Natural Language Processing Projects

Natural Language Processing and Machine Learning are computerized processes that can be used to code and make predictions from both structured (e.g. ICD-9/10 codes) and unstructured (e.g. free text) data. The HER Lab is evaluating the utility of these approaches to predict suicide in adolescents and first episode of psychosis. Improved identification of at risk individuals will reduce the prevalence of suicide while earlier detection of psychosis is associated with improved outcomes.
PI and Point Person at the HER Lab: Nick Carson

Team members: Ben Cook, Brian Mullin, Maria Jose Sanchez, Fred Lu

CMS Disparities (ACOs) by Race/Ethnicity, Gender, and Gender Identity Status

The Health Equity Research Lab was awarded one of the 5 Office of Minority Health seats with the CMS Virtual Research Data Center. This project will study the differences in ACO participation by race/ethnicity, gender (including gender minority status), and mental health status of beneficiaries. This project will also study disparities within ACO's by dual eligibility status, race/ethnicity and gender/gender identity in access, utilization, and quality of mental health services, and also compare access, utilization, and quality of disparities for beneficiaries treated within ACOs to those not treated in ACO's by race/ethnicity and gender. 

CRICO* – Improving Limited English Proficient Patient Safety through Patient Portal Accessibility

Language barriers a risk for poor clinical outcomes, medical errors, and liability across the spectrum of care. We have partnered with Dr. Robert Marlin, a strong advocate of vulnerable populations, to assess the effects of implementing a Spanish-language patient online portal that accurately sends messages to low English proficiency patients in their native language with the aim to improve communication with the care team and mitigate risks that result from miscommunication. The project has recently started, and we look forward to evaluate the initiative by defining patients’ satisfaction and their clinical outcomes.

PI: Robert Marlin, MD, PhD, MPH

Point Person at HER Lab: Fred Lu

Team members: Ben Cook, Jorge Rodriguez, Michael Flores, Brian Mullin, Zhishao (Ken) Luo, Meredith Brown, Fernando Gargano.

Gender Minority Health and Healthcare Disparities Using

Medicare Claims and Electronic Health Records

In 2016, the National Institutes of Health (NIH) formally designated gender minorities (i.e., transgender or gender non-binary persons) as a health disparity population, reflecting mounting evidence that this population has unique unmet health needs and that more research is needed to answer the epidemiological and health services questions which may help reduce disparities for this group. We seek to understand access and utilization of health services by identifying gender minority patients in Medicare claims and in electronic health records. Dr. Progovac has been granted a Harvard Catalyst Award as well as an R03 NIMHD to proceed with the studies. Key outcomes of interest include adequate treatment for depression, treatment for chronic conditions, suicide behavior, victimization from domestic violence, avoidable hospitalizations. We have developed strong collaborations with advocates, policymakers and clinicians who are working closely with and for this population to improve gender minority well-being and health.
PI: Ana Progovac          
Point Person at the HER Lab: Fred Lu     
Team members: Benjamin Cook, Brian Mullin, Maria Jose Sanchez, Fred Lu, Alex McDowell.

Other collaborators: Sari Reisner, Marshall Forstein, Cindy Telingator, Emilia Dunham, Masson Dunn, Joanna Gattusso, Natalie Bird, Pei Chen, Joshua Saffer, Laura Hatfield, Mark Schuster.

The Impact of a Comprehensive Electronic Patient Portal on the Health Service Use of Patients Obtaining Services at Four Hospitals in Madrid: An Interrupted Time-series Analysis

The objective of this study is to determine the extent to which an electronic patient portal was associated with improvements in health service use. Using a quasi-experimental interrupted time-series approach to analyzing administrative hospital data, we assessed health service use before (April 2012-March 2015) and after (April 2015-December 2016) the implementation of a comprehensive electronic patient portal in four hospitals from the Madrid Health System (Madrid, Spain). Our primary health service use outcomes consisted of number of outpatient visits, any hospital use, any 30-day all-cause readmission, and any emergency department (ED) use.
Point Person at the HER Lab: Ben Cook
Team members: Mike Flores, collaboration with colleagues (Spain)

AHRQ Medicaid Payment Policy and Access to Care for

Dual Eligible Individuals

Towards reducing health disparities in racial/ethnic minority populations through policy research, Dr. Cook was granted a R01 NIMHD for 5 years with the aim to assess the impact of recent Medicare policy changes on racial/ethnic disparities in mental health care. Through the implementation of natural experiments, the HER lab will assess the effects of the changes specifically on racial/ethnic minorities in medical care use, quality process measures, clinical events and total and component medical spending. As documented, racial minorities are less likely to receive mental health care than Whites, have lower income and are dual eligible. We hypothesize that after examining longitudinal changes for Hispanic, Black and Asians and Native Americans, these policies will benefit these populations the most. Data to be used will include a 20% sample of fee-for-service beneficiaries linked to provider and area-level data implementing difference-in-difference and within-person statistical analysis. The project is done in collaboration with Dr. Vicky Fung at Massachusetts General Hospital.
PI: Ben Cook, Vicky Fung
Point Person at the HER Lab: Ben Cook
Team members: Mike Flores, Rajan Sonik, Esther Lee, Brian Mullin

Cambridge Health Alliance Health Integration Program Evaluation

The Health Integration Program at the Central Street Health Center at CHA is a Behavioral Health Home that offers recovery-oriented care to adults with serious mental illness, including schizophrenia and other psychotic disorders. The project consists of a “reverse integration” program where primary care services are provided in a mental health care setting. It involves the participation of a multidisciplinary team including healthcare providers, nurse practitioners, case managers and administrative personnel. The HER lab focuses in the short and long-term evaluation of the implemented program, including barriers/facilitators, appropriateness, acceptability, feasibility, sustainability and replicability. A mixed methods approach from analysis of lab tests and narratives from semi-structured interviews with providers and patients, has identified that through enhanced training and care coordination, patients have received more screening tests and monitoring of co-morbid chronic and acute medical conditions. Dr. Progovac has earned a Kaplen award to continue with the analysis of the implemented project and define upcoming priorities as well as potential expansion to other CHA sites.
PI and Co-PI: Ana Progovac, Benjamin Le Cook, Miriam Tepper, Stephen Leff
Point Person at the HER Lab: Ana Progovac

Team members: Dharma Cortes, Sandy Cohen, Fred Lu, Brian Mullin

Patient Preferences Grant (PCORI)

PCORI stands for Patient Centered Outcomes Research and is an organization that helps people make informed healthcare decisions and improve healthcare delivery and outcomes through integrity and evidence-based information. The HER lab was awarded with a three year grant to develop an instrument to understand patients’ treatment preferences in their depression and diabetes care. Incorporating patient preferences into treatment plan requires an understanding of patients’ past experiences and preferences, including experiences of discrimination and the influence of peers, family and community members regarding treatment options. A mismatch between treatment and patient preferences worsens health outcomes via lower patient engagement, poorer adherence, and higher attrition. We have developed a survey that combines conjoint analysis with measurement of prior healthcare experiences and socio-cultural factors to describe the influence of preferences among race/ethnic minorities. The survey was sent nationally to more than 500 patients with either depression, diabetes or both. To refine the survey and understand treatment preferences and prior experiences, we conducted interviews with patients in person and by phone. For Aim 3, we expect to share these findings with providers and rise importance of including preferences as part of their chronic care treatment. The survey was developed using community-based participatory research in collaborations with the Primary Care Practice Improvement team and the Volunteer Health Advisors at Cambridge Health Alliance and the Transformation Center, a peer-support organization that seeks to transform policy and practice in communities through the participation of people with mental health diagnoses.

To learn more, visit the PCORI page

Collaborating partners: The Transformation Center
                                          The Cambridge Health Alliance Primary Care Practice Improvement 

PI: Ben Cook
Point Person at HER Lab: Esther Lee

Team members:  Adam Carle, Ana Progovac, Benjamin Cook, Catherine Rodriguez Quinerly, Danny McCormick, Deborah Delman, Dharma Cortes, Jonathan Delman, Susan Busch, Tali Fleitman Soffer, Timothy Creedon, Sherry Hou, Valeria Chambers, Ziva Mann, Diedre Jordan, Selma Castro, Ruth Nabisee, Farah Shaikh, Afsaneh Moradi, Maria Jose Sanchez, Esther Lee, Michael Flores, Rajan Sonik, Heba Abolaban, Natasha Kaushal

Health Equity Research Lab

1035 Cambridge Street, 2nd Floor

Cambridge, MA 02141

phone: 617-806-8741

fax: 617-806-8740

Current Evaluations and Major Projects

Comparative Effectiveness Research and Racial/Ethnic Health Care Equity (AHRQ R01)

We examine whether specific information in FDA warnings influenced disparity trends in psychotropic drug use and mental health care and to identify how provider characteristics and HMO enrollment act as mechanisms that underlie the differential diffusion of Comparative Effectiveness Research (CER) via health risk warnings. Identifying the influence of FDA risk warnings on trends in psychotropic drug use and related health care provides a platform to understand how CER will influence disparities and will help us to assess whether information regarding the risks and benefits of medications are being equitably disseminated. Our examination of how an increased reliance on CER will likely influence disparities in treatment after health care reform will provide policymakers with actionable information that might avert the negative equity consequences of incorporating CER into routine practice.

PI: Benjamin Cook           

Team members: Mike Flores, Ana Progovac, Nicholas Carson, Brian Mullin..

Follow-up after Acute Mental Health Treatment among Racial-Ethnic Minority Youth

The study assesses the quality of outpatient treatment episodes following emergency department and inpatient psychiatric treatment among Black, White, and Latino children and adolescents in the United States. We used data from multiple two-year panels from the Medical Expenditure Panel Survey to identify youth aged 5-17 between 1996 and 2015 with any emergency department or inpatient psychiatric treatment. We used logistic regression models to estimate any general outpatient follow-up or the beginning of adequate outpatient follow-up (defined as at least 8 therapy visits with a mental health provider or at least four therapy visits and a medication visit) within 7 or 30 days post-discharge. Estimates were adjusted for clinical need variables but not for socioeconomic characteristics, consistent with the Institute of Medicine definition of health care disparities.
Point Person at the HER Lab: Mike Flores
Team members: Nick Carson, Maria Jose Sanchez

Neighborhood-level Predictors of Opioid-Related Mortality

This study compare the neighborhoods of individuals with an opioid-related mortality to the neighborhoods of individuals that passed due to other premature causes. We build upon prior mortality studies by including several less often used neighborhood-level variables, in addition to more commonly used neighborhood-level variables that we believe are associated with opioid-related death. Models additionally incorporate individual-level variables so that comparisons can be made as to the influence of these variables in relation to the neighborhood-level factors.
Point Person at the HER Lab: Mike Flores
Team members: Ben Cook, Aparna Nathan

Safety Net Collaborative Evaluation

The HER Lab is conducting an evaluation of Safety Net, a multiagency integrated model of preventive services for at-risk youth involving mental health providers, police officers, schools, and the department of youth and families. The mission of Safety Net is to foster positive youth development, promote mental health, support safe school and community environments, and limit youth involvement in the juvenile justice system through coordinated prevention, intervention, and diversion services for Cambridge youth and families. This includes collaborations between law enforcement and community organizations and the prevention, early intervention, and criminal justice diversion efforts of Youth Resource Officers.
PI: Jamie Barrett
Point Person at the HER lab: Esther Lee

Team members: Dharma Cortes, Ben Cook, Esther Lee

To learn more, visit the Cambridge Safety Net Page

Strategic Planning for Law Enforcement and Mental Health


Supported by a Department of Justice grant, the HER Lab is partnering with the Cambridge Police Department (CPD) to conduct an evaluation of their community policing efforts. In the last 10 years, the CPD has transformed its community policing philosophy to focus more on prevention, intervention and diversion. Because at least 70% of people in the criminal justice system are mentally ill, the city has been working to connect at risk individuals with mental health care prior to arrest or incarceration. An important part of this is initiative is the CPD Community Services Unit, which includes officers and sworn personnel who are dedicated to addressing the needs of youth, the elderly, the homeless, and individuals with mental health issues by facilitating information sharing and coordination of care between CPD and partner organizations. We will conduct an evaluation of the current state this program, formalize partnerships, and develop toolkits to promote information sharing between law enforcement agencies looking to expand community policing services in their communities.
PI: Jamie Barrett
Point Person at the HER lab: Esther Lee
Team members: Dharma Cortes, Ben Cook, Esther Lee, Fred Lu