Mothers at an NDOH clinic connect through MomConnect, receiving vital health updates directly on their phones.
Reach Digital Health (RDH) serves vulnerable populations, primarily focusing on women, children, youth, and marginalised communities across various health areas. Reach’s programs are designed to deliver essential health information and personalised support to individuals who might otherwise face barriers to accessing healthcare. However, within these programmes, the organisation works closely with local governments to ensure that the services provided are integrated into the healthcare system, functioning as an extension of that system from facilities to communities and directly into people’s homes.
One of its flagship programs, MomConnect, is a national mobile health initiative in South Africa that supports pregnant women and new mothers in public healthcare by delivering personalised health messages through WhatsApp and SMS. These messages provide critical information about pregnancy, newborn care, and maternal health, helping women make informed decisions about their own health and that of their children. The program also offers a help desk, enabling users to ask questions and receive real-time responses from healthcare providers employed by the National Department of Health (NDOH). This personalised, accessible support model ensures that even women in remote or under-resourced areas can access accurate and timely healthcare information, even without easy access to physical health facilities.
Critically, mothers connected to the MomConnect service can provide feedback to the NDOH on multiple issues encountered in the healthcare system, such as quality of care and access to medication and services. Through this feedback loop, not only are mothers in South Africa empowered, but the NDOH can also use real-time data from millions of users to improve the services it offers. For instance, MomConnect has been instrumental in detecting and addressing medication stock-outs nationwide. Immediate feedback from mothers reporting a lack of access to certain medicines, combined with clinic codes identifying precisely which facility is affected, allows for a swift response to low or zero stock of essential items, ranging from iron-folate supplements to vaccinations.
Reach Digital Health aims to enhance MomConnect’s health interventions by integrating AI-driven personalisation and conversational support. By leveraging large language models (LLMs), MomConnect can tailor messages, timing, and delivery channels to better align with individual user contexts. This approach is intended to improve user engagement, increase understanding of health information, and encourage sustained positive health behaviours.
Reach aims to introduce a generative AI tool that uses generative AI to collect user information through more natural, personalized conversations. Instead of static surveys and assessments, the LLM would adapt how it asks questions—adjusting tone, language, and style—based on each user’s responses and level of engagement. By making data collection feel more like a supportive dialogue, this approach is intended to encourage users to share more detailed and accurate information about their health contexts. The richer data gathered would then be used to tailor interventions, enhance user understanding, and ultimately improve health behaviors and outcomes.
Reach is currently building on insights gained from a July 2024 analysis that identified key drivers of user engagement and informed preliminary personalization strategies. The team is testing ways to integrate an LLM-based conversational approach into existing platforms like MomConnect, refining how content is tailored and delivered in real-time. Current efforts focus on developing prompt engineering techniques, exploring multiple languages and modalities, and preparing the groundwork for pilot studies. These studies will measure how conversational data collection and personalized messaging can enhance user participation, improve data quality, and ultimately lead to better health outcomes.