Sahayak: Healthcare Advocacy Chatbot

We are seeking to solve the problem of inequitable access to healthcare. Social and cultural inequalities shape our lives and influence our ability to access key resources such as education, livelihoods, and justice. Access to healthcare is no exception. For instance, as per the Oxfam India report of 2021, one in 4 Indians faces discrimination while accessing healthcare in India based on their religion and caste. This discrimination manifests in various forms, including denial of patient records, case papers or investigative reports. According to this report, 74 % of respondents mentioned that their doctors prescribed treatment, further tests and/or medication without fully explaining their illness, nature and/or its cause. There also exists anecdotal evidence of the discrimination that the Dalit and Tribal communities face while accessing healthcare. The overall poor health outcomes amongst them are a direct result of the discrimination and stigma that they face while accessing healthcare in India.

Similarly, despite recent legal reforms in India, the LGBTQIA+ community, which constitutes around 10% of the Indian population (around 135 million people), continues to face systemic exclusion from healthcare. Discrimination against them manifests in various forms, including but not limited to judgment at the hospital entrance gate, refusal of treatment, lack of understanding about sex- and gender-related diseases, and labelling non-heterosexual sexual orientation as a psychological disorder. Furthermore, weight bias and fat shaming are rampant within healthcare, which influences fat people’s healthcare services-seeking behaviour. Anecdotal evidence shows the bias that fat people face when visiting a doctor in various ways, from attributing all kinds of illnesses to being fat to refusal of treatment if the patient in question does not lose weight to dismissal of concerns due to fatphobia. Such experiences make people apprehensive of visiting doctors, putting them at an increased risk for illnesses.

These examples show that people with different intersectional identities face discrimination while accessing healthcare, which further influences their healthcare-seeking behaviour leading to detrimental health-related outcomes. While such instances of discrimination and bias are highly common, there are laws to prevent such discrimination and redressal mechanisms to address complaints and concerns. However, information that facilitates a patient’s ability to advocate for themselves and hold healthcare systems accountable remains sparsely available. Patient advocacy can ensure not only that patients have access to quality healthcare but also that healthcare providers are accountable for their actions. Hence, through this project, we aim to address a critical gap in the field and build a solution that equips individuals with the information required to advocate for themselves.

Our solution is named Sahayak, a healthcare advocacy chatbot designed to strengthen patient rights within the healthcare system by providing them access to information about their substantive and procedural rights as well as by providing them guidance on negotiating with and advocating for their rights within the healthcare system. This chatbot uses the API from Jugalbandi, an open platform combining the power of Generative AI and Indian language translation models to power conversational AI solutions. When a user tries to interact with this chatbot, they will be presented with 3 options: ‘I want information on legal rights’, ‘I want to information on how to file a complaint’ and ‘I am not sure’. After selecting one of the options, the user will be prompted to describe the situation and ask questions. At the backend, the team has developed a comprehensive database that includes relevant information within healthcare and updated information on hyperlocal legal support that will enable the bot to provide the information as requested. The chatbot does not provide any medical advice to patients. It is designed to offer information, support, and guidance related to healthcare advocacy and patients' rights, but it does not replace professional medical advice or diagnosis.

Further, the chatbot operates in an advisory capacity only. The advice provided should not be considered binding legal advice or a substitute for professional legal counsel. Users will be encouraged to seek personalised legal advice for specific situations. This chatbot does not track any actions taken by the user after obtaining information, and it is meant to equip users with the required knowledge and information to advocate for themselves. However, there will be an in-built feedback mechanism to capture the feedback from the users on the obtained information.

The target audience for the first iteration of the chatbot includes individuals in the age group of 18-45 years. We recognise the limitations of building a tech-based solution that requires users to have access to a smartphone and proficiency in certain languages. Thus, we are targeting individuals residing in urban and semi-urban areas, focusing on marginalised communities who may face socio-cultural discrimination within the healthcare system. Currently, the chatbot functions in two key languages - English and Hindi. However, we hope to incorporate additional languages in future iterations. Since the information on grievance redressal and patient advocacy is sparsely available, we believe a chatbot that can offer hyperlocal information on negotiating and advocating within the healthcare system and strategies to assert their rights will enable individuals to advocate for themselves. The chatbot provides hyperlocal information for healthcare establishments in Delhi, Bangalore and Mumbai. We will expand the database to incorporate other cities as we progress with the development.

We are a team of young women and queer folks who have faced differential treatment from healthcare professionals due to gender, disability and fatness. Our lived experiences, along with anecdotal experiences shared by the community, inform this solution. As an organisation, we work with women and queer folks across our projects and are well-versed in the challenges that they face while accessing healthcare. Within the community, individuals often reach out for requests for inclusive, non-judgemental healthcare providers, given their negative experiences. Further, we are undertaking a research study on Fatness in Urban India, which has also reaffirmed the bias and challenges that fat people face while consulting healthcare providers due to their body size, shape and weight. We have pioneered legal aid provision through our FemJustice Helpline and Labourline, wherein we have provided legal information to survivors of domestic violence and in cases of labour rights violations. Thus, we have prior experience in providing legal information in simpler languages, enabling individuals to take action.

We have a functional chatbot that is operational on Telegram, and we are getting initial feedback by testing it with our communities. The demo video for the same has been attached to the previous question. So far, we have incorporated around 30 use cases, collated hyperlocal data for 3 cities (Delhi, Mumbai and Bangalore) and tested it with around 25 users to gather feedback on different features and elements of the chatbot.

While our team has subject matter knowledge and expertise, we have limited technical expertise for the technical development of the chatbot, such as AI and LLM skills. We would appreciate support in the technical domain for designing an effective solution and user adoption/growth. We also want to build our capacity to scale this solution quickly and efficiently. Currently, we have limited resources to access the required software licenses and are relying on open-access resources. We want to build this solution as an open-source model with an opportunity for other machine learning engineers to contribute to this project. We hope to build a strong network of community and mentors who can offer feedback and advice in scaling this solution.

Existing chatbots focus increasingly on providing patients with access to information about their diagnosis, inputs into managing symptoms, and/or insight into behaviour changes required to better one’s health. However, there are limited solutions that focus on equipping individuals with relevant information to advocate for themselves within the healthcare system. Our solution, Sahayak, offers not only legal information on patient rights but also hyperlocal information on asserting these rights. Given our previous experience of providing legal aid, most people prefer direct mediation or approaching the relevant authorities themselves to address their concerns, before escalating a complaint and approaching law enforcement authorities such as lawyers or police. Thus, the solution offers hyperlocal information based on the user’s location. Since the chatbot is not tracking the user’s location, it will rely on user input to provide such information. We believe this will build users’ confidence in navigating healthcare systems and advocate for themselves and their families and friends without resorting to expensive and often inaccessible legal remedies.

As the user base grows, we believe the user-bot interaction may present some trends related to healthcare discrimination and bias. We hope to collate this information and hold institutions accountable by advocating for changes in curriculum, better training of healthcare providers and compliance of healthcare institutions with the relevant laws.

As individuals become more empowered to stand up for their rights, the incidence of oppression will reduce. By providing accessible information and hyperlocal guidance on advocating for their rights, this chatbot aims to ensure that individuals are able to stand up to discriminatory and biased practices within the healthcare system. When individuals have access to the relevant information and strategies to advocate for their rights, they can hold the institutions and service providers accountable. As more and more people become aware of their rights and capable of advocating for them, there will be a shift in service providers’ practice, and thus, the overall incidence of discrimination will decrease. Thus, the chatbot seeks to foster a culture of transparency and inclusivity within the healthcare system by amplifying patient rights, enabling individuals to advocate for themselves and holding healthcare providers and institutions accountable.

The impact goal for this solution is to ensure that individuals feel more empowered and confident to access and navigate healthcare infrastructure in India. They will have access to legal rights pertaining to healthcare systems. They are able to advocate for themselves and assert their rights in the face of biased and discriminatory practices of healthcare providers. Additionally, we foresee a possible shift in such practices of healthcare providers as an indirect impact goal.

Key indicators for this solution include the following:

1. Number of users accessing the chatbot

2. Number of users who have reported applying/using advocacy strategies provided by the chatbot (via constant monitoring and feedback)

3. Number of users who have filed a complaint on the basis of information provided through the chatbot

Currently, our solution, Sahayak, is operational on Telegram and is built using the Jugalbandi stack. The chatbot uses the API from Jugalbandi, an open platform combining the power of Generative AI and Indian language translation models to power conversational AI solutions. It also uses elements of machine learning to augment the responses given to the user queries. At the backend, the team has worked to develop a comprehensive database of 20 use cases which currently forms the basis of information provided by the chatbot. Additionally, we have collated hyperlocal information on grievance redressal across 3 cities: Delhi, Mumbai and Bangalore.

Currently, we have 3 full-time team members working on it, along with 1 intern to support research requirements and 1 voluntary technical consultant.

Since July 2023

We are a feminist, social purpose organisation with a vision of a world built on social justice, led by communities of care. We are committed to working in a way that is just and equitable. We have adapted our approach to social justice, of ‘Nurturing Radical Kindness’ (i.e., the opposite of violence), into a set of eleven guiding values to inform our ways of working. Some of the guiding values include intersectional feminism, value lived experiences, participatory, intent and impact and care. Our internal policies and workings are informed by these values. We strive to create an inclusive, safe and growth-oriented workplace environment. We follow a participatory decision-making process, not based on consensus, to ensure that everyone has the opportunity to contribute positively to the work. We have a robust employee wellness program named ‘Holistic Security Initiative’ under which we host spaces to heal, connect and grow. We also adhere to a code of conduct that lays down rules for behaviour within the workplace, with a robust grievance redressal process. Additionally, we have anti-bullying, harassment and discrimination policy that prohibits bullying or harassment of any kind.

We are an equal opportunity employer and are fully transparent with our salary structure and job responsibilities so that potential candidates can make an informed decision about their application to work with us. To make the application process inclusive, we ask for specific accommodation needs and strive to address those needs to the best of our capabilities. We also strongly encourage applications from historically oppressed communities, as far as possible.

We would like to keep the chatbot free-to-use, as far as possible. This chatbot will be available to individuals facing healthcare discrimination, who are the primary beneficiaries of this solution. We would also partner with healthcare institutions and providers to improve patient advocacy and healthcare equity. We aim to propose integrating this service with existing patient support mechanisms at institutions and hospitals.

Through the chatbot, we will continue to offer legal rights information, hyperlocal guidance and support and advocacy tools to equip users with information on advocating and navigating the healthcare infrastructure.

Through rapid testing, we want the future iterations to incorporate feedback from users as quickly as possible and ensure that the chatbot is providing services in the most efficient and accessible manner. We have maintained user-friendly language and interface and hosted it on a messaging platform that our target audience is already familiar with. Further, the hyperlocal information feature will ensure the chatbot's effectiveness in providing specific information per the user’s location.

We will uphold data security measures and ensure that all data is stored safely and securely. We will not publish any individual markers of the user, however, we may publish emerging trends from the data that is collected at the chatbot.

We have been applying for various grants with this project. In the future, we will use various ways to raise funds for this project, including crowdfunding and institutional funding. The aim is to have diverse funding sources for this project to ensure financial sustainability. We hope to develop a viable PoC to demonstrate our theory of change and build our case for continued funding from our previous funders. We aim to reach 5000 people within the 1st year of deployment of the chatbot, and we will partner with other institutions, such as educational institutions and healthcare providers, to achieve this goal and establish the viability of this tool. We are also building this solution as open source, inviting contributions from other machine learning engineers. This will help us strengthen our case for future funding as well.

In the past, we have received funding under programmes like Azim Premji Covid Research Funding Programme 2020 (USD 11,000). We also received a small grant amount towards the research for the Chatbot (Sahayak) from Point of View.