Faculty Bibliography

BACKGROUND:Previous studies have suggested that social media data, along with machine learning algorithms, can be used to generate computational mental health insights. These computational insights have the potential to support clinician-patient communication during psychotherapy consultations. However, how clinicians perceive and envision using computational insights during consultations has been underexplored. OBJECTIVE:The aim of this study is to understand clinician perspectives regarding computational mental health insights from patients' social media activities. We focus on the opportunities and challenges of using these insights during psychotherapy consultations. METHODS:We developed a prototype that can analyze consented patients' Facebook data and visually represent these computational insights. We incorporated the insights into existing clinician-facing assessment tools, the Hamilton Depression Rating Scale and Global Functioning: Social Scale. The design intent is that a clinician will verbally interview a patient (eg, How was your mood in the past week?) while they reviewed relevant insights from the patient's social media activities (eg, number of depression-indicative posts). Using the prototype, we conducted interviews (n=15) and 3 focus groups (n=13) with mental health clinicians: psychiatrists, clinical psychologists, and licensed clinical social workers. The transcribed qualitative data were analyzed using thematic analysis. RESULTS:Clinicians reported that the prototype can support clinician-patient collaboration in agenda-setting, communicating symptoms, and navigating patients' verbal reports. They suggested potential use scenarios, such as reviewing the prototype before consultations and using the prototype when patients missed their consultations. They also speculated potential negative consequences: patients may feel like they are being monitored, which may yield negative effects, and the use of the prototype may increase the workload of clinicians, which is already difficult to manage. Finally, our participants expressed concerns regarding the prototype: they were unsure whether patients' social media accounts represented their actual behaviors; they wanted to learn how and when the machine learning algorithm can fail to meet their expectations of trust; and they were worried about situations where they could not properly respond to the insights, especially emergency situations outside of clinical settings. CONCLUSIONS:Our findings support the touted potential of computational mental health insights from patients' social media account data, especially in the context of psychotherapy consultations. However, sociotechnical issues, such as transparent algorithmic information and institutional support, should be addressed in future endeavors to design implementable and sustainable technology.

BACKGROUND:Adults with bipolar disorder (BD) often experience neurocognitive impairment that negatively impacts functioning and quality of life. Previous trials have found that dopamine agonist agents improve cognition in healthy volunteers and that adults with BD who have stable mood and mild cognitive deficits may also benefit. We hypothesized that pramipexole, a dopamine agonist, would improve neurocognitive function in patients with BD. METHODS:We recruited 60 adults (aged 18-65 years) with a diagnosis of BD I or II for an 8-week, double-blind, placebo-controlled trial (NCT02397837). All had stable mood and clinically significant neurocognitive impairment at baseline. Participants were randomized to receive pramipexole (n = 31) or a placebo (n = 29), dose was initiated at 0.125 mg 2 times a day and increased to a target of 4.5 mg/d. RESULTS:At trial end, the primary outcome, MATRICS Consensus Cognitive Battery composite score, had not improved more in the pramipexole group (mean [SD] = 1.15 [5.4]) than in the placebo group (mean [SD] = 4.12 [5.2], Cohen's d = 0.56, P = 0.049), and mixed models, controlling for symptoms, showed no association between treatment group and MATRICS Consensus Cognitive Battery scores. No serious adverse events were reported. CONCLUSIONS:These results suggest that pramipexole is not an efficacious cognitive enhancement agent in BD, even in a sample enriched for characteristics that were associated with a beneficial response in prior work. There are distinct cognitive subgroups among adults with BD and may be related differences in neurobiology that affect response to pramipexole. Additional research to better understand the onset and nature of the cognitive deficits in people with BD will be an important step toward a more personalized approach to treatment.

BACKGROUND:Bipolar disorder (BD) is associated with emotion interpretation biases that can exacerbate depressed mood. Interpretation bias training (IBT) may help; according to the "virtuous cycle" hypothesis, interpreting others' emotions as positive can lead to interactions that improve mood. Our goals were to determine whether IBT can shift emotion interpretation biases and demonstrate clinical benefits (lower depressed mood, improved social function) in people with BD. METHOD:Young adults with BD were recruited for three sessions of computer-based IBT. Active IBT targets negative emotion bias by training judgments of ambiguous face emotions towards happy judgments. Participants were randomized to active or sham IBT. Participants reported on mood and functioning at baseline, intervention end (week two), and week 10. RESULTS:Fifty participants (average age 22, 72% female) enrolled, 38 completed the week 10 follow-up. IBT shifted emotion interpretations (Hedges g = 1.63). There was a group-by-time effect (B = -13.88, p < .0001) on self-reported depression; the IBT group had a larger decrease in depressed mood. The IBT group also had a larger increase in perceived familial support (B = 3.88, p < .0001). Baseline learning rate (i.e., how quickly emotion judgments were updated) was associated with reduced clinician- (B = -54.70, p < 0.001) and self-reported depression (B = -58.20, p = 0.009). CONCLUSION:Our results converge with prior work demonstrating that IBT may reduce depressed mood. Additionally, our results provide support for role of operant conditioning in the treatment of depression. People with BD spend more time depressed than manic; IBT, an easily disseminated intervention, could augment traditional forms of treatment without significant expense or side effects.

OBJECTIVE:Risk calculators (RC) to predict clinical outcomes are gaining interest. An RC to estimate risk of bipolar spectrum disorders (BPSD) could help reduce the duration of undiagnosed BPSD and improve outcomes. Our objective was to adapt an RC previously validated in the Pittsburgh Bipolar Offspring Study (BIOS) sample to achieve adequate predictive ability in both familial high-risk and clinical high-risk youths. METHOD/METHODS:Participants (aged 6-12 years at baseline) from the Longitudinal Assessment of Manic Symptoms (LAMS) study (N = 473) were evaluated semi-annually. Evaluations included a Kiddie Schedule for Affective Disorders (K-SADS) interview. After testing an RC that closely approximated the original, we made modifications to improve model prediction. Models were trained in the BIOS data, which included biennial K-SADS assessments, and tested in LAMS. The final model was then trained in LAMS participants, including family history of BPSD as a predictor, and tested in the familial high-risk sample. RESULTS:Over follow-up, 65 youths newly met criteria for BPSD. The original RC identified youths who developed BPSD only moderately well (area under the curve [AUC] = 0.67). Eliminating predictors other than the K-SADS screening items for mania and depression improved accuracy (AUC = 0.73) and generalizability. The model trained in LAMS, including family history as a predictor, performed well in the BIOS sample (AUC = 0.74). CONCLUSION/CONCLUSIONS:The clinical circumstances under which the assessment of symptoms occurs affects RC accuracy; focusing on symptoms related to the onset of BPSD improved generalizability. Validation of the RC under clinically realistic circumstances will be an important next step.

BACKGROUND:Sleep is vital to youth well-being and when it becomes disturbed - whether due to environmental or individual factors - mental and physical health suffer. Sleep problems can also be a symptom of underlying mental health disorders. Assessing different components of sleep, including quality and hygiene, can be useful both for identifying mental health problems and for measuring changes in well-being over time. However, there are dozens of sleep-related measures for youth and it can be difficult to determine which to select for a specific research or clinical purpose. The goal of this review was to identify sleep-related measures for clinical and/or research use in youth mental health settings, and to update the evidence base on this topic. METHOD:We generated a list of candidate measures based on other reviews and searched in PubMed and PsycINFO using the terms "sleep" AND (measure OR assessment OR questionnaire) AND (psychometric OR reliability OR validity). Search results were limited to studies about children and adolescents (aged 2-17) published in English. Additional criteria for inclusion were that there had to be at least three publications reporting on the measure psychometrics in community or mental health populations. Sleep measures meeting these criteria were evaluated using the criteria set by De Los Reyes and Langer (2018). RESULTS:Twenty-six measures, across four domains of sleep - insomnia, sleep hygiene, sleepiness, sleep quality - met inclusion criteria. Each measure had at least adequate clinical utility. No measure(s) emerged as superior across psychometric domains. CONCLUSION:Clinicians and researchers must evaluate sleep measures for each use case, as the intended purpose will dictate which measure is best. Future research is necessary to evaluate measure performance in transdiagnostic mental health populations, including youth with serious mental illness.

Prior research has identified associations between social media activity and psychiatric diagnoses; however, diagnoses are rarely clinically confirmed. Toward the goal of applying novel approaches to improve outcomes, research using real patient data is necessary. We collected 3,404,959 Facebook messages and 142,390 images across 223 participants (mean age = 23.7; 41.7% male) with schizophrenia spectrum disorders (SSD), mood disorders (MD), and healthy volunteers (HV). We analyzed features uploaded up to 18 months before the first hospitalization using machine learning and built classifiers that distinguished SSD and MD from HV, and SSD from MD. Classification achieved AUC of 0.77 (HV vs. MD), 0.76 (HV vs. SSD), and 0.72 (SSD vs. MD). SSD used more (P < 0.01) perception words (hear, see, feel) than MD or HV. SSD and MD used more (P < 0.01) swear words compared to HV. SSD were more likely to express negative emotions compared to HV (P < 0.01). MD used more words related to biological processes (blood/pain) compared to HV (P < 0.01). The height and width of photos posted by SSD and MD were smaller (P < 0.01) than HV. MD photos contained more blues and less yellows (P < 0.01). Closer to hospitalization, use of punctuation increased (SSD vs HV), use of negative emotion words increased (MD vs. HV), and use of swear words increased (P < 0.01) for SSD and MD compared to HV. Machine-learning algorithms are capable of differentiating SSD and MD using Facebook activity alone over a year in advance of hospitalization. Integrating Facebook data with clinical information could one day serve to inform clinical decision-making.

To develop short forms of parent-rated mania and depression scales, evaluating their reliability, content coverage, criterion validity, and diagnostic accuracy. Caregivers completed the Parent General Behavior Inventory about their youth 5-18 years of age seeking outpatient mental health services at either an academic medical clinic (n = 617) or urban community mental health center (n = 530), along with other rating scales. Families also completed a semistructured Kiddie Schedule for Affective Disorders and Schizophrenia interview, with the rating scales masked during diagnosis. Ten-item short forms and projections of their psychometrics (vs. the full-length 46-item Depression and 28-item Hypomanic/Biphasic scales) were built in the academic sample and then externally cross-validated in the community sample. The mania and two depression short forms maintained high reliability (αs > .87 across both samples); high correlations with the full-length scales (rs> .93); excellent convergent and discriminant validity with mood, behavior, and demographic criteria; and diagnostic accuracy undiminished compared to using the full-length scales. Present analyses developed and externally cross-validated 10-item short forms that maintain high reliability and content coverage and show strong criterion validity and diagnostic accuracy-even when used in an independent sample with markedly different demographics and referral patterns. The short forms appear useful in clinical applications, including screening and initial evaluation, as well as in research settings, where they offer an inexpensive quantitative score. Future work should further evaluate sensitivity to treatment effects. The short forms are available in more than a dozen translations.

BACKGROUND:Research suggests that people with bipolar disorder (BD), like individuals with autism spectrum disorders or schizophrenia (among other forms of psychopathology), often have social cognition deficits that negatively impact relationships and quality of life. Studies of social cognition largely focus on face emotion recognition. However, relying solely on faces is not ecologically valid - other cues are available outside of a lab environment. If the ability to correctly interpret other emotion cues is intact, people with face emotion recognition deficits could learn to rely on other cues in order to make inferences about peoples' emotional states. This study explored whether both facial emotion and emotional body language (EBL) recognition are impaired in people with BD. METHOD:We measured the performance of individuals with BD relative to community controls on a computer-based emotion recognition task that isolated participants' ability to interpret emotions in faces, bodies without faces, and in bodies with faces. RESULTS:Results indicated that the BD group was significantly less accurate on face emotion recognition (Cohen's d = -0.87, p = .023), and was more likely to misidentify neutral stimuli as sad (Cohen's d = -0.58, p = .030). Emotion identification accuracy was equivalent across groups when the body (not just face) was visible. CONCLUSION:People with BD experience deficits in face emotion recognition, and their emotional state may influence their interpretation of others' emotions. However, recognition of EBL seems largely intact in this population. Paying attention to EBL may help people with BD to compensate for face emotion processing deficits and improve social functioning.