By Rachel Blum, Verato
Healthcare technology vendors are focused on building the next-generation platform in care management, revenue cycle automation, electronic health records, population health, and natural language processing. These platforms are built on modern tech stacks leveraging Big Data, artificial intelligence, cloud services, flexible APIs, and blockchain.
Yet as these vendors develop their products, they are realizing that the ability to accurately tie disparate data to a single person — called “patient matching” — is key to delivering their ultimate vision. But many of them do not realize how complex of a challenge it is to develop even minimally decent patient matching technology — called “master patient index” or MPI — and have instead embarked on building their own MPI in their solution.
The decision to use their own team instead of looking for a modern MPI technology to incorporate into their solution — like they may for a cloud provider — is influenced by a variety of factors. Let’s explore and unpack the thought process fueling these decisions:
Misconception 1: “It’s not a main feature, so it must be easy to build”
Patient matching has been viewed as an annoying, behind-the-scenes product feature, not the main feature driving the show. And as such, there is a belief that developers can easily craft a solution for this “simple” back-end feature in a couple of hack-a-thons, especially after having the wherewithal to design the rest of the product. However, building MPI technology is an extremely difficult and often underestimated challenge requiring much more than a simple string match.
Misconception 2: “My solution is unique, so I need a unique patient matching engine”
Every vendors’ solution is unique, which leads to the false belief that their solution requires a unique patient matching engine. This in turn leads to the conclusion that no commercial MPI solution could ever solve their unique patient matching problem. But the truth is, ultimately, that patient matching boils down to the same question no matter what — a question that is ubiquitous across healthcare: “Do these two or more entities/records belong to the same person?” The only differences in answering this question come in the implementation (i.e. when you ask the question) and in the difficulty of answering the question (i.e. the quality of your data that is being matched).
Misconception 3: “Commercially available MPI tech is too heavy and expensive to incorporate into my solution”
It’s true: almost all commercially available MPI solutions are heavy, expensive, difficult to implement, and difficult to maintain. This is because they were built before today’s technologies like cloud computing, Big Data, and machine learning existed. This led them to have to be on-premises, built on relational databases, and painstakingly configurable in order to get accuracy improvements worth their overall cost.
But not all commercially available patient matching technologies are so burdensome and difficult to incorporate. Modern infrastructures and Referential Matching technology have paved the way for SaaS solutions that provide cost-effective and hyper-accurate patient matching, with flexible APIs that make implementation simple. Cloud-based Referential Matching solutions effectively offer patient matching technology as a utility that any vendor can simply plug into their solution — rather than building their own patient matching tech from scratch.
While patient matching is vital to vendors’ end use cases and product lines, it is not the main focus of their business or their main area of expertise. Building a custom MPI involves creating an entire product: buying and maintaining hardware and software, tuning and updating algorithms, hiring data scientists and algorithm experts, and then spending upwards of a year building it all out. Instead, tech vendors should subscribe to modern patient matching services powered by Referential Matching technology and built for the cloud. These modern patient matching services have already invested millions of dollars and countless person-hours toward building world-class MPI solutions that vendors can simply plug into, allowing their developers to spend their time building the features that they were hired to build — the features that differentiate their tech from their competition.
About The Author
Rachel Blum is an executive with Verato.