Led design of Komodo Health’s analytics platform to support technical and non-technical users. Launched with 29 accounts in 10 months and accelerated time-to-insight across user types.

Introduction

MapLab is a powerful healthcare analytics platform developed by Komodo Health to enable a broad spectrum of users—from commercial analysts to data scientists—to rapidly derive meaningful insights from complex healthcare data. The core mission of MapLab is to democratize access to healthcare data insights, empowering users to answer critical business questions swiftly without needing advanced technical skills.

Problem Statement

Healthcare data is notoriously complex and traditionally accessible only to highly technical users. The challenge:

  • Create a single platform that serves two distinct user groups:

    • 🟦 No-code/low-code users (e.g., commercial analysts) needing quick, self-service insights

    • 🟧 High-code users (e.g., data scientists) requiring detailed data access for custom analyses

  • Support data refreshes, external system integrations (CRMs, BI tools), and governance/compliance

  • Make the experience intuitive while managing the overall complexity and discoverability

My Role

I served as Design Lead and Individual Contributor on MapLab, owning design for several mission-critical features:

🔹 Interactive Dataflow Canvas

🔹 Data refresh and export workflows

🔹 Resource sharing & governance

🔹 AI-driven features: Codeset recommendations and Cohort Preview cards

I collaborated closely with PMs, engineers, data scientists, and user researchers to define and execute the UX strategy across these areas.

Cohort preview card

Cohort Preview card designed to display high-level information at a glance.

Users & Personas

MapLab supports a wide range of users across technical skill levels and departments:

🟦 Low-code/No-code Users

  • Commercial analysts, market access teams, medical affairs

  • Prefer intuitive UI and visual workflows

🟧 High-code Users

  • Data scientists, technical analysts

  • Require direct access to underlying tables and advanced customization

User Segments

  • Commercial

  • Clinical Development

  • Medical Affairs

  • Market Access

  • BD&L (Business Development & Licensing)

  • Financial Services

  • HEOR (Health Economics Outcomes Research)

Commercial Analyst Persona

"My brand lead just asked me why our drug's uptake is slow in the Midwest. I don't have time to write a query; I need a dashboard where I can see the market landscape, compare it to competitors, and identify the key drivers in minutes."

Key Insights:

  1. Prioritize Self-Service and Speed. This user needs answers to business questions immediately and cannot wait for a data team. The product must be an intuitive, self-service platform. Pre-built dashboards, simple filtering controls, and a clear, uncluttered interface are essential. Their primary measure of success is "time to insight."

  2. Design for Business KPIs. This user thinks in terms of market share, patient starts, prescriber volume, and competitive threats. The design must surface these Key Performance Indicators (KPIs) clearly and prominently. The goal is to translate raw data into actionable business intelligence. Big numbers, trend lines, and clear data visualizations are paramount.

  3. Enable Easy Drill-Down. While they need a high-level overview, they must also be able to answer "why." The user experience should be built around an intuitive drill-down capability. A typical workflow would be viewing a national trend, clicking to see regional data, then clicking again to see data for a specific prescriber group. The interaction model should be built on filtering and exploring, not building and defining.

Clinical Development Persona

"A phase 3 trial can cost a billion dollars. I need to rapidly model different inclusion/exclusion criteria using real-world data to ensure my trial design is feasible and we can actually find the patients we need."

Key Insights:

  1. Optimize for Rapid Exploration. This user's primary goal is "what if" analysis for trial feasibility. The design must allow them to quickly change parameters (e.g., add or remove an exclusion criterion) and see the impact on the patient count in real-time. Speed and interactivity are more important than deep analytical complexity.

  2. Simplicity Over Granularity. This user is a medical or scientific expert, not a data scientist. They need high-level, directional insights, not perfect datasets. The UI should hide the underlying data complexity and present clear, easy-to-understand visualizations and summary statistics (e.g., patient counts, demographic breakdowns, location heatmaps).

  3. Focus on the Patient Journey. A key pain point is understanding how a trial protocol will work in the real world. A designer should prioritize features that allow this user to visualize and explore real-world patient journeys—what tests they get, what drugs they take, and in what sequence. This context is critical for designing a trial that will succeed.

HEOR / RWE Persona

“I spend months building analysis-ready datasets. I need a tool that accelerates this process but still gives me the methodological transparency required to defend my findings to regulators and payers.”

Key Insights

  1. Methodology is Everything. This user must be able to see, control, and justify every step of their analysis. The product cannot be a "black box." A designer must prioritize features that allow for granular control over parameters, clear data lineage and the ability to export detailed methodology for publications and regulatory submissions.

  2. They Build from Scratch. Unlike other users who consume pre-built dashboards, the HEOR persona builds complex patient cohorts and analytical logic from the ground up. The user experience must be optimized for this "builder" mindset, providing powerful, flexible tools that can handle complex logic (e.g., patient A must have event X but not event Y, then must have event Z within 90 days).

  3. Efficiency is a Core Need. This user's current workflows in code are powerful but slow. The primary value proposition for them is speed and efficiency. The design should focus on streamlining the most time-consuming tasks: defining cohorts, creating variables, and comparing populations, without sacrificing the scientific rigor they require.

Process

We employed a human-centered, iterative design process, including:

  • User research and usability testing to validate direction

  • Cross-functional workshops to align on workflows and priorities

  • Rapid prototyping to explore solutions for AI, data visualization, and interactivity

  • Close partnership with engineering to ensure feasibility and maintain performance

Solution Highlights

Sharing & Governance

  • Seamless resource sharing

  • Compliance with data governance standards

Dataset Refresh & Export

  • Customizable dataset refresh controls

  • Integrated export functions for downstream tools (e.g., Salesforce, Tableau)

AI Chat Experience

Natural language tools offering:

  • Codeset recommendations

  • Cohort Preview cards

Dataflow Canvas

A visual drag-and-drop interface that allows non-technical users to build analytics workflows through microinteractions.

Featured
Sharing
Sharing

To enhance data privacy and collaboration within the MapLab platform, I designed a new resource sharing model that transitioned the system from being public by default to private by default.

This "Share" modal is a key component of the new framework, giving resource Owners explicit control over who can view or edit their work. The design provides granular, user-level permissions (Owner, Editor, Viewer), a simple toggle to share with an entire account, and intelligent "upstream auto-sharing" to ensure that dependent resources are automatically shared for a seamless user experience. This solution empowers users with control over their work, meets critical compliance requirements, and facilitates focused, effective collaboration.

Export
Export

To make MapLab a more integrated part of our users' broader data ecosystem, I designed the platform's Export functionality.

The design, showcased in this modal, provides a clear and powerful workflow for sending datasets to external destinations like Snowflake. Recognizing that users have different needs, I incorporated options for both immediate "One-Time" transfers and automated "Scheduled" exports for recurring data pipelines. This feature is governed by the platform's permissions model, ensuring only authorized users (Editors and Owners) can move data, thereby enhancing both security and interoperability.

Codeset Recommendation
Codeset Recommendation

To ensure users could confidently adopt generative AI for complex medical analysis, my core design principle for MapAI was transparency. For users to trust an AI-generated recommendation, they must understand its reasoning.

This interface is designed to provide a "glass box" experience. Instead of just delivering a final answer, MapAI explains its step-by-step process: it confirms the user's intent, details the exact search parameters used, clarifies adjustments made to the query, and justifies why a new codeset was drafted. This "show your work" methodology is vital for building user trust, as it empowers clinical and research professionals to quickly validate the AI's logic and feel secure integrating its outputs into their workflows.

Dataflow Editor
Dataflow Editor

To make complex data preparation accessible to a wider range of users, I designed the Dataflow editor, a visual canvas for building and managing data pipelines within MapLab.

This node-based interface allows users to intuitively connect different processing steps—from reading a cohort definition to outputting a final dataset—abstracting away the need for complex backend code. The design not only simplifies the creation of sophisticated workflows but also provides a clear, visual data lineage. This enhances reproducibility and makes the entire process easier to debug, validate, and securely share with colleagues using the platform's granular access controls.

Impact

Launched in November 2023, MapLab quickly gained traction:

  • 29 active accounts onboarded in 10 months

  • Strong engagement across both technical and non-technical user groups

My design contributions led to:

  • Increased workflow accessibility

  • Improved discoverability through AI-first UX

  • Stronger user trust via clear and consistent design patterns

“Komodo has significantly impacted my business by enhancing our data analytics and insights capabilities. Since integrating Komodo, I have seen a 20% improvement in data accuracy and a 15% increase in operational efficiency.”
— Senior Director at a Top-10 Pharmaceutical Company

Learnings

This project reinforced the importance of:

  • Flexible design systems that can scale across user types

  • Layered complexity to support both novices and experts

  • Cross-disciplinary collaboration as a lever for innovation

It deepened my expertise in:

  • Healthcare data visualization

  • AI-assisted UX design

  • Enterprise-scale product thinking

Previous

z/OS Management Services Catalog