When AI can predict the cost of care: A new approach to more transparent healthcare
By Dhruv Rastogi, Chief AI Officer, Medi Assist
Healthcare is generating more information than ever across the care journey, from medical records and treatment histories to claims, eligibility and hospital documentation. The opportunity now is to connect these information points in ways that make healthcare more predictable, transparent and responsive.
India is already building the digital infrastructure to support this shift. The Ayushman Bharat Digital Mission has crossed 90 crore ABHA accounts, with more than 100 crore health records linked as of May 2026. The significance of this progress is not just its scale. It points towards a healthcare ecosystem where information can move more securely and consistently across patients, providers and other participants, with consent.
From cost uncertainty to greater visibility
This matters because the financial experience of healthcare is still difficult to anticipate. The latest National Statistical Office health survey found that the average out-of-pocket medical expenditure for a hospitalisation in India was about ₹34,064 in 2025, with significant variation depending on where treatment was received.
The final cost of care depends on several variables – the hospital, treatment pathway, room category, length of stay and the specifics of an individual’s coverage. Much of this information exists across different systems, but it is not always brought together at the point when a decision needs to be made.
This is where AI can play a more useful role. Instead of looking at claims only after treatment, AI can bring together information from treatment pathways, clinical documentation, claims history and coverage to provide an earlier view of the factors that may influence cost. This does not mean predicting an exact bill. Healthcare is inherently variable, and a useful prediction needs to make that uncertainty visible. The value lies in helping patients, providers, insurers, and benefits administrators understand the likely drivers of cost while there is still time to act on that information.
Claims are becoming a source of intelligence
Claims have traditionally been treated as an administrative endpoint: care happens, documents are submitted, and the claim is assessed. With better data connectivity and AI, claims can become a source of intelligence about what is happening across the healthcare journey.
When interpreted alongside clinical, eligibility and utilisation information, claims can reveal patterns across the healthcare journey. They can show where processes are working well, where friction is recurring and where a case may require closer attention. This moves claims from being simply a record of what happened to becoming one source of intelligence for what should happen next.
India’s digital health infrastructure is moving in the same direction. The National Health Claims Exchange is being developed to standardise claims processing and enable interoperable exchange of health claims information across hospitals, insurers and patients. This kind of infrastructure matters because AI is only as useful as the context available to it.
For technology leaders, the distinction is important. AI does not become more useful simply because more data is available. It becomes more useful when information from different systems can be connected, interpreted in context and brought into a decision at the right time.
From prediction to better decisions
This creates an opportunity to rethink how technology supports healthcare administration. Instead of using AI only to automate individual tasks, organisations can use it to identify patterns and surface potential issues earlier.
A documentation gap, an unusual claim pattern or a workflow bottleneck can be identified before it becomes a larger problem for a patient or provider. The role of technology then shifts from simply processing what has already happened to helping the ecosystem respond earlier.
This is where the broader health benefits experience comes into focus. Healthcare is not a series of isolated transactions between a patient and a hospital. It involves patients, providers, insurers, employers and benefits administrators, with information moving between them at different points in the journey. Connecting these touchpoints can make the experience more predictable and reduce avoidable friction.
The human layer still matters
More predictive healthcare does not mean removing human judgement. A prediction is an input into a decision; it should not automatically become the decision itself.
That makes explainability, accountability and human oversight essential. When an AI system identifies an anomaly or produces an estimate, the people responsible for the outcome should be able to understand the factors behind that signal. Complex or consequential cases should retain a clear path to human review.
The quality of healthcare AI will therefore be measured by more than model accuracy. It will also depend on whether its recommendations can be understood, challenged and acted upon responsibly.
Building an anticipatory healthcare experience
The larger opportunity is to move healthcare administration from reactive processing towards earlier, more informed decision-making.
The strongest application of AI in healthcare will not necessarily be the one that automates the most tasks. It will be the one that improves the quality and timeliness of decisions across the ecosystem.
Greater cost visibility can help people plan with more confidence. Better contextual information can help providers and healthcare administrators address issues earlier. More intelligent use of claims and other healthcare information can help insurers and benefits administrators respond to patterns rather than isolated events.
The goal is not to predict every rupee of healthcare expenditure. It is to reduce avoidable uncertainty through better-connected information, earlier insight and more transparent decision-making, while keeping human judgement and accountability at the centre.