AI for equitable health systems: better data, access and health outcomes

In the foothills of Mount Kenya, 25-year-old Kinya Stell serves her community as a public health professional in Embu town, a major trading centre. Like many young people, she has questions about HIV and sexual and reproductive health. To find answers, she asks Beshte — a chatbot powered by artificial intelligence (AI).  

“When talking with Beshte, it’s more of a friend than an automated app, because a human being will judge me according to my story,” she explained.  

Responsible AI-powered tools like Beshte are helping strengthen health systems while promoting more equitable access to health information and services. Designed to be ethical, inclusive, rights-respecting and sustainable, responsible AI incorporates contextually grounded knowledge with a local health systems perspective to promote health equity and achieve better health outcomes.  

Artificial Intelligence for Global Health

The Beshte chatbot is one of more than 50 locally-designed innovations funded under the Artificial Intelligence for Global Health (AI4GH) initiative – part of a larger AI for Development (AI4D) program jointly funded by IDRC and the UK’s Foreign, Commonwealth and Development Office. AI4GH aims to strengthen health systems by supporting implementation research projects that: 

• apply responsible AI solutions to improve sexual, reproductive and maternal health for women and girls 
• support more effective and equitable preparedness and responses to epidemics and pandemics 

AI4GH works across sub-Saharan Africa, Asia, Latin America and the Caribbean and the Middle East and North Africa. The initiative is rooted in the belief that local knowledge, innovations, talent and long-term vision should be the driving force for sustainable solutions that address today’s complex, intersectoral and trans-border health challenges.  

Strengthening health systems

Health systems tend to be under-resourced, and often failing those hardest to reach, and not responsive to the needs of certain segments of the populations.  AI4GH works to strengthen all six of the World Health Organization’s health systems pillars — the building blocks for robust and inclusive health systems: 

• Leadership and governance 
• Service delivery
• Health system financing 
• Health workforce 
• Medical products, vaccines and technologies 
• Health information systems 

The examples featured below show what responsible AI looks like in practice across two of the health system's core pillars — service delivery and health information systems. 

Cross-cutting considerations across AI4GH require every project to consider gender equality and inclusion — encompassing people living with disabilities, LGBTQ+, Indigenous, refugee and displaced communities, responsible AI, capacity strengthening and pathways for scale and sustainability. Funded projects are interdisciplinary in nature and span unmet needs within and across human health, animal health, environmental health — shaped by conflict, climate and community dynamics. 

Service delivery

A chatbot raising HIV awareness among adolescents in Kenya

When developed responsibly, AI-enabled solutions like Beshte improve health service delivery by reflecting local health systems and the social structures they are embedded within. In Kenya, for example, adolescents and young adults account for 40% of new HIV infections, but because of stigma, and health system failures manifested in untrained providers, and ill-equipped facilities, they are often reluctant to get support due to stigma, knowledge gaps, structural barriers and a lack of tailored interventions.

To address this problem, a team at the University of Embu co-created Beshte with input from psychologists, counsellors and high school and university students. Users can ask the chatbot personal questions, like what to do if they’ve engaged in unprotected sex. Beshte answers with compassion and guides them to reputable medical services. The chatbot’s users like how it can answer questions in English, Swahili and Sheng — a slang primarily spoken by youth. The tailored design and training of Beshte influenced its uptake among adolescents.  

The chatbot is available online and able to incorporate more languages spoken in Kenya, supporting future scale up to include more rural and low-literacy users. 

Decentralizing ultrasounds in Guatemala

Access to antenatal care in primary care settings is key to reducing maternal mortality and morbidity. Yet, across many rural and Indigenous communities worldwide, women face challenges in accessing health care during their pregnancy. In Guatemala, the NatalIA project is helping Mayan women overcome barriers to accessing maternal health care, such as cost and distance to clinics. With a 1.6 to 2.1 times higher risk of maternal mortality compared to non-Indigenous women, NatalIA is redressing health inequities among Mayan women.  

A multidisciplinary team at the Universidad Galileo worked closely with community midwives to provide access to affordable ultrasounds. Together, they developed a responsible AI-enabled solution trained on data from Mayan women, allowing midwives to provide high quality ultrasounds with a portable device. NatalIA brings ultrasound services and specialist support to communities that lacked these services.  It shows how low-cost, locally designed innovations — built on Indigenous knowledge and community trust — can expand access to care.  Initial findings across three primary healthcare facilities show promising user satisfaction and screening results for scaling to other centres.  

Health information systems

Improving community surveillance through mobile devices in Ethiopia

An AI-powered reporting tool called Haqila Lab is advancing community-based surveillance in Ethiopia. Initially designed for polio, a team at Jimma University created the app to update the country’s existing surveillance system. Haqila now includes other priority diseases in Ethiopia — anthrax, breast cancer, brucellosis, malaria, measles, mpox and rabies. 

Haqila strengthens capacities of community health officers conducting routine surveillance. It streamlines the surveillance process, allowing them to capture photos and videos during home visits and at clinics.   

The Ministry of Health has adopted the app as part of its commitment to leverage digital technologies and AI to strengthen health systems’ performance, a testament to the app’s scalability in low-resource settings.  

Strengthening mpox surveillance and readiness

Limited vaccine access and gaps in outbreak detection call for surveillance systems that can detect outbreaks and facilitate effective responses. When an mpox outbreak threatened African nations in 2024, AI4GH-funded researchers quickly assembled a multi-country network to address the threat using health information systems and trusted modelling practices. The AI4Mpox initiative emerged from these efforts to improve early warning, hotspot detection and analysis of health systems responses through AI-enabled modeling. 

In the Democratic Republic of the Congo (DRC), recent mpox outbreaks have been devastating — especially for children. The deteriorating security and humanitarian situation, combined with cuts in aid support, have made containing the virus challenging. The AI4Mpox team in the DRC created the AfiaGap application to strengthen health system resilience. The app integrates community-centered surveillance with AI-driven early warning, enabling faster detection, analysis and response to local health threats. In the first year, AfiaGap reached more than 125,000 people, including more than 6,000 people living with disabilities. It has enhanced coordination, improved resource allocation and fostered engagement between communities and health authorities.  

In Senegal, researchers built AI4Mpox-SN, a platform for community-based early detection of mpox and other zoonotic (animal-human) diseases. It combines an AI-powered chatbot, an alert function and a real-time dashboard, and the team has trained 250 border officers to use it. Communities and health workers can now detect and report threats faster than with traditional methods — strengthening the ability of Senegal's health system to catch outbreaks before they escalate future.   

Across these projects, one pattern holds: health systems can be strengthened by the use of responsible AI when these innovations are developed with communities, grounded in local realities and designed, for example, to fill access and data gaps. 

For people like Kinya Stell in Embu — and the communities behind every project featured here — that future is already taking shape: one question answered, one diagnosis made and one outbreak detected before it spreads. 

Contributors: Chaitali Sinha, with Samuel Oti, senior program specialists, IDRC.