Transforming Pharma Analytics with Systems Thinking/ Modeling and Agentic AI

In the rapidly evolving world of data and analytics, technological advancements have predominantly focused on two areas: descriptive analytics (understanding what happened, and when) and predictive analytics (for forecasting what will happen).

Most technological advancements have continued to solve these two ends of the spectrum. For example, allowing natural language-based data interrogation, auto visualizations, auto ml, built in ml functions for prediction etc.

However, a critical gap is still—diagnostic analytics (understanding why something happened).

For business leaders, answering even seemingly straightforward questions such as "What drove pharma product sales last quarter?" can be a complex, multi-faceted challenge. This is where systems thinking, causal loop modeling, and agentic AI can bridge the gap, enabling organizations to uncover deeper insights that traditional analytics often overlook.

The Challenge of "Why" in Pharma Sales

Pharmaceutical sales are influenced by a dynamic interplay of multiple factors: physician prescribing behaviors, direct-to-consumer (DTC) advertising, payer and formulary access, competitive pricing strategies, patient adherence, and regulatory policies.

Unlike descriptive and predictive analytics, which focus on isolated events or trends, diagnostic analytics require correlating multiple data sources and daisy-chaining questions to construct a logical narrative.

For instance, if a particular drug's sales surged in a specific region, the root cause could be a combination of:

• Increased marketing spend leading to higher patient awareness.
• Changes in physician engagement via sales representatives.
• Competitor stockouts or pricing changes affecting market share.
• Regulatory changes or formulary updates improving patient access.
• Therapy adherence improvements due to enhanced patient support programs.

Finding the precise reason requires pulling data from disparate systems, aligning them contextually, and synthesizing insights across domains. This is where systems thinking, and causal loop diagrams become powerful tools.

Using Systems Thinking to Unlock Diagnostic Analytics

Systems thinking offers a structured way to analyze complex interactions by mapping out feedback loops and interdependencies. A causal loop diagram (CLD) helps visualize the cause-and-effect relationships that drive pharma sales and marketing dynamics. For example the following broad causal loops

Causal Loops in Pharma Sales & Marketing

🔁 Reinforcing Loop (R1) – Sales Force Effectiveness Increased sales reps → More physician engagement → Higher prescriptions → Increased revenue → More investment in sales force expansion → Further increases in sales reps.

• Example Data Product: A "Physician Influence Map" that uses engagement history, prescribing trends, and network analytics to optimize rep targeting strategies.

⚖ Balancing Loop (B1) – Regulatory Constraints Increased marketing spend → Higher regulatory scrutiny → Stricter compliance rules → Reduced marketing effectiveness → Fewer prescriptions → Adjusted marketing spend.

• Example Data Product: A "Compliance Sensitivity Model" that analyzes historical compliance interventions, marketing expenditures, and risk exposure.

🔁 Reinforcing Loop (R2) – DTC Advertising & Patient Engagement Higher DTC advertising spend → Greater patient awareness → More patient requests for prescriptions → Increased physician prescriptions → Higher revenue → More DTC ad spend.

• Example Data Product: A "Patient Demand Signal Analyzer" that integrates search trends, social media sentiment, and prescription requests to measure advertising effectiveness.

⚖ Balancing Loop (B2) – Payer Access & Formulary Restrictions Increased prescriptions → Higher payer costs → Stricter formularies and restrictions → Fewer prescriptions covered → Lower sales growth → Companies adjust strategy.

• Example Data Product: A "Market Access Leverage Engine" that models payer contract strategies, formulary tier changes, and reimbursement shifts.

🔁 Reinforcing Loop (R3) – Therapy Adherence & Patient Support Enhanced patient adherence programs → Better medication compliance → Improved health outcomes → Higher long-term drug utilization → Increased revenue → Further investment in patient support.

• Example Data Product: A "Patient Adherence Intelligence Hub" that combines pharmacy refill data, patient support interactions, and real-world evidence to improve adherence strategies.

🔁 Reinforcing Loop (R4) – Out-of-Pocket Costs & Medication Abandonment Higher patient out-of-pocket costs → Increased medication abandonment → Lower therapy persistence → Reduced sales → Further restrictions on access programs.

• Example Data Product: A "Financial Assistance Optimization Model" that predicts patient abandonment risk based on co-pay levels, income demographics, and benefit design.

Introducing Agentic AI: Reducing Time to Value in Diagnostic Analytics

To enhance the value of these data products and reduce the time to actionable insights, agentic AI can play a crucial role. Instead of relying on manual analysis, a network of specialized AI agents can work collaboratively to automate and accelerate diagnostic analytics.

How Autonomous AI Agents Work Together

• Signal Detection Agent – Continuously monitors the data products for changes in functional microsegments/ metrics
• Causal Analysis Agent – Maps changes in key metrics to known causal loops and finds potential root causes.
• Scenario Simulation Agent – Runs simulations to evaluate hypotheses and predict downstream effects.
• Recommendation Agent – Provides data-driven recommendations tailored to business stakeholders.
• Validation Agent – Cross-checks AI-generated insights against historical patterns and expert inputs to ensure accuracy.

For example, if an AI agent detects a sudden drop in therapy adherence, it can trigger an automated workflow: the Causal Analysis Agent checks if a co-pay increase or formulary change played a role, while the Scenario Simulation Agent predicts how different intervention strategies might restore adherence rates.

Ensuring Transparency and Accountability in AI-Driven Systems

At some point of time in our lives, we will have to trust what the machine tells us from a decision support perspective. The biggest problem is that of accountability. When a person gives an answer that is incorrect, he/ she can be held accountable. However, when a machine gives an incorrect answer, where does the accountability lie? Does it lie with the developer, the systems integrator, the platform owner, the technology partner or all of the above?

A combination of human in the loop along with adherence to the following principles will ensure transparency and accountability:

• Explainable AI (XAI) – Every AI-driven insight must be interpretable, showing the reasoning behind conclusions. Reasoning based LLM models make this a reality
• Human-in-the-Loop Oversight – Critical insights require validation from domain experts before being acted upon. A parallel team to cross validate the insights and fine tune the chain of thought
• Feedback Loops for Continuous Learning – AI models should adapt based on real-world performance and expert feedback.
• Data Provenance and Auditability – Every AI-generated insight must be traceable back to source data and processing steps.
• Clear Governance Frameworks – Ownership must be defined for AI-driven recommendations, ensuring accountability at every level.

"Growth makes luck look like strategy"

The future of pharma commercial analytics isn’t just about knowing what happened or predicting what will happen next, it’s about understanding why things happen and leveraging agentic AI to deliver rapid, accurate, and trustworthy insights. This will enable business leaders to take charge and make confident decisions instead of gingerly continuing with status quo.

By integrating systems thinking, causal loop modeling, and autonomous AI agents, pharma companies can move beyond reactive decision-making and embrace a holistic, AI-powered approach to commercial success.