Hybrid clinical-AI solution for public mental health crisis

We are solving the looming public mental health & substance crises - both at the structural level as well as the end user level. 

Structurally, there is understaffing, insufficient or poor methods of psych education, culturally appropriate training,systematic methods to screen at community or population level. Also, currently only episodic clinical data is collected without lived experience data to get insights to risk-stratify and perform proactive referrals into acute care and/or primary care. More daunting is prospect for hiring fast enough to keep up, even as current professionals are experiencing burnout and leaving. Prevention focus to reduce burden on understaffed acute care system is vital but lacking.

At the user level, patient centered approaches with design thinking and appropriate theory of change models to effect therapy-seeking or preventive self-care is lacking. Activating the withdrawn and lonely populations don't exist. Prevention focus knowing that general levels of acuity, substance abuse are better tackled earlier at first symptoms are understood but not implemented, especially for different sub-groups that need different approaches to activate change. 

We deploy knowledge base with clinicians in the loop to use digital, user centered apps to capture extensive lived experience text data and deploy AI/ ML models to risk-stratify and proactively flag escalations and referrals. These data are complementary to in-clinic episodic data and bridges care systems and providers with different user cohorts with specific, actionable insights on mental health / substance abuse challenges and history. 

Related statistics and our thinking are listed below

A 2020 study found that 65% of nonmetropolitan counties did not have access to psychiatrists. [A call to action to address rural mental health disparities, 2020. Morales et al. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7681156/] The Jeeva team recognizes that children and young adults, particularly those with minoritized identities, are severely impacted by the lack of mental healthcare access. Through the pandemic, there has been a loss of third spaces—a community space that is not school or work where one does not need to spend money to participate—and that third space for young adults has shifted online. The rise in isolation, adverse childhood experiences, and lack of community mental healthcare has potential to set up a volatile adulthood for Gen Z. Nearly 80% of young people (~36M in US), nearly a fifth of US population in general expreience issues; with LGBTQ+ and other communities experiencing disproportionately higher. 

In rural areas, in-home assessments are commonly used for mental health screenings. This method creates a high barrier to mental health access. City density does not fully resolve this barrier either—we recognize that mental healthcare is still highly stigmatized and our goal is to lower the barrier at a young age to lower the burden on the understaffed healthcare system, should severe mental illness develop later in life. 

Our approach is to start with the younger, text-friendly and digital savvy population before deploying these to the other groups so we can take the core data and algorithms, refine and prove the models of bridging infrastructure to the user before expanding. 

Our solution has two basic parts. As noted above, we are starting with Gen Z generation as a beachhead to prove out model's capabilities to comprehensively engage users, address their issues for efficacy so we can address the problem statement discussed above. 

For the first part at the user level, it is a user centered mobile app with multiple modalities including (i) AI chatbot for 1x1 confidential 24 hour check-ins and CBT skills learning; (ii) AI enabled community forum for anonymous text based asynchronous peer validation, listening and social support around mental health challenges, with both clinician and AI moderation and (iii) tool kits and resliliency plans built into "challenges" that are suggested based on issues, emotional challenges uncovered through screening (GAD & PHQ screens are built-in) and/or through chat. These were developed through 140 user interviews on 3 campuses (part of MIT's i-corps program) and are currently being piloted with 60 students on three large public university campuses in the US. The mobile app is available for your demo via either playstore on iOS or Android under jeeva digital health. 

The second part is provider ie.,therapist or practice centric and currently being piloted thorugh a private behavioral health practice in licensed in TX and in CA. These are backend dashboard based systems integrated to the mobile apps where each therapist in the practice can assign users (patients) under their care specific weekly supplemental exercises and follow-through to be done between sessions including specific tools, techniques, readings, journaling, daily mood, anxiety and self-care readings. With these supplemental instructions and in conjunction with the app's community support, the hypothesis for the pilot is it will increase adherence, rate of efficacy and sustain the interventions as there will be increased peer motivation and accountability in completing and normalizing the interventions. Currently we have 14 users enrolled in this pilot with user consent. 

Both these pilots will end in early fall '2023. By running these pilots, we solve specifically the technology of digital screening for anxiety, depression, early addiction and mapping user level and structural data capture, deidentifying these so good data, user cohorts,and secure opt-in based referrals is possible.  By reducing stigma through AI chatbot as well as by "normalizing" pervasive mental health challenges via anonymous community & peer settings , we surface early symptoms and capture context and challenges in real-time. Our community forums have demonstrated that we are able to reach users that find mental health issues awkward and inaccessible - including minoritized members that identify themselves as "ACE" individuals or Asian males or Latina girls with adverse childhood experiences. The pilots are also helping us ensure we understand the undiagnosed (user pilots on campuses) vs. diagnosed (private practice) populations, can smooth out the referral, refine systems for content moderation, refine methods to increase adherance and levers for better efficacy outcomes.

Jeeva today in the pilot mode targets young adults, with a goal to use this as beachhead before expanding. Today, We are particularly focused on Gen Z, who are mostly in high school and college. Our goal is to have peer support on generation specific concerns, with the goal to reduce the experience of more severe mental illness in adulthood. Also wanting to prove our premise that early interventions will reduce burden on already under staffed, under resourced health systems, communities, schools, carefgivers etc. The impact will be significant as happy adjusted users will have higher academic and social success, get jobs or better jobs and understand self-regulation, self-care and sensible coping skills with the right education and well formed expecations on and need for therapy, medication, addiction etc. The impact on care-givers will be profound as roughly 50% of mental health suicides are due to those that didn't take up therapy or couldn't. Similarly, as a % of ER admits due to mental health, Gen Z is four timees more than other groups combined; so our impact can be measured in terms of ER admit reduction, the wait times reduction at providers. Today, it takes 10 to 12 years on average according to NAMI between first symptoms and treatment for mental health; we will reduce that. Also, marginalized communities tend to skip therapy or turn away due to stigma suffering the consequences as a result that we can address as well. 

We are very well rounded team with complementary strengths and experiences. Dr Young Jo is an adolescent psychiatrist and a faculty member at Univ of Florida where we are also pursuing few things.  Dr La Cena Jones is a family therapist; We have two AI engineers, two technical advisors both with PhDs in computer science. We also have mobile and full stack engineers and designers. Some of our team members are part-time consultants. I have a background in running an inhouse incubator at Verizon and running global product organization at SeaChange international where I launched a global SaaS product; I am active in Seattle and Boston startup ecosystems where I mentor at EforAll  for under represented founders and work with 360 Impact Studio ventures up in seattle. We hve excellent mentor who is a board member at NAMI, MA. Our team members are listed on our website https://jeevahealth.ai.  I have gone through I Corps program at MIT (in preparation for NSF submission), through CUNY public health incubator, the NEMIC med tech incubator, the Village Capital social impact accelerator. 

So, as a diverse team, we are intentional in building diversity into our teams and building for diversity and inclusivity in our solution; With intersecting identities in our team members that have all provided input in an attempt to reduce inequity in mental health access. As racial minority groups are least likely to seek mental healthcare, we ensure that our team reflects this diversity. Along with racial equity, we represent queer, disabled, low socioeconomic status, first generation students, immigrants, brown, black and caucasian, men and women and queer. 
The team lead is an immigrant south asian (India) and has personal experience with mental health and as care giver for mental health of queer, disabled family member. As represented on our website, our team leads all represent various races and ethnicities across the entire spectrum and we actively source panelists from the user communities we serve and with the challenges that we try to tackle. In addition, we recruit and have a rotating panel of Gen Z students representing diverse lived experiences with mental health, substance including minoritized identities that evaluate our solution design, user experience, chatbot's conversational design, adapting the language for the audience and basically helping with the overall product experience in the context of audience using it day to day. We intend to keep replicating this as we expand to other markets; we already have six therapists looking at the platform and are in conversation with three universities and their student relations staff - similar to Dean of Student Living - to help us frame solutions for the stakeholders. We expect to do the same as we expand beyond this to behavioral health or community health markets. 

Todday we serve ~60 students on 2 university campuses + a smaller 14 users (under six therapists) pilot with private behavioral health practice. Eventually, after pilot it will be deployed to the target markets. Many of these students are from the marganiled communities including LGBTQ+, Asian (males as well) and hispanic. As we know, 22% of hispanic and black community students seriously considered suicide last year. We serve these communities already and are refining these for commercial launch eventually

We see this as an opportunity to connect with the right mentoring resources and opportunity to actively connect and explore partnering with commuity health systems, partnering for research for specific social inequities in mental health, being able to bridge academic research to actionable, real issues on the ground especially in mental health and substance abuse. We also are actively working on applying for NSF, NIH, NIMH grants as we push into public health more broadly and SOLVE will help set us up for this as well.

The following are the innovative approaches: 

AI, personalizaation at the user level

AI informed microcoaching to make teaching of skills stick and be organic

Dual component of both self-care (for those considering therapy or finding it inaccessible or expensive or stigmatizing)  and supplemental care (for those in therapy) 

Digital platform for providers for triaging, monitoring and specifying tools, interventions and following through

Ability to risk stratify and proactively effect referrals to practitioners

Ability to identigy cohort grouops in population around key issues / emotion and pair up with peer support 

Reduce by 10% the black, hispanic suicide rate among students from current 22% to 20% directly due to our early intervention. Similar proportional improvements for other races, marginalized communities - by Making partnerships with Trevor Project, Justice for Girls etc to deploy our AI engine to make similar impact with community organizations and churches

Expand the community health and public health infrastructure's ability to screen, treat and prevent mental health to a larger population (by 15% by 5 years), expand the services ((20% in 5 years), train para-professionals to expand workforce (by 20% in 5 years) and introduce diverse curricula (30% custom content by 5 years) 

Reduce average wait times on campus across the board by 20% and for low acuity cases, by 70% (from avg of 6 weeks to 2 weeks or less) through partnerships with college campuses

Train 3000 social workers, school cunselors and deliver psych education to 30,000 middle school students, 30,000 high school students

Train 3000 community health workers to expand the workforce

Help influence student success at high school, colleges (completion rate, have a great experience, build meaninful foundations for lifelong learning)

We track KPIs on key measures of current "gaps" and pain points. THese range from access wait times, expense, the number of people affected by suicidal thoughts, sexual assaults on campus, the substance abuse case, the years to get mental health diagnoses and treatment, the adherence rates for these solutions, the average number of students that utilize campus health, the average rate of availibility of practitioners available by county in the US, the % of marganalized population left untreated for mental health, the wait times for these communities vs. wait times for general population, the cost barriers. etc. 

Our theory of change relies on specific applications of general models within our context. 

At the user level, patient centered approaches with design thinking and appropriate theory of change models to effect therapy-seeking or preventive self-care is currently limited or lacking and we concentrate on these. Activating the withdrawn and lonely populations don't exist; these are also similar for marganalized populations especially when society tends to become abusive. We apply focus on getting these groups identified, normalizing these topics and then delivering education in a targeted but unstigmatazing ways to get basic engagement; more anonymous engagement helps these groups see peer behavior towards improvement. We engage in group motivations but in anonymous mode as there are specific challeges here. 

Finally, Prevention focus is a fundamental and structural change. Knowing that general levels of acuity, substance abuse are better tackled earlier at first symptoms are understood but not implemented, especially for different sub-groups that need different approaches to activate change; we are piloting these with beachhead populations to demonstrate emperical data on change and design principles that will help

Core technologies included are mobile apps, AI, ML, LLMs, DialogFlow (google GCP) and we are looking at graph knowledge bases. We use clinically developed (proprietary) interventional content to supplement LLM and other AI technologies so the end results are accurate. We use privacy technologies for data de-identification and protection as user empowerment in our context is vital but we can only accomplish this at scale if we protect privacy.