OpenIDEO is an open innovation platform. Join our global community to solve big challenges for social good. Sign Up / Login or Learn more

AI-Assisted Online Communities to Scalably Offer Anonymous Caregiver Peer Support

An online community for small groups of caregivers to anonymously relate, with AI tools to regulate groups and monitor caregiver wellness.

Photo of Shrenik Jain
4 1

Written by

Who is your idea designed for and how does it better support family caregivers as they care for a loved one with dementia?

Our work is designed to offer support to support today's caregivers who struggle in isolation. Cognizant of the stigma and physical accessibility issues that keep these individuals from reaching out, our app allows tailored small group support from anywhere, and anytime via text and voice. However, with access, comes the potential for abuse - so our AI tools run on messages to detect the underlying emotions - flagging inappropriate msgs, and detecting when a caregiver may need additional help

We initially built a mobile application for AI-assisted text group peer support to help those with mental illnesses and substance use issues. Now, however, we see a tremendous opportunity in customizing the technology we have already built to cost-effectively allow every caregiver to receive tailored peer support so they never have to feel alone.

It's known that staying connected is a core-factor in the well-being of caregivers. Stigma, cost-concerns, and limited physical availability mean many caregivers can never access face-to-face support. Online text-based groups have tremendous potential in overcoming these barriers, but existing forums have no way to regulate the sheer amount of conversation generated, which is why sites such as Reddit are plagued by intentionally malicious users, or 'trolls'.

Our app makes group support scalable through artificial intelligence (AI). We’ve built a mobile app for text-based group peer support that allows users to text into a tailored support of 5-7 peers under a pseudonym at any time, from anywhere. Groups are moderated by clinicians or peer supporters sourced from a partner organization. These moderators are aided in managing group content by our AI tools that run on group messages and extrapolate the underlying emotional sentiment, automatically detect clinically relevant or concerning content (e.g. sadness, maliciousness). This allows us to leverage the ability of peers to generate economical engagement while using technology to keep peer-peer interactions safe and allow moderators to focus their time on the caregivers who are most in need of help, exactly when they need it. We have applied our platform to a wide variety of groups (our AI tools learn the slang and jargon between different groups).

Our team has built the first version of this app and its AI tools with the help of clinicians from Johns Hopkins (linked in this submission is the demo of our app at a recent pilot). Our goal is now to test and refine our approach through the OpenIdeo platform so that we can offer caregivers the best possible experience. We at Sunrise Health are also grateful for the support of Johns Hopkins, the Abell Foundation, and the National Institutes of Health.

More info on our founders, Shrenik & Ravi.

What early, lightweight experiment might you try out in your own community to find out if the idea will meet your expectations?

1) Look at existing engagement and usage of online platforms and forums (e.g. reddit) as makeshift support groups within the caregiver communities. 2) Offer our platform to caregivers and place them in small groups, manually monitoring group conversations while we judge how caregivers react to the platform. 3) Analyze existing data from online caregiver support groups with our AI tools to demonstrate both our ability to flag concerning messages, and the need for a technology to do so.

What skills, input, or guidance from the OpenIDEO community would be most helpful in building out or refining your idea?

Our application has been used in a wide variety of populations and contexts: from first responders, to substance users, to at-risk high-schoolers. While this is a testament to the robustness of our small group approach, we recognize that dementia caregivers have unique needs themselves. We'd welcome opportunities to collaborate in 1) refining the UI/UX of our app, 2) developing dementia-caregiver specific content to integrate with the platform, and 3) discussing different rollout strategies

How long has your idea existed?

  • Over 1 year

This idea emerged from

  • A group brainstorm
  • A student collaboration

Tell us about your work experience:

Shrenik and Ravi realized this idea initially to help those with mental illness. Shrenik had watched fellow firefighters deal with PTSD alone, and Ravi had a close friend with depression. We've worked for and exited multiple tech & healthcare startups, and studied this space together at Hopkins.

This inspired (1)

People are for people


Join the conversation:

Photo of Kate Rushton

Hi Shrenik,

It is great to have you in the challenge. This is really interesting.

How are you planning to measure success of this solution and impact on caregivers? How would the finances of the app work? Would the clinicians be paid and would there be a cost to the user?

Photo of Shrenik Jain

Thanks for your response Kate. You've definitely brought up some important considerations. Apologies for typos, as I am on mobile :)

Measuring impact on caregivers - first and foremost, we can measure retention and engagement within the app - if people are consistently talking in groups and coming back, I think that's a sign that we are making progress in reducing the social isolation of caregivers. Another thing we can do is integrate polls and self-Check quizzes into the app. We initially built this feature to integrate diagnostic forms in a substance use context, but it works for any questionare. We can juxtapose an individual answers with their engagement and sentiment scores from our app. Which brings us to the final, and most innovative measure - the sentiment scores our algorithms give messages themselves, which allow us to get an idea of how a caregiver might be feeling based on the emotional sentiment (e.g. more positive or negative). While this is very early stage, I think it is one of the most promising parts of our technology, that we can automatically extract critical content from messages themselves. WIth access to more data, we will get a understanding of how caregivers as a whole respond to different topics - for example, our existing work has drawn interesting links between things like sleep or profanity and how a persons emotional sentiment related to their discussion of such topics.

On the financial aide we currently use a B2B SaaS approach where we license the platform to a central organization, which provides the group moderators. We customize and brand a version of the app for this central organization to better reach their population of interest. In this case there is no charge to end users. It's worth nothing that groups don't have to be moderated by clinicians - it depends entirely on the preference of the organization we work with. We have had groups moderated by peer supporters, counselors, social workers, and psychiatrists - since we provide a technology platform, we can customize to fit best with what's already on the ground. With regards to caregivers for dementia patients, I think healthcare systems and perhaps even payors would see value in this - healthy caregivers lead to better outcomes for both the caregiver and the cared for. Ancedotally, when I left John's Hopkins to work on this technology, the psychiatristry department was undergoing a big push specifically for caring for the caregiver type solutions, so there is clearly some level of interest.
Of course just because we used a certain model before, does not mean we have to use it now. I'd welcome feedback both on the technology and the sustainablity strategy. Just writing this I though that a model that charges caregivers a small amount could perhaps work in training and compensating a peer support to moderate groups. I do however, feel that our technology provides enough value to be sustainable without continuous grants or donations.
I hope this answered your questions! Let me know if I can provide any more context.

Photo of om G

Would you go so far as to provide the ‘Enterprise’ app for far reduced cost in order to gain access to all those suggestions, tips, and advice for your dissemination across the entire user base?

This is also an excellent way to help new caregivers obtain history for that individual.

Great idea.

Photo of Shrenik Jain

Thanks for your response, Om. Could you clarify a bit as to what you are suggesting?