How an AI tool is helping ASHA workers monitor newborn health

Until last year, Jyotsna Suresh Patel, an ASHA (Accredited Social Health Activist) in Daman’s Kachigam village carried a heavy weighing scale to weigh newborn babies. The process was long and she struggled to ensure accurate readings while soothing anxious mothers.

Artificial intelligence (AI) is now helping her accurately weigh babies in seconds. 

Today, all Patel has to do is pull out her smartphone and take a 15-second video. Within minutes, the anthropometric measurements of the newborn that includes weight, length, head circumference, chest circumference, and mid-upper arm circumference are recorded. 

This transformation has been made possible by Shishu Maapan, an AI-powered tool developed by Wadhwani AI that is set to change monitoring of newborn health in India.

Bridging a critical gap in newborn health tracking

In India, under the Home-Based Newborn Care (HBNC) programme introduced in 2011, ASHAs visit newborns on designated days—3, 7, 21, and 42 days after birth—to monitor their growth. Traditionally, they relied on either heavy digital weighing scales or Salter scales, both of which posed challenges.

“The digital weighing machines are heavy, and it’s practically difficult to move them from one place to another, especially when they have to traverse through rough terrain. In order to address this, the government came up with the Salter scale,” explains Dr Sneha Nikam, Senior Program Manager, Wadhwani.

The Salter scale that comes with a hook and a bag to place the child, though feasible in terms of portability, was uncomfortable for the child and did not offer stable measurement, she points out. 

For the past six years, Wadhwani AI has worked at the intersection of AI and social impact, focusing on health, education and agriculture working closely with ministries to ensure their integration into existing government programmes. 

“We create AI products and integrate them into the respective governments’ applications or programmes. All our products we create are mostly open source and we deploy standard software development that can easily be adapted to any application used at the state or national level,” explains Prasaanth Balraj, Group Product Manager, Wadhwani AI.

Last year, Wadhwani AI deployed the Shishu Maapan tool in Dadra and Nagar Haveli and Daman and Diu, closely working with the health department and the administration in the union territory. 

“The first quarter of the year was dedicated to fine-tuning the AI model, where we captured the baby’s videos as well as gold standard equipment were used to capture the anthropometry. Measurements of the child were then compared to the gold standard and the model was fine tuned,” Nikam elaborates. Around 450 ASHAs have been trained and are using the tool so far in the Union Territory.

How Shishu Maapan works

The Shishu Maapan AI tool can be integrated into an application used by ASHAs in a particular state, and where such an application is not available, they can use the Shishu Maapan app, a platform that hosts the AI model.

Using the application, a frontline worker uses a smartphone to capture a short video of the newborn baby who is placed unclothed on a flat surface, next to a wooden ruler. The AI tool extracts key anthropometric measurements from the video in real-time.

Once the AI model processes the data, the baby’s weight, height, and other measurements appear on the app. If the internet is unavailable, the data is stored in the app’s cache memory and automatically syncs to the system when connectivity is restored.

The video is not saved on the device to protect privacy; it is deleted as soon as the measurements are extracted. The measurements are then sent to public health administrators for real-time monitoring.

“We were also mindful of ethical considerations. The video is not stored in the phone’s gallery; it is processed in-app, and once the AI inference is generated, the video is deleted to protect privacy,” explains Balraj. 

According to Nikam, Shishu Maapan reports an average weight error of only 111 grams. The average error for other parameters such as length, head circumference and chest circumference is below 1.3 cm.

Patel, who has been an ASHA worker for 13 years, shares that when she first told a mother they were going to weigh her child with the help of a mobile phone, she thought it was a joke.

“We explain to them how the process works, and only if they are okay with it, we go ahead. Now, once a visit is done, the next date immediately appears on the app. Now they are used to the process, and keep enquiring about their baby’s progress on all parameters,” she says.

“One ASHA worker told us she feels proud using an AI-powered tool, just like professionals using laptops in offices,” adds Nikam.

Challenges in scaling nationwide

While the AI tool has shown promise, several challenges remain in scaling it nationwide. 

Nikam outlines a few. 

Each new region requires fine-tuning with local data to ensure accuracy. To validate AI-generated measurements, calibrated digital weighing scales and infantometers are needed, but many state health departments lack these resources.

Government officials and health administrators often hesitate to adopt AI-driven tools, doubting their reliability and impact. Continuous engagement and capacity building are required to gain their confidence.

Not all ASHAs have smartphones, and those that do may have older models with limited storage or non-functional cameras, posing a barrier to adoption.

Also, while ASHAs are increasingly comfortable using smartphones, some stakeholders remain unconvinced about their ability to use AI-powered tools effectively, creating some resistance.

A scalable solution

Despite the hurdles, Wadhwani AI is hopeful about taking Shishu Maapan to more states.

“Looking ahead, we want to demonstrate that the tool is usable across atleast five-six states so that it can be proposed to the Ministry of Family Health and Welfare (MoFHW) to take it to all newborns across the country,” says Nikam.

However, she also reiterates that Shishu Maapan is not meant to replace existing methods but to serve as a complementary solution that facilitates timely referrals and enhances community-based newborn care.

In the long term, the organisation envisions expanding the tool’s use beyond ASHAs to parents, empowering them to monitor their newborn’s health at home.