Whether to filter out fraudulent beneficiary applications for the Ayushman Bharat scheme or a voice-to-text model that turns a doctor’s speech into digital prescriptions, the Ministry of Health and Family Welfare’s stall is drawing attention at the AI Impact Summit 2026.
Showcasing an AI-based National Anti-Fraud Unit, developed for the insurance arm of Ayushman Bharat, officials said that leakages, whether in the form of abuse, misuse, wastage or deliberate fraud, are an inherent risk in any large health system. “This tool is aimed at reducing errors and delays. Only correct claims will move ahead and bring transparency. “It can track missing documents, unclear reports or wrong patient information. As of 2026, Rs 650 crore worth of public money has been prevented from misuse,” said an official.
“Any publicly-funded programme must ensure that every rupee reaches the intended beneficiary,” he added, outlining the rationale behind the creation of the National Anti-Fraud Unit (NAFU) under the National Health Authority (NHA).
Transparency in claims
Established in the early years of PMJAY, NAFU was designed as a dedicated institutional mechanism to strengthen compliance, detect anomalies and proactively plug systemic gaps. Over time, States were brought into the framework and encouraged to establish their own anti-fraud units, creating a coordinated national structure.
With the rapid digitisation of healthcare under the Ayushman Bharat Digital Mission (ABDM), vast volumes of patient and claims data began flowing through digital platforms. According to the official, this presented an opportunity to use advanced analytics not only for service delivery but also for financial oversight.
Under the new system, all claims data received by the NHA is securely stored in cloud infrastructure, which is then processed as per rules and by AI engines. The system evaluates whether hospitalisation was medically justified as per PMJAY’s standard treatment guidelines and evaluates if the prescribed procedures and benefits align with approved norms.
Common patterns of abuse, such as unnecessarily extending hospital stays to inflate claims or deviating from established treatment protocols, are automatically flagged for review. The AI models are trained to detect such anomalies at scale, allowing authorities to intervene before payments are finalised.
Beyond fraud detection, the technology is also being used to automate claims processing through a mechanism known as auto-adjudication. Earlier, claims were manually reviewed and often took 15 to 20 days for settlement. With AI-enabled auto-adjudication, straightforward claims that meet guideline requirements can now be processed in approximately two hours.
Story continues below this ad
The system does not eliminate human oversight. Routine and compliant claims are cleared automatically, while complex or suspicious cases are escalated for manual examination. “Earlier, the process was entirely human-driven. Now it is humans plus AI,” the official said.
The AI-driven system has been in operation since January 19 and is currently being scaled up across states. Data on fraud cases detected under the scheme is available on official dashboards, while vetted figures can also be accessed through formal communication with the National Health Authority.
The healing voice
The ministry further showcased an AI-based voice-to-text solution. For doctors, this includes a Health Information Management Systems (HIMS) system that allows them to generate digital prescriptions instead of handwritten ones. He pointed out that while digital prescriptions are an important step toward modernization, they also create practical challenges. Traditionally, doctors write prescriptions by hand during consultations. However, with the digital system, they must type out every detail. This can be inconvenient and time-consuming. Doctors often feel burdened by the additional documentation work and patients may feel that the doctor is focussing more on the computer screen than on them, affecting the quality of interaction.
To address this issue, the Ministry has developed a Voice-to-Text (VTT) solution. This AI-based tool converts a doctor’s spoken words into written text in real time. As the doctor speaks during the consultation, the system automatically transcribes the speech and fills in the prescription and patient recommendation sections within the HMI platform. The goal is to reduce the time spent typing and make the process more seamless.
Story continues below this ad
According to the official, this solution has already been integrated into the government’s next-generation eHospital platform developed by NIC and is being used across most government hospitals. However, it is not mandatory for doctors to use the feature. There is a microphone option within the prescription tab, and doctors can choose to activate it if they prefer voice-based documentation.
The official also explained that the idea for this tool emerged from feedback received from doctors. “Given India’s heavy patient load, many doctors see a new patient every one to two minutes, leaving very little time for detailed digital data entry. The VTT system was introduced to ease this burden, reduce turnaround time for patients, and improve overall efficiency in government healthcare facilities,” he added.
