Keynote Presentation

Moving from Intuition to Machine Learning: How Can Quantitative Methods Improve Clinical Decision-Making?

Reliability of clinical diagnoses is often low. There are many algorithms that could improve diagnostic accuracy, and statistical learning is becoming popular with technology companies as well as psychology research. Using pediatric bipolar disorder as a clinically challenging example, we evaluated a series of increasingly complex models ranging from simple screening to a supervised LASSO (least absolute shrinkage and selection operation) regression in a large (N = 550) academic clinic sample. We then externally validated models in a community clinic (N = 511) with the same candidate predictors and semi-structured interview diagnoses, providing high methodological consistency; the clinics also had substantially different demography and referral patterns. Models performed well according to internal validation metrics. Complex models degraded rapidly when externally validated. Naive Bayesian and logistic models concentrating on predictors identified in prior meta-analyses tied or bettered LASSO models when externally validated. Implementing these methods would improve clinical diagnostic performance. Statistical learning research should continue to invest in high-quality indicators and diagnoses to supervise model training.

Dr. Youngstrom recommends reading the following documents in preparation for his talk.
We Are Ready to ROC
EBA of Bipolar Simple to Statistical Learning

Date: Thursday, April 19
Time: 11:00 a.m. to 12:00 p.m.
Location: Jackie Gaughan Multicultural Center - Unity Room (212)

The keynote is free, open to the public and requires no registration.

---

Workshop

During the workshop, Dr. Youngstrom will utilize R, a free statistical software package available at: https://www.r-project.org/. In preparation for the workshop, please download and install the R program as well as the following three packages: Hmisc, pROC and ggplot2.

Instructions for installing the R base program and specific R packages are available at the R Project website. Additional instructions, examples and walkthroughs can be found online.

The workshops are open to the public but require registration. Space is limited.

Register Now

Workshop Part 1

Are you Ready to ROC? Overview of the concepts and uses of Receiver Operating Characteristic Analysis

Bayes’ theorem is three centuries old, but its time has finally arrived. Although Meehl, Dawes and others described how Bayesian methods would be helpful to be able to revise probability estimates of key clinical status (e.g., diagnosis, treatment responder, risk of drop out or self-harm) using assessments or risk factors, they have been slow to permeate psychological research, training or practice. Improvements in technology make it feasible to gather more data from clients, score it in real time and feed it into Bayesian algorithms. These methods have transformed weather forecasting, prediction of elections and sporting events, and now medicine. IBM Watson is using these tools to build dashboards to integrate information and guide care, and Evidence-Based Medicine has developed a range of low tech tools (e.g., probability nomograms) and supporting software. Evidence-based assessment in psychology (Youngstrom, 2013) is melding these methods with traditional strengths of psychological assessment. This workshop will walk through steps from start to finish twice: from a research design and analysis perspective, and from finding a study to applying the results to an individual case.

Date: Thursday, April 19
Time: 3:00 - 5:00 p.m.
Location: Jackie Gaughan Multicultural Center - Unity Room (212)

Workshop Part 2

We’ll Be ROCing Together: Picking Samples and Playing with Techniques

This workshop will be a “hands-on” seminar. Participants can vet their own data to see if it is suitable for these analyses, and we will work through annotated examples. The research-oriented segment will cover optimal research design, checking assumptions, applying receiver operating characteristic analyses (ROC), considerations in choosing cut-points, and using logistic regression to adjust for covariates and test for moderators (Youngstrom, 2014). The clinical application segment will focus on finding studies, triaging them quickly for clinical validity and relevance, and then applying results to cases (including using free online tools to convert traditional effect sizes into ones more helpful in clinical decision-making).

Date: Friday, April 20
Time: 10:00 a.m. - 12:00 p.m.
Location: Jackie Gaughan Multicultural Center - Unity Room (212)