Digital Communication Algorithms Using Semi-automated Chatbot for Community-based Services on Posyandu Application in Indonesia: Model Development with Qualitative Approach

In many parts of the world, keeping a community healthy relies on a simple, human rhythm: a local health worker visits a home, asks questions, and offers advice. This personal connection is the backbone of public health, yet it often struggles against distance, time, and the sheer number of people who need care. When a pregnant woman lives far from a clinic, or a mother of a toddler cannot easily reach a nurse, the flow of vital information can stall. Technology has offered a way to bridge these gaps, promising to deliver health advice through the smartphones that are now in the hands of billions. However, a fully automated computer program that answers every medical question is not yet ready to replace a human doctor; it can make mistakes or miss the nuance of a complex situation. The challenge, then, is to find a middle ground where technology handles the routine, and humans step in for the rest, creating a system that is both fast and safe.

This is the specific problem a team of researchers in Indonesia set out to solve. They focused on Posyandu, which are community-based health posts that serve as the first line of defense for mothers and children in villages. These posts rely heavily on local volunteers, known as cadres, and midwives to track the growth of children and monitor the health of pregnant women. In the Purwakarta district of West Java, the researchers worked with a group of midwives, cadres, and mothers to design a new digital tool. Their goal was not to build a robot that could diagnose diseases on its own, but to create a "semi-automated" chatbot. This is a digital assistant that can answer common questions instantly but knows when to pause and hand the conversation over to a real person. The researchers wanted to understand exactly what kind of conversations needed to happen between these different groups and how a computer could help facilitate them without getting in the way.

To find the answer, the team did not start by writing code. Instead, they spent months listening. They held detailed discussions with eleven midwives, thirteen cadres, and thirty-four women representing different stages of motherhood, including those who were pregnant, had just given birth, or were caring for young children. They asked these participants what information they needed most, what frustrated them about current communication, and how they imagined a digital tool could help. The conversations revealed a clear pattern of needs. Pregnant women wanted reminders about when to take vitamins or check their baby's movement. Mothers of toddlers needed advice on nutrition and how to handle common childhood illnesses. The cadres, who often act as the bridge between the community and the health system, needed quick access to facts about immunization schedules and growth monitoring. The midwives, in turn, needed a way to coordinate with doctors and other health officials when a case became too complex for them to handle alone.

From these conversations, the researchers identified four distinct types of communication that the new system had to support. The first is educational communication, where the system provides health facts and advice to mothers. The second is facilitation, which helps the cadres and midwives work together more smoothly. The third is coordination, allowing midwives to connect with other health professionals, such as nutritionists or pediatricians, when a patient needs specialized care. The fourth is advocacy, a channel for health workers to share data with government officials to help shape better health policies. The researchers realized that a single, rigid computer program could not handle all of these tasks perfectly. A chatbot could easily remind a mother to take her iron pills, but it could not diagnose a rare complication or negotiate a policy change with a district official.

The solution they developed is a digital algorithm that acts as a smart switchboard. In this system, the chatbot handles the routine educational questions, providing instant answers about pregnancy care, breastfeeding, or child nutrition. If a user asks something the computer does not know, the system forwards the message to a human midwife for manual follow-up. This semi-automated approach ensures that the speed of technology is combined with the judgment of a professional. The researchers mapped out exactly how these conversations should flow, creating a model where a mother can get immediate help for a minor worry, while a serious health issue is flagged for human attention. They also designed the system to allow cadres to report community issues to midwives, and for midwives to coordinate with doctors, all within the same digital space.

The study found that this hybrid model addresses the specific limitations of the current health system in the region. In Purwakarta, there are far fewer midwives than the ideal ratio would suggest, making it difficult for every mother to get face-to-face time with a professional. By using a chatbot to handle the frequent, repetitive questions, the system frees up the midwives to focus on the cases that truly require their expertise. The researchers noted that while the technology is promising, it is not a magic bullet. The system depends on reliable internet signals, which can be spotty in some rural areas, and it requires users to adapt to a new way of communicating. However, the interviews indicated that participants recognized the potential for a tool that respects the need for human connection while using technology to remove the barriers of distance and time.

Ultimately, the paper presents a blueprint for a digital health assistant that is designed to fit the reality of community care. It does not claim to replace the midwife or the cadre, but rather to empower them with a tool that extends their reach. By separating the routine from the complex, the semi-automated chatbot allows the health system to function more efficiently, ensuring that information reaches the people who need it most, exactly when they need it. The work suggests that the future of public health in places like Indonesia lies not in choosing between human care and digital tools, but in weaving them together into a single, responsive network.