By Jo Ann Saitta

The life sciences industry is generating unprecedented volumes of real-world evidence (RWE) data, from genomics to claims to electronic health record data, yet it struggles for real-time, actionable insights. This data volume, coupled with fragmented storage in legacy silos, creates a strategic liability.

Data often lacks the quality, consistency and integration needed for enterprise-wide use and AI solutions. Consequently, many sophisticated AI models fall into the ‘proof of concept (PoC)’ trap, failing to progress from PoC to scalable implementation due to a lack of reliable data. The solution to this issue requires building a trusted, integrated data foundation and embedding scalable, responsible AI into critical workflows that accelerate go-to-market execution and deliver strategic advantage.

The starting line for any successful AI journey is data readiness and quality. Before machine learning (ML) or generative AI (Gen AI) can deliver value, data quality and fragmentation issues must be resolved. Scattered, inconsistent data fundamentally erodes trust in the inputs/outputs for AI.

One effective strategy is to unify disparate structured and unstructured sources, such as claims, genomic, lab and electronic health records (EHR), into a unified data fabric. This foundational work involves semantic mapping, harmonisation and quality-checking at scale. This unification enables advanced analytics, such as complex, longitudinal patient journey mapping, moving beyond simple cohort analysis. Platforms like Navigator AI and DataGateway help teams to:

• Transform unified data into actionable insights by applying clinical business rules to answer key strategic questions
• Uncover pivotal treatment moments, such as a diagnosis or medication switch, missed by siloed analysis
• Map the granular patient journey precisely and find patients by identifying intervention points and diagnostic bottlenecks.

While earlier AI focused on basic natural language processing (NLP) and machine learning (ML) for classification and simple predictions, today’s opportunity lies in Gen AI solutions and AI agents, representing a disruptive shift and designed for autonomy. The future includes agentic solutions that can perform complex, multistep reasoning, rapidly synthesise heterogeneous RWE and even initiate actions based on human-defined rules. These capabilities bring a new level of impact across the treatment journey, simplifying provider workflows, reducing system friction and lowering barriers to diagnosis and access.

One example of this is patient finding. In rare diseases, AI patient discovery agents will scan and synthesise unstructured data and correlate it with structured sources across massive data sets, such as EHRs and claims data, to find patients that match a disease fingerprint faster and notify field professionals and associated healthcare professionals (HCPs) once confidence thresholds are met. This will drastically reduce the time to diagnosis. This has the potential to move a diagnosis from years to weeks.

AI is now a strategic driver for business transformation, redefining the commercial landscape by shifting from mass outreach to micro-segmentation and hyperpersonalisation:

• Commercial strategy and patient engagement
  AI elevates the next best action (NBA) to predictive precision, anticipating HCP needs and proactively predicting patient adherence challenges, triggering personalised interventions and enabling hyperpersonalised content creation.

This area is poised for one of the greatest revolutions in our industry. AI and its future agentic solutions can synthesise RWE and regulatory data at unprecedented speeds, drastically cutting the cycle time for regulatory responses. Medical Affairs teams use AI platforms like ClarityNav to distil complex and large volumes of multimodal sources of unstructured data, such as congress summaries, advisory board insights or scientific narrative inputs, into actionable guidance. With a solid RWE data foundation, AI synthesises unstructured data into structured outputs that link to claims and other data sets to recommend guidance and measure impacts to care gaps over time.

In an industry where decisions directly impact patient safety, responsible AI is vital. It means ensuring every AI system is effective, innovative, compliant, safe, equitable and transparent. This requires a human-in-the-loop framework for necessary oversight, accountability and the ability to audit and correct automated outputs. Key considerations include:

• Privacy and compliance
  Ensuring models adhere to strict legal boundaries (eg, General Data Protection Regulation, Health Insurance Portability and Accountability Act) through data anonymisation and maintaining model transparency for auditability
• Ethical considerations
  Actively auditing models to prevent inherent biases that could lead to disparate health outcomes based on demographics.

Jo Ann Saitta is Partner, Global Data Strategy, Analytics and AI Practice Lead at Putnam, Inizio Advisory