Key takeaways

• Patient support is now data-rich but still lacks a connected view. Hubs, specialty pharmacies, co-pay programs, field teams, call centers and digital portals all generate valuable signals, but fragmentation prevents teams from seeing the full patient journey.
• AI-ready patient engagement starts with a trusted data foundation. Organizations need compliant integration, tokenization, patient mastering, consent-aware governance and reusable data products before AI can reliably support action.
• The future of patient services is anticipatory. Connected data and AI can help detect risk earlier, trigger the right intervention faster and support patients and providers before barriers lead to therapy abandonment or discontinuation.

Patient support programs generate vast amounts of data, yet many life sciences companies still struggle to translate information into timely, meaningful action. Despite significant investments in patient support programs and digital engagement, outcomes remain stubbornly slow. According to ZS’s 2026 Future of Health Report, up to one-third of patients never start their prescribed treatment, and 30% of patients say diagnostic delays worsened their condition.

As healthcare shifts toward more proactive and personalized models of care, the challenge is no longer data collection. It’s creating a connected foundation that enables organizations to understand and respond to the full patient journey.

ZS principals Asheesh Shukla and Kirti Prakash discuss forces reshaping patient engagement, from real-time data and AI-enabled decision-making to rising expectations around privacy, consent and trust. They argue organizations must move beyond isolated systems and retrospective reporting toward connected, action-oriented data models that support faster intervention and more personalized experiences.

Below are highlights from the conversation. Watch the full podcast here.

Moderator: Patient services generate an enormous amount of data across the ecosystem. Where is all this data coming from, and what is the state of it today?

Kirti Prakash (KP): Patient services have quietly become one of the most data-rich functions in pharma. Data is now generated across hubs, enrollment, benefits verification, specialty pharmacy dispensing and refill tracking, co-pay and patient assistance programs, call centers and other patient touch points.

That should give organizations a much clearer view of the patient journey. In reality, the data often sits across multiple vendors and systems in inconsistent formats. The hub may have one view, the specialty pharmacy another, and call center notes may contain useful patient context that no one else can see. A patient may be flagged as high risk in one system, while the case manager in another system has no visibility into that signal.

The data is abundant, but what organizations are dealing with is poor decision quality. The cost of that gap is significant. Roughly one-third of prescriptions in the U.S. are never filled, and up to half of patients who start therapy discontinue within the first year. Together, these gaps drive an estimated $500 billion in avoidable costs. Research shows that about 53% of these issues are addressable if organizations can intervene at the right time to help both patients and treating physicians. The problem was never a shortage of data. It is that the data has never been connected.

Asheesh Shukla (AS): The challenge is becoming even more urgent as patient services become more personalized and more closely tied to each patient’s treatment journey. Provider platforms are becoming more connected, interoperability expectations are rising, and privacy and consent safeguards are becoming more important. Combined with recent research from our ZS Future of Health Report, we are also learning that patients will trust an AI agent for medical advice with up to 90% confidence.

At the same time, patients are showing more willingness to share health data when it delivers value and helps them manage their disease with providers. That means the volume of available patient data will continue to grow, but unless organizations improve decision quality and act on the data, the gap will only become more visible.

Moderator: What is fundamentally changing in how data is generated, consumed and trusted, and in how pharma engages with patients?

AS: Four shifts are changing the way organizations need to think about patient data.

The first is a healthcare mindset shift from reactive illness management to more proactive and predictive care. Historically, organizations processed patient data periodically, often in batch cycles. But if the goal is timely intervention, teams need a more continuous relationship with patient data.

The second shift is the move from insight to intervention. Patients and providers need help facing barriers such as prior authorization, benefit verification or affordability challenges. Increasingly, humans and AI agents will work together to support those interventions, which means data can no longer be built only for dashboards and retrospective reporting. It must become execution-ready and action-oriented.

The third shift is consumption. Patients, providers and internal teams increasingly expect insights to appear where they already work, rather than logging into separate portals or dashboards. Large language models (LLMs), co-pilots and multisystem agents are changing how people consume information, but those tools are only as effective as the underlying data and semantic context behind them. Research shows that 91% of people will go to an AI chatbot for medical advice, and what drives that advice is the underlying data. In practice, we routinely see these agents failing to live up to the expectations of both patients and providers, largely because the underlying data lacks the semantic context needed to make a response relevant to the individual’s situation rather than a generic answer given to a million other patients. That lack of semantic context is the fundamental gap, and it means more data needs to be structurally ready, so these new consumption patterns can deliver value.

The fourth shift is trust. In healthcare, trust is a gating factor. If patient data is not well governed, clearly defined and compliant, AI agents can compound risk instead of reducing it. Consent, preference management and compliant data use need to be designed into the foundation from the start.

Moderator: Given this change, how should organizations respond? What does getting ready look like?

KP: For pharma, readiness comes down to two connected imperatives, with compliance embedded in both. The first is building a strong, connected and compliant data foundation. That means linking patient-level data across support programs, hubs, specialty pharmacies, co-pay use, medical records and other sources into a longitudinal view of the patient journey.

But integration alone is not enough. Organizations need tokenization, patient mastering, deidentification and compliant storage in appropriate environments. They also need governance, consent language, consent scope and protected health information (PHI) masking rules that are built into how the data is accessed and used.

The second imperative is the last-mile action. It is not enough to have a clean data core if the engagement itself is not compliant. Case manager outreach, field reimbursement interventions, nurse educator touch points and digital messages all need to operate within the same governance and consent boundaries.

The harder shift is not technology. It is the way of working. Teams need clear ownership for each nudge, a governance and consent model and a way to measure impact. Organizations are starting to move from one-off use cases to repeatable capabilities that help them connect data and act on it more consistently.

Moderator: For a head of patient services or a technology leader supporting patient data, what are the no-regret moves they can start with now?

KP: One no-regret move is to structure patient data as data products. Instead of trying to serve every use case from one giant table, organizations should create reusable layers of data. Foundational data products can capture integrated transaction-level activity across hub, co-pay, specialty pharmacy and other sources. Functional data products can organize enterprise-standard definitions, key performance indicators (KPIs) and business rules around areas such as patient journey, access and affordability, and trade channels.

From there, fit-for-use-case data products can support specific decisions and interventions. Increasingly, organizations also need a fit-for-AI layer that provides the semantic context agents need to read, reason and act more effectively.

The second no-regret move is to be explicit about the personas being activated. Whether the user is in patient care analytics, field reimbursement, patient marketing, case management or another role, the data foundation and last-mile design should be built around how that person will use the insight. Leaders should start unifying patient data now rather than waiting for a perfect data lake or warehouse.

AS: I would add two strategic moves. First, organizations should take a fit-for-purpose approach based on where each asset sits in its life cycle. A prelaunch rare-disease asset requires a different data and intervention model than an established brand. The barriers patients, providers and payers face will vary by product, therapy area and competitive context.

Another strategic element is to start consuming this as a service rather than building from scratch. Every time a new asset launches, organizations rarely have the time or ability to absorb the innovation happening elsewhere. As AI agents rapidly advance, consuming this as a service lets teams move faster without having to rebuild capabilities each time.

Second, organizations should make provisions to own the data, even when parts of patient services are outsourced. Patient services often rely on third-party operators, nurses, field reimbursement teams and other partners. But if manufacturers own the data, they have a stronger ability to own the experience, measure impact and improve how patients are supported over time.

Fundamentally, it comes back to the patient. The goal is to reduce barriers, help eligible patients start therapy sooner and prevent nonclinical reasons for discontinuation when support is available. Done well, this is both the right thing to do for patients and providers, and a sustainable model for the business.

Moderator: Looking two to three years ahead, what will be fundamentally different, and what capability can no organization afford to skip?

AS: Patient support will become genuinely anticipatory. AI will help sense risk, predict barriers and trigger the right intervention before patients and the physicians treating them hit a barrier. That is one of the best services organizations can provide to both patients and providers because it helps them manage disease more effectively over time.

The non-negotiable capability is a connected, compliant and AI-ready data foundation. Organizations need to stay connected, stay compliant and make patient data ready for AI-enabled consumption and intervention, with trust built into the services they deliver.

KP: I agree. The patient relationship and the data behind it will become more important. Manufacturers will need greater control and stronger custodianship of patient data, with built-in consent and less fragmentation across vendors.

It will also be critical to measure and prove the impact. Whether through causal measurement, test-and-control approaches or other methods, organizations need evidence to defend investments and scale what works. Without measurement, it becomes much harder to sustain and expand these capabilities.