Howdy! Welcome to our first blog! For those of you who know us, you know we’re a bunch of data nerds. For those of you who don’t know us: we’re a bunch of data nerds who are obsessed with democratizing data. We don’t think you should need to be a senior data scientist to access data or make basic data-driven decisions.
The reason it’s so costly and time consuming for organizations to answer simple questions with data isn’t a BI tool problem, AI problem and certainly not a sparse data problem: it’s a foundational problem. The technical foundations for ingesting, storing and most importantly consistently labelling and serving up data for analytics are broken.
Organizations’ back-end data systems are optimized for operations rather than analytics. This is particularly true in healthcare: unlike consumer data where the operational goal of understanding consumer patterns and behaviors is intertwined with the end use, healthcare data is often used for something other than its original operational purpose.
Take claims: claims are transactional receipts for the purpose of adjudicating payment of service(s) and claim systems are optimized for the single purpose of paying claims, making any other use of that data very difficult.
Claims hold a lot of valuable information beyond payment: they are the recorded trace of almost every healthcare transaction that occurs in this country. Making sense of that data has implications ranging from population health and disease management to clinical intervention and on-the-ground patient care. However, when an organization has access to claims data and are looking to answer a simple question, you’ll often hear the following reasons why that is difficult, too segmented, no single “data lake”, too slow, can’t interpret fields or metrics, not in the right form, don’t know where it came from, too inaccurate.
In other words, end data users can’t easily access, trust or make any sense of the data.
The result: instead of a hospital CEO quickly getting an answer to a question like, “What were our top three complications associated with inpatient diabetic stays last month?” she gets three different answers from three different people and her team spends their time reconciling data and reports instead of strategizing how to reduce those complications. Why does she get three different answers from three different people? Because they’re each pulling data from different places, with different definitions and metrics and not using repeatable processes to do so.
Bytemap’s Quelle platform eliminates those data conflicts within organizations by automating tedious, but necessary, data processes. Technology and data are as advanced as ever in terms of size, speed, processing, data science and self-learning models, yet we constantly hear about “data problems”. It’s not the companies with the least advanced AI model who will fail, it’s the companies struggling to reliably access their own data now. A strong foundational layer is key to success for any business trying to leverage data (any business).
As much as we believe a strong foundational data platform is key to business success, we don’t think building and maintaining it is a good use of your senior teams’ time or skill. Like the architect or interior designer doesn’t lay the cement foundation or put up the scaffolding, your best tech folks shouldn’t be spending their time wrangling data .
Your data scientists should be running advanced models, interpreting results and working with strategic folks to solve problems and your leaders should have quick reliable answers to data-based questions.