Making Every Voice Count: Muso’s Approach to Equitable Artificial Intelligence (AI)

The patients we have the most to learn from are often those with the least access to healthcare and the technologies shaping its future. They include families living in remote communities and areas affected by armed conflict, often without reliable connectivity, mobile phones, or access to digital tools. Yet their experiences offer important insights into the barriers people face when seeking care and what needs to change.

AI is opening new possibilities for addressing challenges in healthcare, and Muso is exploring how these technologies can help strengthen community health delivery. This work focuses on three areas: supporting frontline supervision, helping health teams better understand patients’ experiences, and anticipating medicine needs.

A proCHW speaking with a patient during a home visit in Yirimadio.

In frontline supervision, AI-assisted tools are already being used in Mali to help supervisors provide more targeted support to Community health workers (CHWs). Since February 2025, these tools have generated monthly supervision guidance in French and Bambara for approximately 440 proCHWs supported by 30 supervisors, serving approximately 500,000 patients. Integrated into existing supervision processes, this guidance helps supervisors identify where individual CHWs may need additional support and coaching.

Muso is also developing AI-enabled medicine-demand forecasting to help health teams anticipate commodity needs and reduce the risk of stockouts. This builds on existing digital stock-management tools that support the tracking and replenishment of medicines, while the predictive forecasting capability remains under development.

Turning Patient Experiences into Actionable Insights

A more recent area of work focuses on making better use of patients’ experiences. With support from the Gates Foundation, Muso is working with Mali’s Ministry of Health and the Masakhane African Languages Hub through Project ECHO to develop AI tools that can transcribe and analyze patient feedback in Bamanakan and French.

Every month, proCHW supervisors across Muso sites conduct more than 2,000 patient interviews, generating nearly 25,000 conversations a year. Historically, supervisors entered patients’ responses into the CHW application as short free-text entries, often limited to just a few words. While this made it possible to record feedback, it limited the amount of detail that could be captured and made it harder to identify recurring issues across thousands of conversations.

By identifying recurring themes across interviews, the tools being developed through ECHO aim to help supervisors, program managers, and health authorities better understand barriers to care, including those related to cost, distance, and experiences with healthcare providers, and identify where services may need to improve.

Because Bamanakan remains underrepresented in AI systems, a central part of this work is ensuring that the tools can accurately capture and analyze feedback in both Bamanakan and French. The project also integrates a gender-barrier taxonomy into the AI analysis, allowing community feedback to be classified by both health topic and type of barrier to accessing care. This will help identify challenges that particularly affect women and girls, including delays in obtaining spousal approval, transportation costs, concerns about adolescent confidentiality, and barriers to attending antenatal care visits or postpartum follow-up. By bringing these patterns to light, the tools can help health teams better understand the barriers women face and make services more responsive to their needs.

Together, these efforts reflect Muso’s commitment to applying AI where it can make a practical difference: helping supervisors better support frontline workers, enabling health teams to anticipate medicine needs, and making patients’ experiences more visible in decisions about care.