Research program

Precision Mental Health

Can genetic, clinical, and patient-reported information help identify an effective treatment sooner?

COMPASS and related studies examine genetic, clinical, and patient-reported information as possible predictors of mental health treatment response.

Study-design schematic: genetic data, health records, patient reports, and treatment context inform repeated follow-up of mental health treatment response. Models are evaluated in a separate sample when a suitable dataset is available.
Study design combining several kinds of data to predict treatment response.

Why treatment response is difficult to predict

Finding an effective treatment can take time, and response varies widely from one person to another. COMPASS asks whether combining genetic, clinical, and patient-reported data can help predict those differences. A useful model must also perform well in people and settings that were not part of its development.

How COMPASS studies treatment response

Combine several kinds of data

Genetic and clinical data, patient reports, and digital measures provide different information about treatment and recovery.

Define treatment response carefully

We define treatment response to match the clinical question instead of treating every kind of response as the same outcome.

Validate in a separate sample

When a separate dataset is available, we use it to test models built in the first sample.

Follow outcomes over time

Repeated follow-up measures show how symptoms and treatment response change over time.

These papers describe the COMPASS study and ways to collect patient-reported information alongside electronic health records (EHRs) and genetic data. They also describe associations between genetic liability to depression and diagnoses in clinical records.

Current work

Projects and collaborations

Precision mental health

Active

COMPASS

COMPASS studies how genetic, clinical, and patient-reported information relates to mental health treatment response. Its full name is Comprehensive Mobile Precision Approach for Scalable Solutions in Mental Health Treatment.

Amy Bohnert, Srijan Sen, and Lars Fritsche are the principal investigators. Lars leads the genetics and EHR data work, including polygenic scores, pharmacogenetics, and analyses of diagnoses and medication histories.

National Institute of Mental Health
U01 · U01MH136025 · 2024-present

Publications

Selected papers

All publications →
2026

Precision approaches for scalable digital and clinic-based interventions in mental health

Biol Psychiatry Cogn Neurosci Neuroimaging

Why it matters: Digital tools can widen access, but low engagement and modest average effects mean that scale alone is not enough. Written by the three COMPASS co-leads, the perspective asks whether mobile, genetic, and clinical data can help tailor both conventional and digital treatment.

2023

Epidemiologic Questionnaire (EPI-Q) - a scalable, app-based health survey linked to electronic health record and genotype data

Epidemiol Health

Why it matters: Clinical records include diagnoses, visits, and prescriptions but omit many aspects of a patient's experience. EPI-Q links standardized self-reports—including mood, depression, anxiety, stress, pain, and substance use—with genotype and EHR data. Its 10% response rate also means that participation bias must be considered.

2022

Polygenic Liability to Depression Is Associated With Multiple Medical Conditions in the Electronic Health Record: Phenome-wide Association Study of 46,782 Individuals

Biol Psychiatry

Why it matters: In 46,782 MGI participants of European ancestry, a depression polygenic score was associated with diagnoses well beyond depression, including after patients with recorded depression were removed. The result informs COMPASS study design, but broad population associations should not be treated as predictions of an individual's response to treatment.

Updates

News and talks