SNEHA Didi se Poocho (Didi means ‘elder sister’ and Poocho means ‘ask’ in Hindi)

India ranks 128th in terms of meeting the UN health-related Sustainable Development Goals (SDG) by 2030. Although we have made significant progress in reducing maternal mortality, challenges remain in achieving the SDGs target. The SDG target 3.1 aims to reduce the global maternal mortality ratio (MMR) to less than 70 per 100,000 live births by 2030, which as of 2016-18, was 113 per 1,00,000 live births.  One-third of all women of reproductive age in India have anaemia.  India tops the list of 10 nations contributing 60% of the world’s premature deliveries. This data suggests priority initiatives to ensure the best health outcomes for our women, mothers and children. 

Bhiwandi, though a part of the Mumbai Metropolitan Regional Development Authority is one of the most vulnerable cities in India with low transport connectivity, resources and infrastructure.  With a large population of over 800,000, staying in informal settlements and on the foothills, difficult to traverse and navigate. A significant proportion of women, who hail from these settings, are migrants, displaced, underserved, and coupled with low literacy levels (30% of women are illiterate).  They have inadequate access to personalized, relevant and comprehensive health information during the critical first 1000 days of life, from pregnancy to children two years of age. This contributes to increased morbidity and mortality. 

• Exacerbated by low literacy levels, and the prevalence of age-old myths and misconceptions, they face challenges in finding reliable information on essential health aspects necessitating a low-cost, easily accessible, credible and timely source of information.
• Due to less exposure, women are hesitant to ask questions related to their sexual and reproductive health.  Even if they access primary health care services, they face inhibitions due to the crowds,  presence of male doctors, and inability to express themselves clearly in an environment that is not conducive to privacy and confidentiality.
• Despite having access to technology, many women lack the necessary skills to utilize it effectively, particularly in the context of healthcare. This knowledge gap presents a barrier that limits their ability to fully leverage digital solutions for their health and well-being.
• Linkages between government frontline workers and the community seeking health services are weak, leading to communication barriers and difficulties in accessing care. This is due to factors such as communication barriers, lack of trust, or insufficient awareness about available health services.
• ASHAs – Accredited Social Health Activists, who are one of the government frontline workers, often lack comprehensive training and resources, leading to gaps in their knowledge and skills. This can hinder their ability to provide quality health services.

Improving health outcomes requires addressing these multifaceted challenges with a sustainable approach that involves relevant information and nudges for behaviour change that is easily accessible at low cost through the AI-enabled conversational bot.

Our solution is "SNEHA Didi se Poocho", a WhatsApp-based voice and text conversational bot that provides women access to health information based on their needs through Generative AI with LLMs.   This endeavour will seek to strengthen two groups in their empowering journey; a. Women (married, pregnant & lactating women and mothers of 0-2 year old children) will be empowered with knowledge, potentially leading to behaviour change and improved health outcomes b. ASHAs - government frontline workers, will be empowered with knowledge that will enable them to improve service to communities, thereby resulting in improved health outcomes.   It has been observed that mothers who have interacted with ASHAs are more likely to adopt healthy behaviours related to seeking healthcare.

a. Women: An internal survey informed us that 65% of our women are Android users while 87% of these use WhatsApp.  These women will be empowered to effectively engage with and use the bot to access information related to sexual, reproductive, maternal and child health and social schemes.

Human-in-the-loop Approach:  However, for the 35% of women who do not have access to Android phones, ASHAs will provide support through in-person interactions, offering door-to-door counselling and delivering health information and services.

 Module: This module will feature a two-way text and voice-based Gen-AI bot, designed to facilitate seamless engagement between users and the bot. It will allow users to ask questions, and receive responses.  This user-friendly approach aims to enhance user engagement and provide accessible, personalized health information endorsed by the Ministry of Family Health and Welfare of the Government of India.  This digitized platform will augment women’s capacities to bridge information gaps, minimize their hesitancy and allay fears in using information technology.

b. ASHAs: ASHAs will be trained in digital literacy and empowered to facilitate the learning and engagement of bot users. The training will be geared to enhance their capacity to build knowledge and skills.

Module: Chatbot-enabled training modules, curated in line with the government’s training modules will be used to enhance ASHAs' health-related activities, incorporating a knowledge base of life-saving preventive health behaviours.  Additionally, the module will include a scorecard to continuously assess the ASHAs' learning, retention, and recall capacities. This feedback mechanism will help in enhancing the modules over time, ensuring that ASHAs are equipped with the necessary knowledge and skills to effectively improve health outcomes in their communities.

This initiative will democratize health information and enhance behaviour change communication ultimately improving maternal and child health outcomes.

The SNEHA Didi se Poocho chatbot will help in:  

1. Making information available to women based on their needs:

• Making health information accessible 24/7: This will provide 24/7 access to information to mothers, especially needed for mothers of very young children, women with low literacy levels and those who may have limited access to healthcare facilities. 
• Nudges for action: The solution will provide reminders and notifications to women, to encourage them to avail of services and take health action
• Enable scale: Given the high mobile penetration, the solution could potentially reach a large number of people. 
• Reaching the Migrant populations: this solution has the potential to reach women who migrate to other areas with messaging and information related to health.  

2. Providing a platform for ASHA workers to build agency and knowledge

 A solution to build knowledge and agency of ASHA’s using the Standard Operating Procedures outlined by the government, updated health information and modules to enhance their skills and knowledge in healthcare delivery.

By providing accurate and credible information, the chatbot will empower both mothers and ASHAs, giving them agency to make informed decisions and take appropriate action.

In the two instances as illustrated below, the AI-Bot served as a valuable resource for women with limited literacy, providing them with accurate and accessible information on important health topics.

 Vignette 1 – An illiterate woman in the urban setting asks a question which she hesitates to ask anyone. She does not have the right words, injunctions or placing a context.  But she definitely has an ally in the AI-Bot who answers her question. 

‘I’ve just recently delivered. What is the best time for the next pregnancy?’

Vignette 2 – A woman with limited literacy (note the spelling mistake) is still able to ask a question, even though she struggles to find the right words.  And hey pronto, the bot has picked a nerve there in answering her question.

‘I’m having a ‘wait’ ‘descharge’, I’m scared what should I do’

Understanding the needs of mothers: SNEHA’s approach to improving health outcomes focuses on the community to improve health-seeking behaviour through one-to-one counselling; group meetings and community events. Our intervention strategies are based on our understanding of community needs and the drivers and barriers to behaviour change. The team works very closely on a daily basis with the mothers/ caregivers we serve.

1. Understanding the gaps in the public health system: We work in close partnership with the public health systems, from the frontline workers (ASHA’s) to senior health providers to improve the quality of service to underserved communities. The team works in close coordination with public system staff to ensure regular delivery of quality services.

 This gives us a good understanding of the needs of both these stakeholders’ i.e community and public system, which is the basis of the proposed WhatsApp chatbot initiative.

The USP of our project positioning is:

 The SNEHA Didi se Poocho bot is endorsed by the Public Health Systems which has acknowledged its crucial role in advancing maternal and child health by improving access to information and support, promoting healthy behaviours and providing personalized care and support to individuals and communities.

• A key aspect of our approach is building strong relationships with communities to ensure buy-in and adoption of Technology solutions that enhance healthcare access.  This approach has enabled us to build trust and confidence among the community members, paving the way for successful implementation and adoption of IT solutions.   
• We have a full-time CTO with a data team
• The Board is also committed to supporting a technology solution to scale impact
• We have leveraged the expertise of Tech4Dev (https://projecttech4dev.org/) and Glific Teams, (https://glific.org/) who curate tech-enabled solutions for the development sector, on implementation on WhatsApp and data science.  
• Robust Monitoring and Evaluation and Program Implementation Team with rich experience designing and implementing evidence-based sustainable solutions. 

Mapping the Ground: Systematic situational analysis and need assessment were conducted to determine the use of information technology-based interventions.  A mix of qualitative and quantitative measures was adopted to gain an in-depth understanding of the needs, cultural barriers, myths and misinformation, common health practices and behaviour adopted by the community.  This informed that 65% are Android users while 87% use WhatsApp.  This survey also informed the scale of digital penetration, language preferences, and availability of time to engage with the Chatbot.  

Based on the above information, two pilots were conducted in the intervention areas:

1. The first pilot was rolled out to determine the engagement levels of the users in raising queries related to immunization schedules, myths, misinformation and others.
2. The second pilot was rolled out with a broader knowledge base using AI-LLMs to address cross-cutting health issues related to Immunization, Institutional Delivery and Exclusive Breastfeeding.   

Pilot 1 – September to October 2023

• 400 Mothers of 0 – 2 years children
• Topic: Routine Immunization
• Period: Six weeks
• Human intervention for responding to queries after 48 hours

Learnings

• Incorporate creative and engaging elements
• Users expected a response immediately 
• Incorporate Text and voice-based messages
• Scope for languages
• Enhanced knowledge base
• Need-based nudges 

Pilot 2 – February to March 2024

• 300 women
• Mothers of 0 - 6 months’ children and Pregnant Women in 3rd trimester
• Topics: Institutional Delivery, Exclusive Breastfeeding, Routine Immunization
• Period: Five weeks
• Gen-AI-generated responses within 15 to 20 seconds

Learnings incorporated

• AI/ML integration for immediate response to queries and collating information in real-time
• Nudges for behaviour change
• Expanded knowledge base and content relevance for comprehensive understanding and response
• Creative elements to enhance user engagement and elicit response

The two brief pilots successfully tested engagement rate and response accuracy.  In conclusion, the two pilots yielded significant outcomes: a) a strong buy-in from public health systems to implement the chatbot with a larger number of beneficiaries, and b) community acceptance of AI-ML-enabled technological solutions for improved maternal and child health outcomes.

Through this endeavour, we hope to:

• Seek guidance and mentorship from experts in the field to effectively help us build on the current solution
• Seek to learn from other Solve partners about solutions that others have used in advancing healthcare in low-resource settings
• Help in designing a viable business model for this solution to take it to scale
• Support in ensuring ethical aspects are incorporated into the solution
• Support in enhancing the creative use of communication tools to increase the use of the chatbot
• Support in enhancing data analytics from the AI/ML-based WhatsApp chatbot solution
• Support to identify and get buy-in from long-term strategic donors

Co-creating Experience: SNEHA Didi se Poocho chatbot is co-designed in tandem with the public health systems and the community boosted the chatbot's effectiveness and uptake, catalysing a better buy-in from the primary stakeholders.

Content Creativity:  the nudges that were embedded in the bot aimed to drive and sustain health behaviour change.

Remote Consultations:  enabling remote consultations and health monitoring.  For example, AI-powered chatbots have enabled us to provide basic health information and guidance to detect early signs of complications and seek immediate assistance for our users.

Information Kiosks: manned by government health workers, these kiosks will be placed at the health posts and places where immunization camps are held, mentoring and supporting young mothers and women to navigate the chatbot. 

Support Capacity Building of ASHAs: specifically tailored training modules with enhanced content and features viz. a.  Creative training elements embed with audio, and videos; b. Integrated scorecard to gauge the increase in knowledge and skills; c. incremental incentives to boost the morale of the ASHAs.

Knowing that some beneficiaries may not have access to the internet or smartphones, the trained ASHAs will ensure that these women are covered with health information and services during their regular home visits.

We see the promising changes that the integration of Gen AI could bring to the market landscape in several ways:

Improve access and affordability:  Additionally, an increase in access to information powered through technologies could potentially lead to a rise in demand for health services, which could further drive improvements in access and affordability, resulting in increased access to services, and thus reducing healthcare costs for both providers and patients.

Shift in Healthcare Workforce:  it will lead to a shift in the healthcare workforce, with a greater emphasis on roles that require skills in mapping patients’ needs, and greater engagement scoping for a greater need to upskill and re-skill the proficiencies of the healthcare workers.

Increased Focus on Preventive and Promotive Health: healthcare providers can potentially identify and address health risks before they escalate, leading to a greater emphasis on promotive and preventive care and designing information strategies in the engagement with patients.

Scoping Digital Health Solutions: such as mobile health apps, remote consultations, training and learning apps, will change the way healthcare is delivered and consumed, with a greater emphasis on reaching the last mile beneficiary with greater access to healthcare. 

The SNEHA Didi se Poocho chatbot solution is designed with two primary impact goals in mind.

Promoting women's empowerment and community engagement in maternal and child health by fostering education and raising awareness.  Additionally, it encourages active participation in decision-making processes regarding healthcare-seeking behaviours and practices.  Through accessible information and support provided by the chatbot, women and communities can make informed choices about their health, ultimately leading to improved maternal and child health outcomes. 

Enhancing the capacity of ASHAs by providing them with training and support through AI chatbots.  By leveraging technology to augment ASHA's skills and knowledge, they can effectively deliver maternal and child health services, including accurate assessment, counselling, and referrals.

The Ask SNEHA Didi chatbot solution is designed to improve maternal and child health outcomes in urban underserved communities of India. Our primary impact goals are as follows:

• Improved rate of institutional births
• Improved full immunization coverage of children under two years of age
• Improved maternal and child health service delivery by ASHAs

 Additionally, we will be tracking the engagement and effectiveness of the AI chatbot with its users:

• Number of interactions with the AI chatbots (Mothers and ASHAs)
• Users receiving timely and appropriate information enabling them to make informed decisions
• Satisfaction levels of the users on using the chatbot for healthcare information.
• Accuracy of information provided by the chatbot compared to standard guidelines

 The pre and post-evaluation design will be used to assess the effectiveness and impact of the AI chatbot solution on improving maternal and child health outcomes and enhancing the capacity of ASHAs. Before the implementation of the intervention, baseline data will be collected to assess the current status of key maternal and child health indicators. Following the implementation period, post-intervention data will be collected using the same indicators to measure changes and improvements.

 Our goal is to engage 10% of women of reproductive age from the total population of 800,000 across 15 health posts under the Bhiwandi City Municipal Corporation. These women will be those seeking health services at the health posts.

After having successfully rolled out the AI-enabled chatbot with 700 women in Bhiwandi City, in partnership with the Public Health Systems, we seek to scale it up across seven cities in the Mumbai Metropolitan Region in a phased manner and thereafter at the State Level across India, subject to government approvals.     

Core technology powering SNEHA Didi se Poocho includes:

 Whatsapp enabled conversations:

We use Whatsapp Business API (https://business.whatsapp.com/products/business-platform) an offering of Meta to create personalized conversations with our end-users.

 Conversational Flow Design:

SNEHA staff uses Glific (https://glific.org/) to create engaging conversations that address the knowledge gaps, deliver timely nudges to effect behaviour change, and get analytics from the engagement that the beneficiaries have with the chatbot. Glific is also an open-source API-based platform, which provides the potential for SNEHA to integrate with other data sources that could help influence messaging behaviour and more comprehensive analytics using modern machine learning methods.

 Generative AI LLMs:

 We currently leverage OpenAI Large Language Models for the generation of responses in conjunction with the Retrieval Augmented Generation (RAG) architecture from Jugalbandi (https://www.jugalbandi.ai/) developed by OpenNyAI in collaboration with Thoughtworks and Microsoft Research Labs. This helps us obtain accurate information with guardrails on sources of data to known validated health sources particular to the Indian context instead of the entire body of knowledge available to Open AI. Our solution is LLM and RAG agnostic and in the future, we envision leveraging any and all models from open-source or paid sources such as Llama, Claude, Anthropic, Gemini etc. and other RAG frameworks.

 Text to voice and voice to text:

 For handling text-to-voice and voice-to-text features of the chatbot, the Bhashini platform (https://bhashini.gov.in/), maintained by the Government of India, is used which utilizes open-source translation and transliteration models developed by AI4bharat particularly for Indic languages spoken in India.

Three internal product/data teams (part-time)

Five internal program implementation teams (full-time)

One CTO (full time)

One Fractional CTO (part-time)  

Two staff from Glific and Tech4Dev consultants who develop products and work in the social sector

It’s less than a year since the organization has been using AI-powered solutions for healthcare. 

 Nonetheless, for 25 years the organization has been endeavouring to strengthen an eco-system comprising the public health systems and the community, and also establish sustainable health models of intervention.    

SNEHA is a secular organization.  It has a non-discriminatory policy that applies to religion, gender, caste, class and creed. This means that we recruit and treat all individuals fairly and equally, regardless of their background. We strive to create an inclusive environment where everyone is respected and valued for who they are, and where opportunities are accessible to all, without any form of discrimination.

Among the 490 full-time employed staff, there are 397 females and 93 males, indicating a strong female presence. Similarly, among the 81 consultants, 76 are females and 5 are males, further highlighting the organization's efforts to leverage and empower women's capabilities. (Staff figures are for Apr, 23 to Mar, 24) 

The strong female presence in the organization's workforce extends to its senior leadership, which is noteworthy. The Founding Chairperson, CEO, and Executive Director are all women.  Their leadership roles showcase the organization's efforts to empower women and leverage their capabilities in key decision-making positions.

The organisation is characterised by a diverse array of skill sets among its members. It includes medical professionals such as doctors and clinicians, information technology (IT) specialists, social scientists, public health specialists, researchers, implementation researchers, scientists, clinical psychologists and others.  Together, these diverse skill sets enable the organization to take a comprehensive approach to addressing health and behaviour change challenges. 

At SNEHA, we prioritize strict adherence to ethical standards in implementation practices. To this end, we will develop a comprehensive study protocol, encompassing detailed study processes, indicators, and an analysis plan.   To ensure the privacy of our beneficiaries, we adhere to the confidentiality principles of our participants. We believe that data sharing should maximize the benefit to the public so the datasets analyzed in the course of this endeavour can be made publicly accessible as part of our commitment to transparency and responsible research practices.  SNEHA has a whistle-blower policy which is designed to enable us to be aware of possible malpractice and to take appropriate steps to deal with it immediately.

Funding: We secure funding from a variety of sources including trusts, foundations, corporate social responsibility initiatives, high-net-worth individuals, well-wishers, individual donors, campaigns, and awards. These funds enable us to provide free services to communities and public health systems. Self-generated income from our investments is used to partly cover our administrative costs.

 SNEHA Staff: We have a team of 490 employees and 81 consultants who design, implement and evaluate our programs. They are hired based on their qualifications, experience and competencies for the role and are critical to the success of our work

 Government partnerships: The government is a key partner. Our beneficiaries use their services and our role is to support them in a)improving the quality of service b) ensuring coverage of all beneficiaries c) Linking the community to government services.

The government also utilizes our services in the form of adopting our replicable and scalable models to achieve impact at scale. (For e.g. Our maternity referral model for high-risk pregnancies has been adopted by the public health systems.). Similarly, SNEHA was appointed as an ‘Implementing Agency’ to run the One Stop Center, on behalf of the government for women and children survivors of violence.

We also undertake training of government staff, frontline workers, supervisory staff, medical professionals and administrative staff to improve services to these vulnerable communities.

 Value added to Community volunteers providing free community service: We identify, train and keep motivated, a cadre of community volunteers comprising women, men and adolescents, living in the communities we serve. We build their knowledge of health and their capacity to support their own community members in healthy behaviours and practices. In a volunteer study we conducted to understand the motivation of these volunteers who provide pro bono services, the main drivers are the knowledge they gain, recognition in the community and the deep satisfaction they gain from serving others and being agents of change. 

 Value added to beneficiaries: Our primary stakeholders for whom we work are the residents of vulnerable informal settlements. SNEHA seeks to add health-related knowledge and promote positive behaviour change.  By providing them credible, relevant and timely information on health, nutrition and safety, and supporting their efforts to access appropriate services, we achieve desired health outcomes.

 NGO Collaboration: SNEHA has accumulated evidence-based healthcare models and has developed technical expertise with strong support from the medical community. We collaborate with other non-profits and development partners to share knowledge and insights, enhancing their projects. For example, SNEHA serves as a knowledge partner and technical expert for addressing malnutrition in collaboration with some NGOs. Development partners seek to utilize tailored versions of our solutions in areas where our teams are co-located and aligned with government priorities.

  Impact communication: We regularly report to donors, government partners and NGO partners on the impact of our work through donor reports, research studies, and surveys, baseline, midline, and end-line evaluations, which are conducted and widely disseminated across a large stakeholder base.

We propose to fund the scale-up of this technology through:

 Program donors i.e. we will add the cost of technology to our other program implementation costs since it will become a core strategy for behaviour change.

• Seek NGO partners working on maternal and child health and enable them to introduce this technology solution, funded by their program donors.
• For our government partners, we will advocate for this cost to be included in the budgets of primary health facilities.
• Explore opportunities with telecom providers for data packs for mothers and improve network connectivity to increase the use of chatbots in vulnerable informal settlements.