Pain Intervention & Digital Research Program: Precision tools to measure behavior and improve treatment

Melanie Fu, BS

Brigham and Women’s Hospital – Research Assistant
Fu_Melanie_Poster

Scientific Abstract

BACKGROUND

Chronic pain affects up to 28% of U.S. adults, costing ~$560 billion each year. Depression and anxiety are markedly frequent among people in chronic pain: a third of patients have either depression or anxiety and another third have both. The interaction between mental health and pain remains a barrier to improving individual patient function. We introduce the Pain Intervention and Digital Research (Pain-IDR) program—a research clinic that seeks to develop tools to measure behavior in patients with chronic pain and psychiatric conditions with a focus on the mental status and functional status exam.

METHODS

As reported in previous work, we used open-ended questions to elicit natural, nonstructured spoken responses from admission to discharge in 48 sessions from 8 acutely ill psychiatric inpatients.  We used inexpensive, consumer-grade equipment and publicly available, automated tools to derive face, acoustic, and linguistic measures. We tested whether derived measures infer clinically relevant phenotypic information about mental state within sample and in an independently collected, external validation set (142 sessions in 81 patients).

RESULTS

Face psychomotor activity (mean head pose velocity and mean gaze velocity along the y-axis) tracks clinician-rated depression score. Overall, faster head and gaze movement was observed as depressive mood score decreases among participants. We further showed that speech content analysis quantifies diminution clinical observations such as perseveration.

CONCLUSION

The Pain-IDR seeks to develop tools to better assess and treat patients with chronic pain and psychiatric conditions. We showed that consumer-grade devices and open-source software can detect clinically relevant features of mental state. Current work will use the Beiwe app (beiwe.org) to capture functional status via measures of mobility, emotion, and sociability. We are currently enrolling patients (IRB 2021P003530) and invite clinicians to refer patients for a comprehensive standard-of-care assessment and treatment plan. Please contact us (mfu5@mgh.harvard.edu or dbarron2@bwh.harvard.edu) for more information.

 

SoundCloud Transcript

Hello, my name is Melanie Fu. I am the lead research assistant for the Pain Intervention and Digital Research Program. I’ll be walking you through our poster:

Precision tools to measure behavior and improve treatment

By myself, Melanie Y. Fu and Dr. Daniel S. Barron, who directs the program.

 

BACKGROUND

Chronic pain affects up to 28% of U.S. adults, costing ~$560 billion each year. Depression and anxiety are markedly frequent among people who suffer from chronic pain: a third of patients have either depression or anxiety and another third have both depression and anxiety. Understanding the interaction between mental health and pain remains a barrier to improving individual patient function. Our overarching goal is to develop tools to quantitatively measure what we already ask about during a routine clinical assessment.

We introduce the Pain Intervention and Digital Research (Pain-IDR) program—a research clinic that seeks to develop tools to measure behavior in patients with chronic pain and psychiatric conditions, with a focus on the mental status and functional status exams.

METHODS

As reported in previous work, we used open-ended questions to elicit natural, nonstructured spoken responses from admission to discharge in 48 sessions from 8 acutely ill psychiatric inpatients.  We used inexpensive, consumer-grade equipment and publicly available, automated tools to derive face, acoustic, and linguistic measures. We tested whether derived measures infer clinically relevant phenotypic information about mental state within sample and in an independently collected, external validation set (142 sessions in 81 patients).

RESULTS

The figure at the top of the poster illustrates how transcribing speech and using natural language processing tools to measure semantic structure can be valuable. In a patient with mania and a grandiose delusion of being a world-class consultant, we used semantic analysis to quantify the distance from the patient’s speech to concept of “consulting” in relation to other patients. As the patient’s perseveration decreased, this topic became ls frequent in his interactions. Thus, we showed that speech content analysis quantifies clinical observations such as perseveration.

The figure at the bottom of the poster illustrates how quantifying face expression and movement can be valuable. By using automated algorithms to label different parts of the face, we can quantify how the face moves over time to create expression. Here, we showed that face psychomotor activity (mean head pose velocity and mean gaze velocity along the y-axis) tracks clinician-rated depression score. Overall, faster head and gaze movement was observed as depressive mood score decreases among participants (except patient 8, who had a personality disorder).  Individual patients are assigned their own color and are represented in Patient-Session format (e.g., 1-3 represents Patient 1, Session 3). The background density plot (blue hues) represents patient pose acceleration and velocity from the independently collected Northwell dataset (n=89). 

CONCLUSION

The Pain-IDR seeks to develop tools to better assess and treat patients with chronic pain and psychiatric conditions. We showed that consumer-grade devices and open-source software can detect clinically relevant features of mental state. Current work will use the Beiwe app (beiwe.org) to capture functional status via measures of mobility, emotion, and sociability.

We are currently enrolling patients (IRB 2021P003530) and invite clinicians to refer patients for a comprehensive standard-of-care assessment and treatment plan.

We are looking for patients who:

1) have a chronic pain and psychiatric condition

2) are 18-120 years old;

3) own a smartphone;

4) are fluent in English;

5) have access to Wi-Fi internet and a valid email address;

6) do not have an active substance use disorder

Patients referred to the Pain-IDR can expect:

1) A rigorous pain history and physical exam;

2) Multi-modal treatment plan recommendations;

3) Follow-up on treatment plan until their pain condition is stabilized;

4) To return to their referring provider once they are clinically stable.

Please contact either myself mfu5@mgh.harvard.edu or Dr. Barron dbarron2@bwh.harvard.edu for more information. Thank you!

Live Zoom Session – March 9th

research Areas

Authors

Melanie Y. Fu, BS, Daniel S. Barron, MD, PhD

Principal Investigator

Daniel S. Barron, MD, PhD