The Journey of Building a Healthcare Business Intelligence Platform

How we started building our data analytics platform to help address Biopharma companies' business needs

The world of healthcare market access is highly dynamic and complex. Particularly in oncology and other specialty therapeutic areas, the complexity has only grown as policies evolve and patient care models shift to become more value-driven. In order for Biopharma companies to stay competitive, they need to adapt and become more data-driven in how they develop strategies around reducing barriers to access to drugs for patients. At Pulse Analytics, our goal is to build the next-level business intelligence (BI) platform that allows Biopharma companies to make smarter decisions around market access. To reach this audacious goal, we had to start with a fundamental question: what makes a great BI platform?

Core Components of a Great BI Platform

When we think of BI platforms, we often think of analytic dashboards with figures and charts. But a great BI platform is more than composing charts on a web page. At its core, it’s about telling a story with data and providing users self-service capabilities and insights to make critical decisions. Over the years, we’ve identified three core components to building a great BI platform:

1. Simplicity in Design

Delivering a great user experience is central to the success of building any product and for building a BI platform, it’s no different. In this context, delivering a great user experience involves presenting complex data with simplicity. We want to help our users answer their most pressing business questions which means that the data visualizations presented must be thought through carefully. We also want to present information in a layered approach and reveal granular details through drill-downs.

2. Quality Data

At the core of a BI platform is its data. Without data, the platform has no utility. To help our users make critical decisions, it is imperative that the data we collect and present is of high quality. This means that data must be assessed carefully based on the six dimensions of data quality: Accuracy, Completeness, Uniqueness, Timeliness, Validity, Consistency

3. Custom Report Capabilities

Insights are pointless if the information doesn’t lead to action. To make data-driven decisions, users must align with relevant stakeholders and communicate key findings. This means that our platform must allow exportable charts and have customizable report generation capabilities built-in.

Engineering Challenges

With these principles in mind, we had set our sights on building this platform from the ground up. The journey to building a custom BI platform that specifically addresses the market access needs of different Biopharma companies face was not easy. Here are several of the key engineering challenges we’ve faced throughout our journey:

1. Building Custom Data Visualizations

To tell a rich story about data, sometimes a simple pie chart is not enough. In particular, when we’re designing more complex views (e.g. analyzing a drug’s regional access information), we have to take on the role of a data visualization engineer and develop more customized data visualizations. How do we create visualizations that allow users to derive insights quickly? How do we introduce interactivity to ensure that the right amount of details are being shown?

2. Collecting Data

Our data comes from many different sources ranging from primary and secondary research to third-party APIs and websites. It’s crucial that we provide a platform for business analysts to input data and at the same time, acquire data from external sources in an automated fashion. How do we ensure that the data ingested is of high quality? How do we design a centralized layer to maintain data coming in from various sources?

3. Modeling Data

A BI system’s data model is core to the success of ensuring that data quality is maintained. Because both OLAP (Online analytical processing) and OLTP (Online transaction processing) systems are needed to power the entire platform, we are constantly thinking about the optimal data models for the different systems. How do we optimize a data model for heavy writes vs heavy reads? Given the domain we’re in, how do we appropriately apply the business constraints to our data models? What type of databases should we adopt (NoSQL vs. SQL vs. Graph)?

4. Building a Role-Based Access Control (RBAC) System

With different users, different companies, and different needs, we have to be able to build a robust permission control system that allows users to see the information that’s tailored for them. How do we determine the appropriate resources for users, teams, and clients? How do we manage the security risks and policies for each user?

5. Building a Reporting Layer

When we’re building custom data visualizations from the ground up, the export functionalities to various file formats need to be customized as well. Having report generation capabilities also means that we need a way for different front-end components to be composed into a report as well (i.e. PowerPoint Deck, PDF, etc.). How do we ensure the exports are standardized across the different types of data visualizations? How do we add customizability to the reports?

Conclusion

The journey has been both incredibly challenging and rewarding over the past few years. We’ve made a ton of mistakes and learned a lot along the way. To this date, we’re still facing some of the same challenges and are constantly looking for new ways to optimize and improve our existing systems. If these challenges sound exciting to you, check out our careers page, and come join our team!