What Frontline Health Workers Told Us: Acceptability Evidence from 100+ ASHAs

A tool that frontline workers cannot or will not use will not achieve its purpose, however accurate its predictions. The AI-NETRA Pune Pilot Study therefore treated acceptability and usability as a main strand of the evaluation, through structured work with the people responsible for using the tool in the field.

Preparation in the field

Before deployment, one hundred workers across four tuberculosis units were trained on their own devices. They learned to read hotspot maps, open satellite views, and identify the next household for screening. This preparation was treated as a condition for the tool to work in the field.

Programme leadership

Fifteen members of programme leadership reviewed AI-NETRA's outputs, generated from their own district's data. Overall willingness to adopt the tool was 93 percent. Landmark-based hotspot identification and key-population mapping received the strongest support. Leadership preferred delivery on a mobile phone over paper methods or web-only portals.

Frontline workers

One hundred frontline workers, mostly ASHAs, community health workers, took part in sixteen group discussions between April and May 2026. All participants owned smartphones. Most already used mapping applications. All preferred outputs in Marathi, the local language. Seventy-five percent said they would use the tool, and 96 percent said they would recommend it to a colleague. Average ratings for usefulness and ease of use were both 4.5 out of 5.

After the campaign

The most informative evidence comes from a survey of 101 ASHAs after they had used AI-NETRA plans under real field conditions.

• 83 percent agreed the identified hotspots were genuinely higher-risk

• 92 percent reported a reduction in workload

• 86 percent reported a saving of time

• 94 percent found navigation straightforward

• 96 percent reported identifying the same suspected cases while screening a smaller population

• 98 percent considered themselves adequately trained

The finding that 96 percent of workers identified the same suspected cases while screening fewer people matches the reduction in people screened per confirmed case in the quantitative analysis. The two findings support each other. It should also be noted that only 41 percent of workers felt there were more people needing a test inside hotspots than elsewhere. That perception is consistent with the study's result that total case detection, when every form of confirmed tuberculosis is counted, did not differ much between the two arms.

What did not work well

Two issues stood out.

A display that showed workers a target number of people to screen was regarded favourably by only 19 percent. Workers described it as an unrealistic addition to an already demanding workload. The report recommends removing it.

Turn-by-turn navigation failed for five of seven workers in one dense urban unit. A separate time study found that simply opening the relevant plan took about a minute of extra work, because the process required several steps across more than one application.

Constraints at programme level

District and state leadership named geospatial data quality as the main constraint. Address records in the notification system are often free text and inconsistent. Permanent addresses are used rather than current ones. Geo-coordinates exist in the underlying system but are missing from standard data exports. Leadership also named timely payment of ASHA incentives, reimbursement of data costs incurred by frontline workers, and reliable access to mobile chest X-ray facilities as conditions for realising the efficiency gains shown in the pilot.

Implications

These findings point toward an offline, field-first application in the local language: one that opens directly to the day's plan, offers voice-guided navigation, works without a network connection, lets workers record households they find independently, and writes results back to Nikshay in a single action. Photographs from the field deployment are available on the Pune Pilot Study page, and the full acceptability analysis is in the study report.