Click on a link below to jump to the corresponding section.
[separate into 2 tracks of research, phenotyping in wild and Neuroimaging Correlates of Psychosis, or 1 track on all research, 1 outlining key concepts for our work and our ethos/approach to them?]
Deep (Digital) Phenotyping “In the Wild”
- The Big Problem
- Quantifying Behavior
- Neuroimaging Correlates of Psychosis (Genetic and Clinical Risk)
- Building & Validating Tools (Data Storage, Processing, and Visualization)
- Detecting Agitation Before It Happens [Datta]
- Multisense: Nonverbal Feature Extraction in Psychosis
- Bipolar Longitudinal Study: longitudinal fluctuations in mood and cognition in bipolar patients
- Valera: Using Mobile Technology to Improve Quality of Care
The Big Problem
Our health care system is inefficient at providing optimal care for behavioral & psychiatric symptoms. Lack of actionable data in the hands of clinicians and researchers is a significant driver of this problem.
Conceptually, deep phenotyping seeks to address this dearth of data. However, capturing meaningful behavioral change is a complex task, and current phenotyping approaches are labor-intensive, cross-sectional, and error prone. Training staff adequately can take months, and even then administering and scoring a battery can take 3-4 hours per participant to complete. The subjectivity implicit in scoring SCIDs and scales also leads to relatively poor inter-rater reliability, making symptom fluctuations challenging to measure over time.
These downsides to phenotyping the make longitudinal data less valuable to researchers and less likely to translate into improved care. The conceit is good; proper execution requires a reformed approach.
As the old joke goes, “a therapist knows precisely how her patient is feeling every Thursday at 3 o’clock.”
Patient self-report is a critical but insufficient measure for x, y, and z. Even under the best conditions objective self-assessment is functionally impossible, and is bound to leave out important contextual factors and latent influences.
To mitigate this inherent error Baker Lab is pioneering digital methods for self-report undergirded by objective data. Drawn from smartphone These
Neuroimaging Correlates of Psychosis
Investigating the neurobiological underpinnings of psychotic disorders is another core task of Baker Lab…
Building & Validating Tools
There is a profound need for leadership in the emerging area of technology-based tools in psychiatry. For our part, Baker Lab is working in conjunction with the Institute for Technology in Psychiatry to develop best practices and software solutions for automated data processing and analysis. We see scalable, easily managed data conventions as critical infrastructure for helping the field and our patients realize the huge potential benefits of deep phenotyping and other novel approaches.
Detecting Agitation Before It Happens [Datta]
Automatic Characterization of Patterns of Human Behavior seeks to determine whether individuals with different neuropsychiatric illnesses may deviate from healthy individuals in their movement patterns in ways that may reflect their illness severity, current treatments, and side effects.
Multisense: Nonverbal Feature Extraction in Psychosis
The Multisense Project, or Feasibility and Clinical Utility of MultiSense for Nonverbal Feature Extraction in Psychosis, seeks to develop and test a software application called MultiSense developed by our collaborators at Carnegie Mellon University (CMU). Validation of MultiSense technology will 1) make ongoing clinical research and care more feasible and effective, and 2) provide new chronic disease insights.
Bipolar Longitudinal Study: longitudinal fluctuations in mood and cognition in bipolar patients
More commonly referred to as the Bipolar Longitudinal Study (BLS), Circuit dynamics underlying longitudinal fluctuations in mood and cognition in bipolar patients intends to characterize the natural course of changes in mood and cognition, associated environmental variables, structural brain anatomy, and functional network architecture in individuals with severe bipolar disorder or related conditions.
Valera: Using Mobile Technology to Improve Quality of Care
Using mobile technology to improve quality of care in individuals with mental illness is a clinical trial executed in collaboration with EthoSmart Health LLC. It seeks to validate their mobile application Valera as a tool for remotely guiding and supporting patients on an ongoing basis. Under the present protocol we are testing Valera as a means bothto improve the doctor-patient connection and improve patient clinical outcomes for individuals who have recently been discharged from McLean’s inpatient Schizophrenia and Bipolar Disorder Program.
Glossary of Terms
- deep phenotyping –