and intolerance of uncertainty

Allison E. Meyera and John F. Curry

14. Bidirectional Associations Between Inflammatory Biomarkers and Depression Symptoms in Adolescents: Potential Causal Relationships

Daniel P. Moriarity, Marin M. Kautz, Naoise Mac Giollabhui, Joshua Klugman, Christopher L. Coe, Lauren M. Ellman, Lyn Y. Abramson, & Lauren B. Alloy

Data Blitz #1: Three Fast Talks, Three Slides Each, Followed By Discussion (9:30 AM)

[edit | edit source]

Talk #1: Dispositional And Environmental Predictors Of And Structural And Functional Correlates Of Adolescent Self-Harm?

[edit | edit source]

Ted Beauchaine, Ph.D., Ohio State University

Notes

[edit | edit source]
Click "Expand" for notes
  • Talking about pilot data of adolescent girls who engage in self harm
  • Neurodevelopmental theory of NSSI, SB, and BPD
    • 2009 paper in Psychological Bulletin-
      • Most likely to develop when trait impulsivity presenting as combined ADHD symptoms interact with environmental stressors
      • About 1 in 5 girls engage in self injury, around 30% in maltreated and ADHD samples. ADHD and maltreated rates rise to around 50%
      • Suicide attempt rates rise to around 10% for maltreated or have ADHD, but combined has rates around 35%
  • NSSI, Impulsivity, and Neural responses to incentives
    • Study had 19 girls who engaged in self injury and 19 control
    • The experimental group had less activation to reward cues than control
  • NSSI, Emotion Dysregulation, and Cortical Structure
    • 20 self-injuring and 20 control.
    • Found smaller bilateral insula and right inferior frontal gyrus that serve self and emotion regulation functions

Talk #2: Substantive Questions Tested With Applied Measurement Methods

[edit | edit source]

Tom Olino, Ph.D., Temple University

Notes

[edit | edit source]
Click "Expand" for notes
  • Old study examined generational effects on psychopathology over 3 generations
    • Used parent report and could be influenced by mood state effects of the parents. Need to be able to test what is a true effect and what could be a confounding artifact
    • Can use measurement invariance methods
    • Is there equivalence of a construct or factor? Are our parent reports of kid behavior based on their own experiences with psychopathology
  • ABCD project
    • Over 8,000 mother reports of child and self report of their own internalizing using CBCL and ASR
    • Identified 5 dimensions-
      • Internalizing problems
      • Detachment
      • Somatization
      • Externalizing
      • Neurodevelopmental problems
    • Used moderated non-linear factor analysis to determine measurement invariance and extend it
    • Results- very small impact of maternal bias based on mother’s detachment and internalizing
      • Minimal evidence of for maternal mood-state bias influencing maternal report of youth problems
    • Future direction could look at demographic effects

Future Directions And Discussion

[edit | edit source]

Elizabeth Mccauley, Ph.D., University Of Washington

Notes

[edit | edit source]
Click "Expand" for notes
  • Goal- to think about importance of your work in developmental psychopathology, and what is needed in the real world to learn from this work
  • Data suggests in clinical realm that we should be moving from categorical to dimensional measurement
  • We also need to rethink how we talk about and work about psychopathology, we are being urged to think about common elements and traits like emotion regulation, disordered thinking, etc. that are common and manifest as what we call comorbidity
  • These traits are shaped by environmental and biological mechanisms that lead to some individuals having problems
  • Challenge- timing is off- we have a dramatic increase in clinical demand
    • Reports of 50% increase in last 5-10 years of suicidality in girls, 30% in boys
  • Clinicians need guidance in how to engage in clinical work- intervention and classification models
    • How do we talk and think about these things and tell the public about them?
    • Where do we do the work? Integrated settings? Schools? Community wide education/prevention programs?
    • What do we treat? Top problems? Traits (impulsivity, emotion regulation, etc.)? Key symptoms? By disorder construct (ADHD, anxiety)?
    • Who do we treat? In depression, focus is on patient and not parents so much. Should we just treat parents? They struggle with how to manage. Do we treat community systems?
    • How do we treat? We have manualized interventions for most common things that work- we need access, education, and improving access that kids have to these intervention. Or do we use common elements treatment?
  • What kind of change is needed? How do we define and measure change? What is “Clinically significant change?” Should we focus on it? Big change for a few kids or small change in big group? Functional improvement vs. symptom change?
  • How do we address access? Stepped care models?
  • Where do we stand on treatment of suicidality and depression?
    • Suicidality- more promising outlook than depression. DBT, mentalization, CBT showing promising outcomes
      • Lessons learned from DBT vs. Supportive treatment study
        • Need to have a more chronic care approach given persistence of suicidal ideation
        • DBT is focused on improved emotion regulation and it was improved as well as sleep, and parent engagement/training
    • Depression?
      • Disappointing effect sizes and about 30% youth who don’t respond to any treatment
      • Looking at new care pathways- Screen/Eval then CAMS then brief intervention then BA, IPT, CBT then DBT
      • Enhancing parent training
      • Challenge- need strong data to rethink how to do clinical work and train trainees. How we educate consumers and fiscal reimbursement systems also important
  • Audience ideas for where depression research needs to go-
    • Should we think of psychopathology in terms of severity or categorically? Continuous idea is not new, why are we not further along? We also can’t trash categorical system as we have a history full of people who fit the description of disorders. What do you do with severity in the clinic? Practical ramifications of using severity in clinical settings
    • Advantage of dimensional perspective- easier to integrate heterogeneous groups and data so that we can define phenotypes well instead of just saying depression
    • What are the questions clinicians want answers and the form in which they want them?
    • Combining mechanistic approach and clinical trials- what kind of mechanism to investigate is the question.
    • Need to encourage clinicians to gather repeated data so that we can see when patients exceed typical levels of stress or depression so it’s not comparing a patient to a sample but to themselves
      • Don’t see this happen as much in kid research
      • Ecological momentary assessment?
    • Having a more symptom, domain-specific way of conceptualizing what would be helpful to patients (vs. dx-based) would be helpful, b/c this maps on to what we already do as clinicians (knowing the symptoms that are most distressing to them, etc.)
    • Can identify individuals who are at risk earlier with good prevention practices. Need to focus more on prevention than treatment
      • Early intervention!!
    • Multiple mechanisms at play when treating a real person. No initial CBT trials impacted mechanisms but were kept because they helped with symptoms. We may slow down opportunities for treatments if we focus on the mechanisms than outcomes
    • How can school counselors continue working with kids and not be in fear because any risk is a great risk
      • Watching kid over time might be a helpful way to figure this out
    • Symptom profiles in different dx; how symptoms look different across disorders

Current NIMH Priorities & Discussion (10:30 AM)

[edit | edit source]

Stacia Friedman-Hill, Ph.D., National Institute Of Mental Health Eric Murphy, Ph.D., National Institute Of Mental Health

Notes

[edit | edit source]
Click "Expand" for notes
  • Over the summer, NIH released a policy about diversity and inclusivity, how it’s valued
    • Will look at panels/participants to make sure it’s diverse; end “Manels”
    • Will only attend workshops/conferences respecting diversity
  • Money, budget
    • Continuing resolution; fiscal year starts Oct. 1
      • No approved FY20 budget yet; congress approved a continuing resolution
    • Budget is different than appropriations
    • Congress agreed on broad outlines of a budget for 2020 and 2021
    • Appropriations = what each agency will finally end up with
    • Budget sometimes passed in one big piece but 12 diff parts
      • House has already approved their package for 10/12 pieces of the budget
      • Things still in committee in the Senate; need to align House and Senate
    • Cautiously optimistic looking at what’s been proposed
      • Possibility of increase in funding for NIH and NIMH
    • Expect that success rates will stay on trend, looking for ways to manage that
    • 2015 last strategic plan released-- is now a virtual, living document on NIH site
      • Early 2020 update
  • Ways to get the pulse of NIMH and things they might be interested in funding
    • Join the listserv
    • Director’s messages, Josh’s blog
    • Funding announcements and notices
      • Yesterday: notice of intent to publish FOA related to rapid-acting interventions for suicide
    • Look at meeting summaries: website → news and events → meeting summary
      • Questions raised during workshop, conclusions
      • Often hold workshops thinking about gaps in portfolio
    • Before an RFA or FOA is published, has to be cleared by council (meets 3x/year)
      • List of council clearances on website (brief description of what FOA might be about)
      • Not everything that has been cleared will end up as a fundable FOA
  • Research priorities
  • Methods
    • Intensive longitudinal monitoring, understanding signals for different temporal levels of proximal risk; how to pick up different types of signals; especially useful for things like suicide, psychotic break, relapse; understanding where potential interventions might be useful based on signals we’re not aware of yet
    • Improved classification of SI and relationship to suicide attempts (including in youth); what kids are able to verbalize and how that relates to fleeting SI or actual risk
    • Improved ecological validity of real-time stressors (e.g. monitoring social media and social interactions); understanding in real time how the experience of stress in real-life environment relates to some of these outcomes measures
      • Social media, social interactions and ways to understand that and classify those; understanding relationships
  • Concepts
    • Concepts: Considering translational pipeline of research; understanding how predictive biomarkers may be related to development of novel interventions and how they could be tailored to different stages, different populations (e.g., what kind of intervention may be useful in the ER vs. outpatient)
    • Understanding age-specific risks; don’t just assume that young adult phenotype translates to child or adult
    • Demographic risk in special populations (sexual and gender minorities, racial and underrepresented minorities)
    • Social interactions and context (e.g., peer and family, social contagion); understand naturalistic settings in which people experience issues; contextual factors
    • Sex differences in all of the above, sex x age interactions; want to be powered to look at these things and get an idea of where there are differences even if we don’t have apriori thoughts
  • Not of interest
    • Small sample sizes
      • Power studies for the question you’re asking; need a lot more power to look at complex relationships, have more sophisticated statistical models; well-justified N for project
    • Candidate genes, polygenic risk scores asked on poorly supported components
      • Need a well-documented level of effect to say you’re going to target a specific risk gene or profile
    • Symptom-based studies
      • RDOC & DSM divides…
      • Symptoms in and of themselves won’t always be useful in terms of what’s going on
      • Part of picture, can be combined with things that are easier to measure more objectively but purely symptom-based studies don’t have as much methodological rigor
    • Examination of single risk factors or well-studied risk factors
      • Similarly...we understand that there are a lot of depths of context that you need to understand some questions about risk
      • Examining a single risk factor doesn’t drive a lot of innovation in the field
      • Understand context in which these things happen
      • If you think a specific risk factor has been overlooked, give the whole picture, why you think that
    • Studies focused solely on group-level risk and static risk
      • We know these things ebb and flow and groups are at best tenuously defined; so many different ways to understand a group; taking a super heterogeneous risk vs/ someone else doesn’t drive as much innovation
      • Risk factors predicting lifetime prevalence are helpful but not moving us towards what is driving underlying psychopathology
  • Several large cohort studies exist that will provide public data sources; NIMH happy to have you delve into these
  • RDoC
    • Research framework
    • Set of domains chosen mostly because they had a fairly good idea of neural under circuitry of those
    • Examples of what you can do with RDoC but not all you can do (doesn’t have to be written in the matrix)
    • NIMH happy to talk about if a project feels RDoC-y, not where in the matrix it has to fit
    • Gives us a different way to think about how populations are broken up and where you get useful signal from different subcategories within groups
      • Allows you to break out bio-types that we know exist across all kinds of disorders but may have similarities within certain domains (whether or not they represent dx category that an individual may be labeled with)
    • RDoC recognizes that there’s not a single target; it will be multivariate; we look at patterns within domains
      • Expect it to explain heterogeneity, comorbidity, transdiagnostic approach where interventions may be helpful across multiple domains, dx
    • Show a tool either validates or engages a target; those can be cognitive of behavioral, don’t always have to be neural signatures
      • Being able to show you’re engaging that target w/whatever intervention you propose informs a go/no-go decision
      • Once you show target engagement you can go on to full clinical trial
  • *Helpful graph of which kinds of grants fall under which headings-- will have access to later
  • Guidelines for conducting research w/participants at elevated risk for suicide
    • Exclusion criteria of SI is common but not helpful; would be great if everyone included
    • Lots of useful info on how to do these kinds of projects
    • Lower barriers for studies asking these questions
  • *Program officers and contact info at NIMH for different areas
  • What is gold standard for classifying research as a clinical trial?
    • It’s complicated; within clinical trials there are diff categories; whether or not you randomize to a treatment group classifies as a “classic” clinical trial; some things are mechanistic clinical trials, don’t count as a classic trial
      • May be using an intervention not for clinical tx but to show that you can engage a pathway (e.g., using TMS to demonstrate a particular brain region important to task performance)- considered a mechanistic clinical trial
    • Read funding announcements for clinical trials very carefully to see what kind of trials they’ll accept
      • Some things you may use are “procedures” and not tx
      • Just b/c you randomize subjects doesn’t mean it’s a clinical trial
    • NIMH can send some of the links; there’s a series of 70 or 80 FAQs
      • Definitions are NIH-wide
      • There’s a webpage with a branching series of questions to help you determine which kind of clinical trial you might have
      • Check w/program officers; sometimes the meat of the methods helps determine if it’ll be considered a CT or not
  • Thinking behind go/no-go decisions?
    • Idea is that being able to show level of target engagement is a necessary piece to go forward
    • Need to demonstrate that there’s some evidence that a target is relevant
    • Replication crisis-- if we have a lot of people incentivized to find what they say they’re going to find, what does this create in the field?
      • Not that different than people who get a successful R01 will probably get another R01; 2 years for first part, perhaps shortens timeline, makes it more imperative to crank out results quickly
      • NIMH funds with the hopes that these will prove to be effective; won’t give additional 3 years if it looks like intervention won’t show engagement, though
    • NIMH has funded “fast-fail”; idea that target engagement will be demonstrated before additional funding
      • Not intended as punishment; idea of experimental therapeutics is that question should be posed in a way that suggests regardless of results, it’s important information that will move the field forward
        • If something doesn’t work, we want to know why, that we can move on and that it’s not just chance
          • Understanding a null result isn’t necessarily a bad thing
      • Null result is just as important
      • Was circuit/target engaged, and if not, it can suggest what future studies should be
    • If we back away from symptoms, how do we reconnect this to clinical care?
      • Is right way of approaching this to do “yes, and…”? (e.g., yes, we have symptom measures, AND we have biological measures etc.)
      • It’s not that they don’t want to see symptoms, they just don’t want to see them only
      • Show target is relevant by showing it’s clinically relevant-- which brings us back to symptoms and function
  • Going back to small N and low base rates…
    • How can we address sophisticated questions using large samples with limits of budget, years?
      • Consortium approach- put in something that allows you to get around budget constraints
      • Some people ask for more money than suggested when science requires it-- recognize that there are serious questions around how much $ is needed to answer certain questions
    • If you could imagine large, multisite study enhanced for risk of suicide (based on past attempts, diagnosis, impulsivity), what could you get out of it? Is there enough agreement in the field on measures, etc.? (QUESTION FROM NIMH)
      • Field is ready for it-- can either go after really big, healthy samples, or high risk samples
        • Huge increase in death rates after people leave psychiatric hospitals
      • People are working on proposals for this!
      • What kind of power do you really need?
        • Multi-site studies, collaborative R01
        • Proliferation of main-effects models
          • Looking at complex interactions between factors and over time
          • Within-person variability
        • Subtypes and heterogeneity within attempters: multi-site approaches may be a more scientifically justifiable way of looking at it
        • Large samples followed over time to ensure base rates, proximal risk (could be 5-sites or 20-sites); this happens in imaging research
      • Does suicide risk look the same across diagnoses?
        • Exploring heterogeneity
  • Who are the right people to get together, invite to conversations like this?
    • PATIENTS!! And families
    • Schools (points about re-entry after patient care in teens, kids and engaging them); would like to hear more about prevention and early intervention
    • ER doctors

Poster Viewing (Session 1) (12:15 PM)

[edit | edit source]

To view the posters on OSF, click here.

Click "Expand" for list of session 1 posters

1. Emotional response inhibition interacts with momentary negative affect to predict nonsuicidal self-injury urges

Taylor A. Burke, Kenneth J.D. Allen, Ryan W. Carpenter, David Siegel, Marin Kautz, & Lauren Alloy

2. Predictors of sleep quality: Effects of perceived stress exposure and the additive influence of six serotonergic polymorphisms

Gail Corneau and Suzanne Vrshek-Schallhorn

3. Attentional Biases and Individual Differences in Adolescents’ Mood and Physiological Reactivity to Stress

Cope Feurer, Kiera M. James, Claire E. Foster, & Brandon E. Gibb

4. Detecting suicidal thoughts: The power of ecological momentary assessment

Ilana Gratch, Tse-Hwei Jonathan Choo, Hanga Galfalvy, Liat Itzhaky, & Barbara Stanley

5. Machine learning predicts condom use among men who have sex with men

Hunter Hahn and Woo-Young Ahn

6. Steady-State Visual Evoked Potentials Reveal Deficits in the Ability to Inhibit Attention to Negative Interpersonal Stimuli in Adolescents Who Engage in Nonsuicidal Self-Injury

Kiera M. James and Brandon E. Gibb

7. Influences of Childhood Maltreatment, Emotion Dysregulation, and Recent Life Stress on Pregnant Women's Respiratory Sinus Arrhythmia

Parisa R. Kaliush, Elisabeth Conradt, Sarah Terrell, Robert D. Vlisides-Henry, Dylan Neff, Betty Lin, Nila Shakiba, & Sheila E. Crowell

8. Sex differences in the link between current suicidal ideation and guilt in youth: ecological momentary assessment technology approach.

Anastacia Y. Kudinova, Leslie Brick, & Gracie A. Jenkins

9. Emotion dysregulation and functional connectivity in children with and without a history of depression

Katherine Lopez, Joan Luby, Andrew Belden, & Deanna Barch

10. The longitudinal associations of inflammatory biomarkers and depression revisited: systematic review, meta-analysis and meta-regression

Naoise Mac Giollabhui, Tommy Ng, Lauren M. Ellman, & Lauren B. Alloy

11. Influence of Parental Depressive and Anxiety Disorders on Youth Personality and Psychopathology: Mediational Path Models Via Youth Personality

Daniel M. Mackin, Brady D. Nelson, Megan Finsaas, Greg Perlman, Roman Kotov, & Daniel N. Klein

12. Responses to Affect Subtypes Differentially Associate with Anxious and Depressive Symptom Severity

Rebekah Mennies, Samantha Birk, Julia Case, & Thomas Olino

13. Tell me yet again it's okay: Predicting reassurance seeking from anxiety and intolerance of uncertainty

Allison E. Meyera and John F. Curry

14. Bidirectional Associations Between Inflammatory Biomarkers and Depression Symptoms in Adolescents: Potential Causal Relationships

Daniel P. Moriarity, Marin M. Kautz, Naoise Mac Giollabhui, Joshua Klugman, Christopher L. Coe, Lauren M. Ellman, Lyn Y. Abramson, & Lauren B. Alloy

Discussion Of Flipped Keynote #2 (12:30 PM)

[edit | edit source]

Machine Learning View before the meeting here. Oscar Gonzalez, Ph.D.

Keynote times and topics

[edit | edit source]
Click "Expand" for breakdown

Time Slide Theme

0:00 #1 Intro

3:00 #6 Data types

3:45 #7 4 V’s of big data

6:00 #12 Area of research

12:00 #15 General outline of sections of talk

15:00 #18 The buzz about ML

24:10 #25 On Learning

35:30 #36 Performance evaluation ß emphasizing prediction over p value

39:00 #39 Is regression a ML model

45:00 #43 General model

48:30 #46 Overview of different ML models

51:00 #49 Supervised (prediction) vs. Unsupervised (descriptive)

54:45 #54 LASSO and other regularization (why would we shrink weights?)

56:00 #55 Variable selection (filters, wrappers, embedded)

57:00 #56 Decision tree & random forest

58:10 #58 Neural networks

1:00:00 #60 Stacking (“superlearning”)

1:01:50 #64 Unsupervised (clustering, PCA, mixture modeling)

1:03:40 #68 Current applications

1:04:30 #69 Ilgen et al. (2009) – decision trees predicting suicide rates

1:05:30 #70 Schnack et al. (2014) – Support Vector Machines (SVMs) discriminating bipolar, schiz, healthy

1:06:30 #71 van Loo (2014) – multiple methods, looking for subtypes of depression

1:08:45 #72 ML and psychometric assessment

1:10:00 #73 ML and mediation effects

1:12:45 #76 ML closing thoughts

1:13:45 #77 Data still king

1:15:00 #79 List of books and code for methods

Notes

[edit | edit source]
Click "Expand" for notes
  • Society produces a lot of data - how do you take it and make it useful
  • There is currently a big data science initiative because of the presence of data
  • Data science- Collection, analysis, processing, visualization, and interpretation of vast amounts of data
  • Machine learning might help your question.. Or not
    • We can use it to examine patterns in part of the data and evaluate what we find
    • Focus is on generalization
    • Can use the model in the future to predict
  • Machine learning- why?
    • Uses framework of learning
    • Rooted in results generalization
    • Focus on prediction rather than generalization
    • Exploratory
    • Significance testing may not be useful in large datasets
  • Rough idea-
    • Traditional analysis
      • Input and model go into computer and get output
    • Machine learning turns it around
      • Input and output go in computer and yield model/algorithm
  • Let’s be upfront-
    • Lots of researchers “get the bug” and find that machine learning isn’t what the expected and there’s less emphasis on significance testing and sometimes the models might be able to be interpreted easily
    • We still have typical problems- explanation of behavior, psychometric issues, measurement bias
  • Learning (Mitchell 1996)
    • Learning- performance improves based on experience
    • Need to give computer dataset to gain experience
    • Might not be able to give measure to everyone but can have a large database of the predictors
  • Is regression a machine learning model
    • Traditional focus on significance of regression coefficients and make theories about why x is related to y
    • Treat regression as machine learning algorithm
      • Develop model/learn pattern in one dataset
      • Evaluate in another
      • Not too complex
      • Prevent learning the idiosyncracies
  • Machine learning flavors
    • f(x) may be linearly or nonlinearly related to the outcome, difficult to interpret, or have few important predictors
    • Machine learning looks for relationship between predictor and outcome
  • Two types of machine learning methods
    • Supervised learning models- predictive
      • Focused on prediction of outcomes
      • Learn by example
    • Unsupervised learning methods- descriptive
  • Stacking (aka Superlearning)
    • For any prediction could use a number of regression types, neural networking, vectors
    • Stacking combines all the models and combine them because you may not know from the start which to try
  • Machine learning and assessment
    • Developing assessments to evaluate mechanisms of behavior change
  • Main message
    • Machine learning is highly exploratory so be clear about what we’re doing
    • One can be unethical with confirmatory approaches… ethical with exploratory
    • When you see a machine learning paper ask:
      • What’s the task?
      • What’s the experience?
      • What’s the performance metric?
  • What is a research question that could be answered by prediction?
  • Link to free version of ML book will be put online (gives examples in R)
    • Random forest and vector
    • Carrot (very comprehensive) and has an ebook
    • KNN
    • Packages and examples help to learn
  • What would we like to see in a year? Other questions?
    • Broader acceptance of the alternative methods
    • How do we write a grant using these alternative methods? NIH has a person specifically for quantitative psychiatry. Prediction in and of itself is less helpful than showing what you predict and showing what you do with it in terms or treatment
      • So interpretability is crucial
      • Machine learning is becoming more common
      • Machine learning is especially helpful in classification (RDoC)
      • Test and retest samples are important to test classifiers
      • Can’t just say “and we will also do machine learning” as a buzzword. Want to have a 2 stage analysis approach- main findings and looking to improve prediction
    • What is a way to evaluate how well your machine learning is doing? Better than chance? Try on different types of datasets and see how much of a loss of predictive power you have
      • Perhaps compare to gold standards
    • Is getting to a clinical level of prediction of a clinically important outcome without the clinician helpful or a goal? Depends how you look at it. How does this play out in funding?
      • Machine learning can be used to build risk calculators so we can judge the value of it by its predictive power
      • Example- grant for early risk of psychosis with digital imagery of faces and comparing them to see if there’s a relationship in risk of psychosis
    • Kaggle- give data away and say what question they want to answer and offer a prize
      • Zillow has offered largest prize ever to predict selling price of a house
      • Way of crowdsourcing
      • Youngstrom has a dataset on the open science framework, and the lasso model won but not in a clinic replication
        • Contest- dataset is up there, whoever comes up with the best method to predict clinic data wins

Resources

[edit | edit source]

Articles

[edit | edit source]

These are examples of applications of statistical learning methods in the mood disorders concept space.

Ilgen et al. (2009)[1] – decision trees predicting suicide rates

Schnack et al. (2014)[2] – Support Vector Machines (SVMs) discriminating cases with bipolar disorder, schizophrenia, and healthy controls

van Loo (2014)[3] – multiple statistical methods, looking for subtypes of depression

Books

[edit | edit source]

An Introduction to Statistical Learning with Applications in R

Applied Predictive Modeling

The Elements of Statistical Learning: Data Mining, Inference, and Prediction

Machine Learning: A Probabilistic Perspective

Break & Poster Viewing (Session 2) (1:30 PM)

[edit | edit source]

To view the posters on OSF, click here.

Click "Expand" for list of session 2 posters

1. Emotion-Related Impulsivity Relates to the p Factor of General Psychopathology

Jennifer G. Pearlstein, Charles S. Carver, Kiara R. Timpano, & Sheri L. Johnson

2. Exploring changes in rumination across hospitalization: A study of adolescent self-injurers

Olivia H. Pollak, Eugene D’Angelo, Matthew K. Nock, & Christine B. Cha

3. Control When it Counts: Executive Control Under Stress Predicts Depressive Symptoms

Meghan E. Quinn and Jutta Joormann

4. Health, crime, and social-welfare inequality cluster in a population segment: Analyses of 4 million citizens from 2 nations

Leah S. Richmond-Rakerd, Stephanie D’Souza, Signe Hald Andersen, Sean Hogan, Renate M. Houts, Richie Poulton, Sandhya Ramrakha, Avshalom Caspi, Barry J. Milne, & Terrie E. Moffitt

5. Mental health service utilization by race/ethnicity in a nationally representative sample from 2004 to 2017

Ana Sheehan, Rachel Walsh, Christina Sanzari, & Richard Liu

6. Brief Mobile Mindfulness Intervention for Ruminative Adolescents: (Very) Preliminary Results from a Randomized Control Trial

Caroline M. Swords, Eleanor Horner, Liesl Hostetter, Eliana Whitehouse, Jialun Yang, & Lori M. Hilt

7. Cerebral Blood Flow is Altered According to Mood States in Adolescents with Bipolar Disorder

Simina Toma, Bradley J. MacIntosh,; Anahit Grigorian, Lisa Fiksenbaum, Andrew D. Robertson, & Benjamin I. Goldstein

8. Experience-Sampling Approach to Emotion Differentiation Across Bipolar and Unipolar Mood Disorders: Associations with Emotion Regulation and Variability

Cynthia M. Villanueva, Douglas Mennin, Greg Murray, Renee J. Thompson, & June Gruber

9. Pregnant women’s self-reported emotion dysregulation predicts newborn neurobehavior

Robert D. Vlisides-Henry, Brendan D. Ostlund, Sarah Terrell, Mindy A. Brown, Elisabeth Conradt, & Sheila E. Crowell

10. Temporal Trends in Peer Victimization among Sexual Minority & Heterosexual Adolescents

Rachel Walsh, Christina Sanzari, Ana Sheehan, & Richard Liu

11. Reduced Hippocampal and Amygdala Volumes as Mechanisms of Stress Sensitization to Depression following Childhood Violence Exposure

David G. Weissman, Hilary K. Lambert, Alexandra M. Rodman, Matthew Peverill, Margaret A. Sheridan, & Katie A. McLaughlin

12. Borderline symptoms at age 12 signal risk for poor outcomes during the transition to adulthood: Findings from a genetically sensitive longitudinal cohort study

Jasmin Wertz, Avshalom Caspi, Antony Ambler, Louise Arseneault, Daniel W. Belsky, Andrea Danese, Helen L. Fisher, Timothy Matthews, Leah Richmond-Rakerd, & Terrie E. Moffitt

13. Examining the Relationship between Reaction to a Suicide Attempt and Subsequent Suicidal Behavior in the Context of a Suicide-Specific IOP.

K. Wolfe, K. Rial, K. Goga, C. Tran, A. Moorehead, B. Kennard, B., & G.J. Emslie

Data Blitz #2: Three Fast Talks, Three Slides Each, Followed By Discussion (2:45 PM)

[edit | edit source]

Talk #1: Digital Prediction/Prevention

[edit | edit source]

Matt Nock, Ph.D., Harvard University

Notes

[edit | edit source]
Click "Expand" for notes
  • Significant challenges in the field
    • Theories for suicide are overly simplistic
    • Designs and methods are too and are contrived
    • Treatments are weak and have inconsistent effects
    • Progress is slow, stagnant
  • Strength of predictors not improving
    • Recent meta analysis found:
    • We’ve been looking at the same risk factors for the past 50 years
    • Haven’t been observing behavior most of interest to us, naturalistic observation
    • Most studies haven’t look at outcomes of interest or in clinically meaningful window
    • Significant opportunities to ID gaps, use new approaches
      • One way to is use EMA
    • Gap: Need better method of combining risk factors
      • One study used machine learning to produce risk score- 50% of those with highest scores died by suicide within a year
    • Gap: Need data on real time unfolding of suicidal thoughts
      • Have 5 subtypes of this thinking
      • Can also use passive monitoring like GPS movement, call/text, heart rate, accelerometer data
    • Gap- Need scalable real time prevention/intervention
    • Recent study: looked at people on social media platform and used machine learning to identify people who might be high risk and give them a brief automated intervention
  • Questions-
    • Is machine learning good at identifying people at high risk? Yes that’s how they use it and then give it to clinicians
    • How do you find signal amongst noise? Follow people post hospitalization and use EMA and then hope to build models and and then generalize

Talk #2: Non-Suicidal Self-Injury Among Sexual And Gender Minority Youth

[edit | edit source]

Richard Liu, Ph.D., Brown University School Of Medicine

Notes

[edit | edit source]
Click "Expand" for notes
  • Have long understood that these people are at high risk but substantial research is needed- designated as health disparity populations
  • Characterize scale, trends of this issue and disparities
    • Include physical and mental health conditions (NSSI included)
  • Systematic review/meta-analysis = 54 studies on NSSI in gender/sexual minorities
  • Lifetime prevalence across different minority groups (compared to cisgender reference groups)
    • 2-3x as much lifetime prevalence across different gender/ethnic minorities
    • Bisexual/transgender at particular risk
    • Moderator analyses
      • Age was one of the strongest
      • Sexual minority status stronger among adolescents than adults
  • Temporal trends in NSSI among sexual minority youth
    • Core sets of questions across states but individual states could ask their own questions too
    • Massachusetts Youth Risk Behavior Survey (YRBS) ahead of the curve in asking questions about gender identity, minority status
      • Also asking NSSI before others
      • Some people who engage in same-sex behaviors don’t identify as sexual minorities, and vice versa
    • NSSI range among sexual minorities 38-52%
    • Significant decline in heterosexual youth NSSI from 2005-2017
  • Gender differences?
    • No sex differences in these studies b/c sample was too small, but other researchers have found profound sex differences and we know they exist
  • Did they look at sexual/gender identity over time?
    • Current datasets cross-sectional so no, but in current studies they are

Talk #3: Emotional Reactivity And Suicidal Ideation: A Prospective Neuroimaging Study

[edit | edit source]

Adam Bryant Miller, Ph.D., University Of North Carolina At Chapel Hill

Notes

[edit | edit source]
Click "Expand" for notes
  • Investiagtes nueral mechanisms underlying suicide
  • Disruptions in emotion regulation
    • Linked with suicidal ideation and behaviors
    • Most adolescents have reaction to stressors- some might be over or under exaggerated
  • Emotion reactivity trials
    • Sample of 49 adolescents
    • Individuals with and without history of suicidal ideation
    • Looking at negative pictures in a scanner
    • Those with history of ideation may not engage regions we know to be implicated in emotion control
    • Also interested in prediction- found that less activation of dorsolateral prefrontal cortex was associated with greater future suicidal ideation severity
    • What mechanisms link childhood adversity with ideation? Used a different sample and found that lower activation of DLPFC was linked
  • Questions
    • What about other dependent variables like depression or anxiety? Mediation model was significant after controlling for depression. More interested in neural markers of specific risk factors
    • Are there treatments that promote increased activation?
      • Also have trials to tell them to use skills while watching the same negative/neutral images. Kids with suicidal ideation were able to bring activation in the areas of interest in the experimental setting
    • Executive functioning deficit may also contribute to not being able to do the CBT skills they found helpful

Future Directions And Discussion: Anthony Spirito, Ph.D., Brown University School Of Medicine

[edit | edit source]

Notes

[edit | edit source]
Click "Expand" for notes
  • Has been conducting research with suicide attempters for 35 years
  • Intersection of clinical care with clinical translational research
    • Goal is to improve patient care
    • Challenges to that & questions we should consider
  • Don’t get seduced by technique and technology
    • We are clinicians and need to step up sometimes
  • Quotes to think about:
    • Tom Insel said in 20 years at NIH, we didn’t move the needle in suicide
    • Think about how neurocircuitry underlies deaths of despair
      • Take into account the populations you’re working with (e.g., alcoholism and deaths of despair in white middle aged men)
  • Risk factors we know a lot about from rock and roll
    • Loneliness- 33% of millenials are lonely; how do we reduce loneliness, increase social connectedness to reduce suicide over next 13 years
      • Consider biological treatments like oxytocin all the way to social media intervention
    • Sleep- sleep always comes up as an indicator of SITBs
      • All highly suicidal patients should be treated by sleeping pills? Y/N
      • Develop behavioral intervention that will improve sleep that doesn’t rely on patient motivation
        • Takes motivation to change this
  • Implementation is a science too-- what can it do over next 13 yrs
    • Equal focus on training mental health professionals to treat substance use
    • Training of professionals to identify/treat high risk for psychosis
    • Training to treat trauma
    • Assessment (EMA especially): what can we do for impulsive individuals?
      • Focus on most at risk and how to identify them and will they use the technology
  • Determine for what subset of patients with SI continuous monitoring is helpful and for which it’s not
  • AI techniques and machine learning in EMRs could be helpful because could show novel markers for intervention in suicide reduction
    • EMR could also contribute directly to increase in SI because of the time it takes away from clinicians providing patient care, association w/clinician burnout
  • Treatments to think about in the next 13 years
    • Neuroscience vs psychotherapy
    • AI replacing therapists
    • Expert therapist may become highest calling!
    • Ketamine-- could be good until overprescribed, replicate opioid crisis
  • Psychiatry invented case conceptualization and personalized medicine but it has found diminished importance in training and care- push toward evidence based treatment
    • Used to think of the person not the regulatory process
    • RDoC could actually help with new case conceptualizations
    • Consider human condition
  • Discussion
    • Underscore importance of case conceptualization to figure out what this situation means for the person. We have diagnostic criteria for depression that 100 kids might meet but with 100 different stories
    • Environment and suicidal behavior. Predicting and preventing the behavior are two different things. We know that when we return people who have been hospitalized to the same environment they came from, we know that they are highly likely to attempt suicide again because we haven’t fixed anything about the environment
    • How do we address multiple targets/multiple risk factors and also use new approaches to see how these things interact with each other?

Break & Poster Viewing (Session 2) (2:45 PM)

[edit | edit source]

To view the posters on OSF, click here.

Click "Expand" for list of session 2 posters

1. Emotion-Related Impulsivity Relates to the p Factor of General Psychopathology

Jennifer G. Pearlstein, Charles S. Carver, Kiara R. Timpano, & Sheri L. Johnson

2. Exploring changes in rumination across hospitalization: A study of adolescent self-injurers

Olivia H. Pollak, Eugene D’Angelo, Matthew K. Nock, & Christine B. Cha

3. Control When it Counts: Executive Control Under Stress Predicts Depressive Symptoms

Meghan E. Quinn and Jutta Joormann

4. Health, crime, and social-welfare inequality cluster in a population segment: Analyses of 4 million citizens from 2 nations

Leah S. Richmond-Rakerd, Stephanie D’Souza, Signe Hald Andersen, Sean Hogan, Renate M. Houts, Richie Poulton, Sandhya Ramrakha, Avshalom Caspi, Barry J. Milne, & Terrie E. Moffitt

5. Mental health service utilization by race/ethnicity in a nationally representative sample from 2004 to 2017

Ana Sheehan, Rachel Walsh, Christina Sanzari, & Richard Liu

6. Brief Mobile Mindfulness Intervention for Ruminative Adolescents: (Very) Preliminary Results from a Randomized Control Trial

Caroline M. Swords, Eleanor Horner, Liesl Hostetter, Eliana Whitehouse, Jialun Yang, & Lori M. Hilt

7. Cerebral Blood Flow is Altered According to Mood States in Adolescents with Bipolar Disorder

Simina Toma, Bradley J. MacIntosh,; Anahit Grigorian, Lisa Fiksenbaum, Andrew D. Robertson, & Benjamin I. Goldstein

8. Experience-Sampling Approach to Emotion Differentiation Across Bipolar and Unipolar Mood Disorders: Associations with Emotion Regulation and Variability

Cynthia M. Villanueva, Douglas Mennin, Greg Murray, Renee J. Thompson, & June Gruber

9. Pregnant women’s self-reported emotion dysregulation predicts newborn neurobehavior

Robert D. Vlisides-Henry, Brendan D. Ostlund, Sarah Terrell, Mindy A. Brown, Elisabeth Conradt, & Sheila E. Crowell

10. Temporal Trends in Peer Victimization among Sexual Minority & Heterosexual Adolescents

Rachel Walsh, Christina Sanzari, Ana Sheehan, & Richard Liu

11. Reduced Hippocampal and Amygdala Volumes as Mechanisms of Stress Sensitization to Depression following Childhood Violence Exposure

David G. Weissman, Hilary K. Lambert, Alexandra M. Rodman, Matthew Peverill, Margaret A. Sheridan, & Katie A. McLaughlin

12. Borderline symptoms at age 12 signal risk for poor outcomes during the transition to adulthood: Findings from a genetically sensitive longitudinal cohort study

Jasmin Wertz, Avshalom Caspi, Antony Ambler, Louise Arseneault, Daniel W. Belsky, Andrea Danese, Helen L. Fisher, Timothy Matthews, Leah Richmond-Rakerd, & Terrie E. Moffitt

13. Examining the Relationship between Reaction to a Suicide Attempt and Subsequent Suicidal Behavior in the Context of a Suicide-Specific IOP.

K. Wolfe, K. Rial, K. Goga, C. Tran, A. Moorehead, B. Kennard, B., & G.J. Emslie

Data Blitz #3: Three Fast Talks, Three Slides Each, Followed By Discussion (3:00 PM)

[edit | edit source]

Talk #1: The Inflamed Brain: A Neuroimmune Model Of Depression

[edit | edit source]

Robin Nusslock, Ph.D., Northwestern University

Notes

[edit | edit source]
Click "Expand" for notes
  • We usually focus only on the brain; it would be a good idea to focus on other organs, too
    • Communication with immune system functions in emotional problems, too
  • Early life adversity sensitizes cells in the amygdala to be more responsive, heightened threat response
  • People at elevated early adversity also display increased inflammation
    • Primes immune cells; elevated inflammatory biomarkers
    • When dysregulated, elevated inflammation can lead to chronic mental and physical health problems, early risk of depression
  • Amygdala influences release of hormonal products, pro-inflammatory phenotype in the body
  • Brain in state of defense = body in state of defense
  • Inflammation shown to access amygdala, increased threat processes
    • Another target involves circuits in the brain involved in reward processing (ventral striatum, basal ganglia)
      • Dysphoria, anhedonia, psychomotor slowing; adaptive for when you’re fighting off something to help with conserving energy
  • Heightened peripheral inflammation - reduced executive control
  • Negative emotion -> self regulation (sometimes negative) -> all inflammatory
    • Way you’re regulating emotions increases inflammation which increases depression; perpetuating cycle
  • 2-hit vulnerability; can lead to novel neuroimmune interventions

Talk #2: Predicting Bipolar Disorder Onset In At-Risk Youth.

[edit | edit source]

Anna Van Meter, Ph.D., Northwell Health

Notes

[edit | edit source]
Click "Expand" for notes
  • Missed opportunity for early intervention
    • Most people report that they had symptoms prior to full onset but most reported symptoms aren’t specific to bipolar disorders
      • So most people who have these symptoms won’t develop bipolar, so who do we target?
    • Pittsburgh developed a risk calculator which was then applied to the longitudinal assessment of mania symptoms? Study
      • Modifications were made to apply it to LAMS and resulted in prediction a bit better than chance
      • LAMS youth recruited when seeking treatment while other sample had parents with bipolar- after removing these factors the AUC was improved somewhat
      • We expect risk of bipolar to increase in adolescence- so they got rid of age and was left with depressive and bipolar symptoms and had a better AUC- so the symptoms of bipolar predict bipolar best
    • Better quality data lead to more accurate predictions
      • We primarily rely on self report. It’s hard to predict rain coming unless it’s already sprinkling from an individual perspective

Talk #3: Where Are The Windows Of Opportunity For Suicide Prevention In BD?

[edit | edit source]

Ayal Schaffer, MD, FRCPC, Sunnybrook Health Sciences Centre

Notes

[edit | edit source]
Click "Expand" for notes
  • Dr. Schaffer is a psychiatrist in Toronto
  • Where should we go? Windows of opportunity for active intervention, prevention for patients with BD
  • Since 1998 have a collaboration with coroner in Toronto where they do suicide chart reviews
    • 4,200 suicides
    • Coroner’s data provides a lot of info about the person, detailed info about death, but doesn’t tell you much about what happened beforehand
    • Brought in healthcare administrative data (captures every bit of health data)
    • Combined these 2 databases to produce comprehensive picture of person, suicide, and contact with healthcare system in year prior to death
  • Open to collaborations w/people who want to look at coroner’s data
  • Toxicology, suicide note data, etc. Looked at visits to hospital, outpatient, etc.
    • 2,800; 176 with BD
  • Most of the people who died from suicide here did not touch acute care healthcare system, but were seen in community, outpatient practices or PCPs
    • Contrasts a lot of data we have that visits to ER are a strong predictor of suicide death
  • Time from last contact = 14 days (outpatient)
    • Greatest window of opportunity in ambulatory care; general psychiatry, looking at what’s happening in the person’s life; need to infuse evidence-based suicide prevention strategies into this care rather than waiting for them to come out of emergency care

Future Directions And Discussion: Sheri Johnson, Ph.D., University Of California, Berkeley

[edit | edit source]

Notes

[edit | edit source]
Click "Expand" for notes
  • 3 possible directions
    • Social environment
      • Trauma and early adversity (Kessler, McLaughlin et al., 2010)
      • Negative life events (Brown and Harris, 1989)
      • Income inequality (Wilkinson & Pickett, 2019)(Patel et al., 2018)
    • Broadened scope of psychological dimensions
      • Social dominance (Gilbert, 2000)(Sloman, 2000)(Johnson, Leedom, et al., 2012, Psychological Bulletin)
        • Childhood experiences of subordination
        • Life events of subordination and rejection
        • Same more than sadness
        • Lab indicators of subordination
        • Animal models of social defeat and depression
        • Testosterone deficiency
      • Emotion related impulsivity
        • Elevated in MDD
    • Integrity and Quality
      • Tackett papers on replicability
      • Incentive structures- for instance, tenure
        • Making sure we recognize someone’s contributions
        • We need to move from quantity to quality indices
    • Discussion
      • We don’t often publish our negative findings
      • Dominance and lithium? Dominance is defined as a mouse that will fight for shared resources. Hyper-dominant animals given lithium decreases dominance. Also literature about how being more dominant can be desirable.
        • Could also look at occupational issues and outcomes

End Of Conference, Brief Closing Remarks (4:00 PM)

[edit | edit source]

Mitchell J. Prinstein, Ph.D. & Eric A. Youngstrom, Ph.D.

  1. Ilgen, Mark A.; Downing, Karen; Zivin, Kara; Hoggatt, Katherine J.; Kim, H. Myra; Ganoczy, Dara; Austin, Karen L.; McCarthy, John F. et al. (2009-11-15). "Exploratory Data Mining Analysis Identifying Subgroups of Patients With Depression Who Are at High Risk for Suicide [CME"]. The Journal of Clinical Psychiatry 70 (11): 1495–1500. doi:10.4088/JCP.08m04795. ISSN 0160-6689. PMID 20031094. PMC PMC3057750. http://www.psychiatrist.com/JCP/article/Pages/2009/v70n11/v70n1102.aspx. 
  2. Schnack, Hugo G.; Nieuwenhuis, Mireille; van Haren, Neeltje E. M.; Abramovic, Lucija; Scheewe, Thomas W.; Brouwer, Rachel M.; Hulshoff Pol, Hilleke E.; Kahn, René S. (2014-01-01). "Can structural MRI aid in clinical classification? A machine learning study in two independent samples of patients with schizophrenia, bipolar disorder and healthy subjects". NeuroImage 84: 299–306. doi:10.1016/j.neuroimage.2013.08.053. ISSN 1053-8119. http://www.sciencedirect.com/science/article/pii/S1053811913009166. 
  3. Loo, Hanna M. van; Cai, Tianxi; Gruber, Michael J.; Li, Junlong; Jonge, Peter de; Petukhova, Maria; Rose, Sherri; Sampson, Nancy A. et al. (2014). "Major Depressive Disorder Subtypes to Predict Long-Term Course". Depression and Anxiety 31 (9): 765–777. doi:10.1002/da.22233. ISSN 1520-6394. PMID 24425049. PMC PMC5125445. https://onlinelibrary.wiley.com/doi/abs/10.1002/da.22233.