That time when we created a therapist chatbot for family caregivers that offers emotional support and provides targeted reliable resources to help reduce burnout

Project Context


Over 43 million family caregivers in the U.S. are providing unpaid care to an adult or child. Support websites or mobile applications for family caregivers are highly fragmented and scarce, exacerbating the health impacts of caregiving.


A chatbot platform delivering an AI-driven mobile application providing on-demand, interactive, and emotionally intelligent support as well as health solutions for reducing caregiver burnout and promoting good health and well-being of family caregivers.


Family caregivers
Family Caregiver Program
Human resources department
Employee Wellness program vendors
Caregiver’s social support network
Managed Care Organizations
Medicare and Medicaid
Healthcare Providers

Project Type

Volunteer Graduate School Project


Volunteer Driven

Time Frame

1 Year

My Role

User Researcher
Project Manager
User Experience Designer


Interdisciplinary Professors
Interdisciplinary UW Students

Project Breakdown


Problem Context

There are 43 million caregivers in the U.S. with more than 850,000 family caregivers In Washington state alone. They are the backbone of many states’ long-term care systems.

For many Family caregivers their health is adversely affected while fulfilling the responsibilities of caring for a chronically ill family member. 70% of working caregivers suffer work-related difficulties due to their dual roles.

The pressures and stress inherent to caregiving duties place caregivers at risk of neglecting their health and well-being. In turn, poor mental health may harm the caregiver and those depending on them for care. However, family caregivers have low utilization rates for conventional support and work services averaging around 5%; usage is especially lower for those isolated family caregivers residing in underserved communities with limited access to healthcare.

As a result, Caregivers’ health has a direct impact on the cost to local, state, and federal budgets. Caregiver absenteeism costs the U.S. economy an estimated $25.2 billion in lost productivity (based on the average number of work days missed per working caregiver).

Figure 1: There’s profound research showing if caregivers don’t have good health and physical and mental functioning, they are not able to provide good care to the ones who are relying on them.


Research Process

Recognizing the severity of this healthcare challenge, Dr. Yuwen was eager to address caregivers’ unmet health needs in a way that would be accessible and remotely available. 

Utilizing the University of Washington’s network, she recruited a group of very talented faculty and students together.  The team was awarded a Population Health Initiative pilot research grant in March 2019 to conduct an initial pilot research study on the acceptability of this proposed solution. The results of this initial study confirmed the utility and feasibility of a specialized chatbot for caregivers.

The initial study further inspired Dr. Yuwen and her colleagues to develop CocoBot for real-world applications and build something that is for the social good. 



After identifying a mechanism that could address caregivers’ healthcare needs, Dr. Yuwen and an interdisciplinary team of faculty and students collaborated to create “CocoBot”, a chatbot that utilizes artificial intelligence to detect caregivers’ feelings and propose self-help solutions or direct caregivers to appropriate healthcare resources.

Interdisciplinary team

The UW provided support on both the research side and also with commercialization, which helped scale up and make CocoBot accessible for families.

The Cocobot project was a interdisciplinary collaboration between UW students, Magaly Ramirez, assistant professor of health services at the UW School of Public Health, community partners Puget Sound Asthma Coalition and Mujeres Latinas Apoyando La Comunidad, Dong Si, assistant professor in the Division of Computing & Software Systems at UW Bothell’s STEM; Sunny Chieh Cheng, assistant professor of Nursing and Healthcare Leadership at UW Tacoma; and Teresa Ward, professor and chair of the department of child, family, and population health nursing at the UW School of Nursing. 

Figure 2: Caregivers can connect with a Care Professional for advice or chat at their own pace with Cocobot, their mental health AI-assisted conversational chatbot.



With Cocobot, family caregivers can connect with care professionals for advice or chat with Cocobot on their own time, their mental health AI-assisted conversational chatbot. “Cocobot” combines cutting-edge conversational AI technologies, evidence-based therapies, and insights from top care professionals to support many caregiving needs.

I had varying responsibilities as a yearlong volunteer team member for the family caregiver startup. Specifically, I recruited caregivers (during peak covid times) for their paid participation in a survey and semi-structured remote Interviews. Additionally, I collaborated with designers, researchers, and healthcare workers using Figma and WordPress to develop the “” website. Lastly, I helped tag conversational data to support the Machine Learning process when requested.

Figure 3: I collaborated with designers, researchers, and healthcare workers to develop the website “” using Figma and WordPress.



During peak Covid times, our team participated in the remote National Science Foundation (NSF) I-Corps program. For customer discovery research, I recruited caregivers and conducted a paid semi-structured interview. Another teammate and I virtually presented the research findings to the NSF I-Corps judges, who awarded our team an NSF grant to continue our work.

As a direct result of my research and presentation contribution, our team earned a $2,500 grant, which further aided the customer discovery research. Subsequently, the NSF grant helped keep our team progressing forward as we later became the recipients of the University of Washington’s CoMotion Innovation Grant ($50,000).


So far, the team’s proposed solution has earned them various grants, from a Population Health Initiative pilot research grant, a National Science Foundation grant, and UW’s CoMotion Innovation Gap Fund.

As a result, the Coco team has developed an AutoML platform that allows experts to quickly train an automated system that can be used stand-alone or as an interface to provide recommendations to non-expert operators. Coco founders are currently seeking a patent for this crucial differentiator. The interdisciplinary team continues forward, furthering the development of CocoBot for real-world applications.

Check it out

For more information about the CocoBot application, visit

Future Iterations

Competing products fail to scale while maintaining the quality of their service with personal health coaches. Coco’s competitive advantage over competing products is from offering its product free of charge to caregivers by approaching it in a B2B2C model for rapid and sustained growth. For a pilot program, Dr. Yuwen and her team are combining forces with UW Human Resources to develop a personalized UW version of their application specifically with UW-specific resources for family caregivers.

Figure 4: Cocobot is a mental health AI-assisted conversational chatbot offering virtual therapy, telehealth, and self-care support through mindfulness exercises and wellness activities personalized for their needs.