Artificial Intelligence Powered Mental Health Chatbot

Psychological problems are steadily in Pakistan due to financial problems, generational trauma, and numerous socio-cultural pressures stemming from institutions such as patriarchy. Research demonstrates that around 30-50 million people may potentially be affected by mental illness. 40% of Pakistan’s population of 220 million people comprises of adolescents and youth (13-30 years). Underprivileged youth are disproportionately affected by mental illness with recorded rates of substance abuse and suicide attempts steadily increasing recently. A nationwide survey conducted in June 2020 focusing on underprivileged communities revealed more than 60% of respondents were affected by psychological problems. Women are particularly affected, with the highest rate of mental illness reported in 20–30-year-old married women.

Stigma and lack of awareness about mental health are the biggest barriers with research showing that it can take up to 7 years for a patient to seek mental health support after developing symptoms, increasing the likelihood of their condition worsening. Public expenditure on mental health care is less than 1% of the health budget with most existing services being provided by the commercially driven private sector. These services are often costly, inaccessible to youth living outside of major urban centers and focused more on tertiary care, being offered in chaotic hospital-based settings which makes receiving them a daunting prospect for distressed individuals. There is little to no focus on mental health promotion and primary prevention, an important aspect of which is early detection and management which can reduce the burden of mental illness. Due to the high demand of services and low supply, it often takes months for people, who can afford services, to get appointments with mental health service providers. All these factors prevent young people affected by mental illness from receiving the support they need, compromising their human right to universal healthcare, preventing them from achieving their full earning potential, and causing them to remain entrapped in the cycle of poverty.

To bridge this gap and provide a sustainable and scalable solution we have developed Pakistan’s first mental health chatbot (https://taskeen.org/chatbot/). The pilot was launched in June 2021 and the first iteration focused on screening people for mental health problems and connecting those screened positive with our free of cost tele-mental health services for support. The chatbot was marketed on social media with more than 10,000 people using it and 1000 users requesting mental health support to date (30th March 2022), 90% of which are young people between 10-30 years of age. We were also contacted by several educational institutions requesting us to roll it out for their students. The results of the pilots have demonstrated that chatbot based mental health literacy and screening can prove to be an effective way of increasing the agency of youth affected by mental health problems by allowing them to access mental health resources in a confidential and user friendly. The chatbot can serve as a first line mental health resource for those affected by mental health problems, since it will not be hampered by capacity limitations and can be accessed 24/7. In the increasingly digital world where messaging conversations predominate it can also be more convenient for tech savvy youth to build trust and destigmatize help seeking for mental health. We now aim to scale the chatbot by expanding it into a mobile app and WhatsApp-based format, incorporating local languages, integrating mental health literacy resources, and testing it in educational systems functioning in low-income communities. We also want to increase the capacity of our free of cost nationwide tele-mental service and introduce virtual youth peer support groups to provide mental health support to young people affected by mental health problems more effectively. We hope that if we are able to succeed in doing this in Pakistan, we can scale this solution to other low- and middle-income countries in Asia, Africa and South America. 

A recent survey of Pakistan’s digital landscape highlighted that Internet penetration is increasing by 17% annually with around 78 million people currently having access, over 80% of which are youth. The mental health chatbot will be freely available and accessible nationwide 24/7 for those who can access social and mass media. It will be flexible by being available through web, mobile app, and WhatsApp based mediums. By integrating it in educational institutions working in underprivileged regions, we can also reach youth who don’t have internet access/personal devices.

By increasing the capacity of our free of cost tele-mental health services and connecting users affected by psychological problems to it, we can enable underserved youth to access mental health support from their homes. Introduction of virtual peer support groups can facilitate connection between distressed youth and enable training of a cadre of peer leaders who can provide support to others in their communities.

Taskeen Health Initiative is a non-profit aiming to change attitudes and behaviours around mental health and wellbeing and is the pioneer of mental health promotion and mental illness prevention in Pakistan. Taskeen Health Initiative has a proven track record in the area of interest for the past 7 years. It has active programs working on mental health literacy and stigma reduction, through community-based interventions and social and mass media campaigns and providing mental health support through free of cost tele-mental health services with over 80% of their target audience being youth. The mental health chatbot pilot was launched by Taskeen in May 2021.

Taha Sabri, the team lead, is a 31-year-old doctor working in public health in Pakistan. He got his MBBS degree followed by an MPH in Global Mental Health from the University of Washington. Over the past 8 years he has helped design various mental health programs in more than 10 non-profit organizations in Pakistan. Through this he has gained experience in the primary, secondary, and tertiary levels of mental health care including mental health awareness and stigma reduction campaigns, a school mental health program, mental health integration into primary care, mental health helpline, disaster response mental health interventions, and psychiatric rehabilitation. Taha is a Fulbright Scholar, an Acumen Fellow, an Aspen Fellow, and serves as an Advisor on the Lancet Commission of Stigma and Discrimination in Mental Health.

In addition to his professional achievements, he has lived experience of mental illness since he was 21 years old. This enabled him to learn first-hand about the challenges that youth face in accessing mental health support in Pakistan. This also informed the philosophy of Taskeen in the importance of using the voices of lived experience in designing effective mental health programs. We firmly believe that no intervention can be entirely successful without the input of the end users involved. In line with this belief, we take inputs from people with lived experience for all our programs through focus groups and prioritize them as potential employees. 90% of our management team are youth (under the age of 35) with 75% of our team members having lived experience of mental illness.

While we have established the vision for the program and have the capacity to roll it out, we require assistance in the form of intellectual and financial resources to realize our vision. This is what we hope that MIT solve can help us with. The financial aspect of the prize, along with support in fundraising, can help relieve the pressure of funding and enable us to focus on program development and implementation. Support in developing an effective marketing strategy for the solution will also be helpful in distributing it more widely. Access to world renowned scholars, mentors and institutions can help us develop technical skills, particularly with regards to the artificial intelligence powered mental health chatbot that we intend to build along with help with human centered design. We also need support to evaluate the impact of the chatbot by working with experts to identify the key indicators we should be measuring. We would also be grateful if MIT solve can help us establish connections with successful digital mental health models working in other parts of the world so we can learn from them and adapt their teachings to our cultural settings. To achieve sustainability, while ensuring equity, we also need support in developing an effective business model that can help us scale the chatbot. 

Most mental health services in Pakistan are hospital-based treatment services in urban centers, suitable for those with severe mental illnesses. This is too costly, inaccessible, and often inappropriate for youth from underprivileged communities. There is little focus on increasing mental health literacy, addressing mental health stigma, developing accessible mental health services, or developing peer networks. The few interventions that focus on these aspects are resource intensive and geographically limited, limiting their ability to be sustained and scaled.

Our solution addresses some of these gaps since the chatbot and tele-mental health service is free of cost and customized for youth. By allowing users to avail mental health literacy, screening, support, and connection to peer support groups through mobile devices we can increase accessibility for youth outside major urban centers. This also allow us to sustain and scale our solution more easily without any geographical limitations.

Mental health chatbots such as Woebot have been shown to be effective for increasing mental health literacy, conducting preliminary screening, and delivering basic therapeutic interventions for those affected by mental health concerns. However, these are mostly developed in HIC; our goal is to utilize existing evidence to culturally customize this technology for underserved youth in LMIC like Pakistan.

The current mental health chatbot offers mental health screening and is web based, being more accessible to middle income youth. We now aim to make it accessible to low-income youth by developing app-based and WhatsApp based versions and translating it into local languages, increasing its inclusiveness. The app-based version will function offline and allow us to reach underserved youth, in areas lacking internet, through educational institutions. The WhatsApp version will increase access for those having internet access but low literacy since this version will have voice messaging options. In addition, introduction of educational resources will allow the chatbot to improve mental health literacy and decrease the stigma against help seeking.

By increasing the capacity of tele-mental health services, by hiring more psychologists, we can allow those screened positive for mental illness to receive mental health support at the touch of a button. Our appointment-based service is available Monday to Saturday 11 AM to 11 PM. By connecting youth facing similar problems with each other through virtual peer support groups, facilitated by a psychologist, we can allow them to learn from “experts by experience”.

By developing this innovation, we can radically transform the face of mental health for underserved youth because it would be free of cost, available 24/7, provide ease of access, would not be limited by capacity issues reducing waiting times, and remove logistical and financial barriers preventing underserved young people from accessing mental health resources and support.

Our objectives for this project are:

• To increase mental health literacy of target youth by:
  • Increasing awareness of the importance of psychological wellbeing.
  • Educating them about the causes, signs, symptoms and management of depression and anxiety.
  • Enabling them to adopt healthy coping skills to manage stress.
• To screen users for symptoms of anxiety and depression and connect those screened positive to tele-mental health services.
• To improve mental health outcomes of beneficiaries utilizing tele-mental health services and virtual peer support groups.

At the end of next year, we hope to see the following impact:

• 15000 beneficiaries using the mental health chatbot.
• 9000 (60%) beneficiaries demonstrating a meaningful increase in mental health literacy.
• 3000 (20%) beneficiaries screened positive for anxiety and depression.
• 1500 (50%) beneficiaries screened positive for symptoms of anxiety and depression utilizing tele-mental health services.
• 900 (60%) beneficiaries utilizing mental health services and virtual peer support groups demonstrating reduction in anxiety and depression symptom severity and improvement in wellbeing.

At the end of 5 years, we hope to see the following impact:

• 150000 beneficiaries using the mental health chatbot.
• 90000 (60%) beneficiaries demonstrating a meaningful increase in mental health literacy.
• 30000 (20%) beneficiaries screened positive for anxiety and depression.
• 15000 (50%) beneficiaries screened positive for symptoms of anxiety and depression utilizing tele-mental health services.
• 9000 (60%) beneficiaries utilizing mental health services and virtual peer support groups demonstrating reduction in anxiety and depression symptom severity and improvement in wellbeing.

These projections are based on previous experience we have had with community based mental health literacy interventions and include current benchmarks we use to measure the success of our tele-mental health services.

A very useful aspect of the mental health chatbot is that all the interactions conducted on it can be recorded giving us access to data that can be used to analyze its effectiveness. Some important metrics that we will track are:

• Number of users reached through marketing the chatbot through social media and educational institutions.
• Number of users using the chatbot.
• Changes in mental health literacy of users (ascertained through statistical analysis by conducting a pre and post-test examining knowledge, attitudes, and behaviours of users regarding mental health)
• Percentage of users finding the chatbot to be useful.
• Percentage of users likely to recommend the chatbot to others.
• Percentage of users taking the mental health screening test
• Percentage of users screening positive for mental health problems
• Percentage of users screened positive utilizing tele-mental health services
• Changes in mental health outcomes of users utilizing tele-mental health services (measured through quantitative and qualitative methods depending on severity of mental health status)

The priority outcomes will be evaluated on those beneficiaries accessing tele-mental health services. We will be using a mixed methods approach to measure a change in the outcomes with the specific outcomes measured depending on the mental health status of the beneficiaries. Mental health status will be divided into mild, moderate and severe categories based on an assessment of impairment in functionality.

We will measure increased wellbeing for mild, moderate, and severe patients by using the WHO-Five Wellbeing Index. We will be measuring decreased symptom severity for moderate to severe patients by using the PHQ-9 and GAD-7. An internally developed qualitative tele-phonic survey will also be conducted with all categories of beneficiaries to assess improvement in distress, satisfaction with service quality. The tools will be administered at the first, after every fourth or at the last consultation. Statistical analyses will be used to assess differences in pre and post test scores.

Our theory of change is as follows:

• Inputs:
  • Development of new iteration of mental health chatbot.
  • Expansion of tele-mental health services through hiring more staff.
  • Formation of partnerships with educational institutions.
• Activities:
  • Marketing of chatbot on social and mass media.
  • Marketing of chatbot in partner educational institutions.
  • Mental health awareness of users
  • Mental health screening of users
  • Referral to tele-mental health services of users screened positive for mental health problems.
• Outputs:
  • Number of individuals using the chatbot
  • Percentage of users demonstrating improved mental health literacy.
  • Percentage of users screened positive seeking mental health support.
  • Percentage of users utilizing mental health support demonstrating improved mental health outcomes
• Outcomes:
  • Increased mental health awareness and stigma reduction in young people.
  • Early detection of users at risk of mental illness.
  • Improved mental health outcomes of users seeking mental health support.
• Impact:
  • Good health and wellbeing

The core technology that powers our solution is artificial intelligence which will be trained to provide appropriate and instantaneous responses to user queries. We will be integrating the mental health chatbot into existing messaging services such as WhatsApp messenger and will also be developing a dedicated mobile application for it along with improving the existing web based version. 

At Taskeen, we believe that innovation happens when diverse perspectives are allowed space to work together. That is why we have established a multidisciplinary team of individuals ranging from different backgrounds including psychiatrists, psychologists, public health professionals, marketing professionals, business development experts, artists, community development experts, financial management experts, journalists, lawyers, and social activists. We are an equal opportunity employer and believe in equity over equality. This is why 70% of our staff members are female, 25% are male and 5% are non-binary/gender non-conforming. Our team speaks more than 6 languages and come from a variety of religio-cultural backgrounds representing the Pakistani society. To facilitate our staff we have robust policies to support them including a mental health policy, anti-harassment policy, period policy (for menstruators) etc.

At Taskeen, we believe that healthcare is a right, not a commodity and that is why we strive to provide develop services that can be useful for the most underserved populations. An important aspect of this is that our services are free of cost with even the cost of telephonic consultations being borne by our organization. This is to allow maximum flexibility to our users and remove any barriers that may get in the way of help seeking. The aim of the chatbot is to be freely available to our target youth and adolescents. However, in order to ensure equity during initial consultations we conduct financial eligibility assessments of our beneficiaries and refer affording beneficiaries to partners offering fee for service while we manage non-affording beneficiaries inhouse. 

Our innovation does not require a lot of funding for maintenance as the major costs are in the design and development stage. We aim to raise funding for scale up through institutional grants and public donations which will mostly include increasing the capacity of tele-mental health services by recruiting and training more mental health professionals. In addition, the wellbeing of the youth is also on the government’s agenda as evidenced by the “Successful Youth” initiative and they are increasingly looking for digital solutions for problems as evidenced by the “Digital Pakistan” initiative. Therefore, we can leverage this and get state funding for sustainability in the future. We also plan to develop a commercial model by incorporating the chatbot into our existing employee wellbeing program, delivered to corporates, and utilize funds raised to sustain the program for underprivileged youth.

Over the past 2.5 years of our legal existence, we have received more than $300,000 in funding which we have utilized for our mission. Since our services are meant for underserved communities, we cannot monetize them easily and our ultimate goal is to integrate them within the public sector. One of our major achievements in this regard is that the government has shown interest in adopting our mental health helpline to become the national mental health helpline funded by the government. This would help us ensure scalability and sustainability. We want to achieve a similar result with the mental health chatbot.