Opportunity: Transform raw clinical data into trusted, workflow-ready decision support.
Risk: Build an AI engine clinicians wouldn’t adopt.
The Challenge
Clinicians make high-consequence decisions under time pressure, often navigating fragmented data and limited real-world applicability. Clinical trials rarely reflect the nuance of real patient populations.
For this product to influence behavior, insight had to be trustworthy, workflow-aligned, cognitively efficient, and measurably tied to patient outcomes.
Without that, we risked scaling a system no one would meaningfully use.
My Role
I led early discovery and product definition — establishing the validation framework, aligning stakeholders, and defining product principles before development scaled.
My focus was clarity before scale.
Strategic Approach
We did not begin with UI. We began with clinician decision-making.
If this system was going to influence care outcomes, it had to reflect how clinicians reason under pressure — not how we assumed they did. Through interviews, workflow analysis, and structured concept validation, we examined where decisions break down: how evidence is weighed, how risk is perceived, how time pressure distorts judgment, and where workflow friction overrides best intent.
Before investing in interface, we tested core assumptions. Could AI-driven insight meaningfully reduce cognitive load? Would projected outcomes increase trust? Would surfacing trade-offs improve decision quality, or simply introduce more complexity?
Adoption would ultimately depend on whether insight was contextual to the patient at hand, integrated naturally into EMR workflow, capable of demonstrating value quickly, and supported by evidence transparent enough to build trust.
This work clarified product principles, narrowed scope, and reduced strategic risk before engineering investment scaled.
Key Decision #1
Define decision archetypes, not generic personas.
Through synthesis of interviews and workflow analysis, we identified distinct decision archetypes among clinicians. Rather than segmenting by demographics, we segmented by decision behavior — how evidence is weighed, how risk is perceived, and where workflow friction enters.
Key Decision #2
Shift the product from prescription optimization to Total Care.
Clinicians do not think in medication lines — they think in care trajectories. The opportunity was not optimizing prescriptions, but supporting holistic care decisions: medication combinations, affordability, adherence, and long-term outcomes.
This expanded the product scope beyond prescriptions to the dimensions clinicians actually manage.
Notable Key insight - “Total Care”
Total care extends beyond prescription writing. It requires evaluating medication combinations, affordability, adherence, evidence strength, lifestyle impact, and long-term outcomes.
Clinicians don’t optimize drugs — they navigate trade-offs across the patient’s full trajectory.
Strategy
The product needed to integrate directly into clinician workflow, align with EMR systems, and support both individual patient decisions and population-level insights. My focus centered on the standalone clinician experience — ensuring insight was contextual, trustworthy, and actionable within time constraints.
This strategy established clear design principles: balance patient and clinician needs, make treatment trade-offs immediately discoverable, and ensure AI-driven insights were transparent, defensible, and trustworthy.
Evolved Product Direction
The reframed scope translated into a clinical decision interface grounded in contextual insight, explainability, and workflow alignment.
Information Architecture
The Information Architecture clarified how insight flows from search to decision — patient context, treatment options, evidence, projected outcomes — within a single decision surface.
User Flows & Wireframes
Core flows prioritized discoverability, trade-off clarity, and progressive disclosure of evidence.
Design
The primary decision surface centers on the patient record, surfacing A1C trajectory, current treatment context, projected outcomes, cost implications, and formulary alignment in one unified view.
Clinicians can compare treatment scenarios over 3–6 month projections, view trade-offs instantly, and access supporting evidence without leaving workflow.
Evidence & Validation
Direction was validated through iterative clinician testing and structured feedback
Design Principles that Shaped the System
Affordability & Cost Transparency - Surface total patient cost, co-pay implications, assistance programs, and long-term affordability at the moment of decision. For many clinicians, this was the primary constraint.
Adverse Outcomes & Risk Visibility - Expose projected side effects, renal impact, blood pressure implications, weight changes, and secondary risks before medication selection.
Trade-off Clarity - Make efficacy, quality of life, side effects, cost, A1C trajectory, renal impact, and medication burden comparable in a single decision surface.
Formulary Alignment - Indicate whether a treatment is in or out of formulary and surface associated friction, cost impact, and administrative burden in real time.
Evidence Transparency - Clearly communicate the data source and level of evidence supporting each recommendation to build clinician trust.
Patient Compliance & Behavior - Highlight adherence risks and support clinician–patient interaction through reminders, behavioral considerations, and follow-up visibility.
Outcomes
This tool helped clinicians improve health outcomes for their patients and reduced costs of care by $70 million annually for Anthem. In addition, there was a 55% reduction in care gaps and improved patient compliance in multiple areas including wellness, prescriptions, and preventive care.
