Problem solved: How we built a trusted data platform that produces accurate healthcare insights

Part 2 by Vinod Subramanian

2023/01 6 minutes read

Social Media Blog Graphics 16

In part one of this three-part series, I shared the key differences between a data platform and a healthcare data platform. Now, I’m going to dive deeper into why interoperability, data normalization and knowledge management - along with AI - are critical capabilities that must be built into a healthcare data platform. And how these capabilities transform disparate real-world data into data that can be used to power the work of life sciences, clinicians and researchers. 

Before we go on, let’s quickly align on key definitions. 

  • Interoperability is the ability of computer systems to exchange and make use of data and information. What it’s not: only data ingestion 
  • Normalization is the process of data organization
  • Knowledge management transforms complex and inconsistent data into easily-understood information

These definitions are simple, however, the process is complex. Let’s explore this further.

How do interoperability, data normalization and knowledge management enable the transformation of real-world data into real-world evidence? 

It’s not enough to extract and ingest vast amounts of multi-source data. Where real-world data in healthcare is concerned, information needs to be ingested but also shared easily and seamlessly between any number of disparate systems across the technology spectrum. This is where true interoperability comes into play. 

Where RWD is concerned, extracting and ingesting data isn't enough. Healthcare data needs to also be easily and seamlessly shared across any number of disparate systems. This is where interoperability comes into play.

Data is not only derived from multiple sources but exists in multiple formats (different naming conventions for therapies, patient designations, even dates & times) normalization and a proven knowledge management solution are critically important in reducing ambiguity and making the data usable. 

Interoperability is the enabler. By accessing and exchanging multi-nodal real-world data, it speeds the transformation to real-world evidence, which can then be used by health systems, life science companies and regulators to improve their understanding of every aspect of patient care, and turn that learning into action.  

Normalization and knowledge management are the tools. They make it possible to aggregate and validate immense quantities of widely disparate data, interpret that data accurately, then deliver it in consistent and easily-understood formats. All of which can only be achieved through deep clinical understanding, data science expertise and teams of health informaticists—features that should be inherent in a healthcare data platform—you’re not likely to find in a more standard platform. 

Being able to extract, share, validate and trust data, from multiple sources and diverse systems, is essential to the transformation of real-world data into real-world evidence. And if the intent is to advance patient care, interoperability, normalization and knowledge management are the amplifiers of that intent. 

How does AI impact healthcare data and real-world data more specifically? 

COVID-19 was the catalyst for change in how we think about using healthcare data to advance patient care. The historically rapid development of vaccines, antiviral medications, and even accurate and reliable tests would not have been possible without real-world data coupled with AI to interpret that data in new and innovative ways. 

It’s now commonplace to think of AI as playing a central role in the future of care innovation and delivery. 

As with other features of healthcare data platforms, the productive use of AI is very much guided by our ultimate intent, which is to help patients overcome life-threatening disease. 

It’s also determined by the quality of data we feed into the tool. With a platform built for healthcare, real-world data from multiple sources—structured, semi-structured and unstructured—gives AI a powerful foundation with which to work its predictive magic, identifying complex patterns in the data, like finding myelodysplastic patients. And determining the probability of cancer or any other ailment before they actually happen, like predicting metastatic status.

In short, AI is only as good as the data we give it. It would be difficult to imagine a traditional platform delivering high-quality real-world data at scale, given the complexities of a healthcare environment.  

AI can only be accurate in predictions and identifying complex patterns if the data we give it is of the highest quality. 

I’ve identified five areas where AI is currently being used to great effect in healthcare. 

Early Detection of Disease—providing insights into what is to come based on what has already occurred in a patient’s life. 

In Clinical Trials, AI can help trial coordinators more accurately identify the right candidates at the right time in their patient journeys. 

AI can augment and improve Natural Language Processing, by accelerating the flow of unstructured data such as hand-written physician notes and pathology reports into the data stream. An example would be using NLP to extract complex biomarker information for patients with cancer. 

We’re seeing AI increasingly being applied to Data Normalization, by helping to organize widely disparate, clinically complex and highly contextual data types into common terminologies that healthcare professionals can use to advance their life-saving work. 

And finally, Target Identification & Drug Development—using predictive analytics to help pharmaceutical companies develop precision therapies for a narrow subset of patients.

Transforming complex healthcare data, and seamlessly sharing it with various healthcare organizations to ease the fear and burden of serious disease requires many capabilities. Unfortunately, needed capabilities for health systems and life sciences are often unavailable in traditional data platforms and would require additional solutions, along with the IT (and clinical, regulatory, etc.) expertise to implement those various solutions. When you’re researching potential real-world data and real-world evidence companies, be sure to ask about their ability to methodologically show you how their capabilities will work to power your ability to accelerate insights discovery. You can read more about Syapse Raydar, the Syapse data platform here

In the final blog post of this series, I’ll cover the critical topics of data privacy and compliance.