Tom Burton and a group of healthcare veterans started Salt Lake City-based Health Catalyst in 2008. Initially, they wanted to revolutionize clinical process models using analytics. But during development, they were challenged by the quest for a data warehouse that could handle the complexity of healthcare data.
Burton noted that many hospitals were still organizing their staff around departments rather than care processes and how care is delivered to patients. Physicians were pretty much left alone on their own to venture in the sea of ever-expanding medical information. More importantly, medicine favoured traditional ways of doing things, even if they are inefficient and result in a variation of practice.
Burton wanted things to work differently. At the very least, patients should receive similar quality of care. There should be standardized ways to gather, evaluate, and disseminate knowledge to assist clinicians and analytics to guide how resources should be distributed and outcomes are measured.
So Burton and his team geared their resources to build a data warehouse that organizations can leverage to eliminate the manual processing and interpretation of data. Analysts can devote more time to uncover new patterns, gather information essential for decision makers and conjure models that underline potential clinical and financial risks.
The Health Catalyst data warehouse is backed by the Adaptive Data Architecture, a form of late-binding design. Binding is a method of mapping data within a data warehouse to standardize vocabularies and business rules, so information is consistent for analyses. Data warehouses consistently try to determine all the possible vocabularies and business rules that will be needed for mapping. But the practice, known as early-binding, is time-consuming and expensive. In healthcare, both vocabularies and business rules are constantly changing. As such, data mapping may need to take place, again and again, to keep the resulting models updated.
However, late-binding, the architecture which Health Catalyst’s data warehouse relies, only maps the data at the point of solving an actual clinical or business problem. This allows models to be agile, flexible, and be implemented in a matter of weeks compared to months or probably years that they usually require. It also avoids wasted time and effort through making lasting decisions about a data model up front. For example, certain tests need to be ordered every time in the case of acute appendicitis, but others are optional depending on the patient’s condition.
The Health Catalyst data warehouse and analytics tools provide the percentage of times various tests were ordered and the results of ordering and omitting certain tests based on past evidence. The technology not only renders insights to avoid potential wastage but also provides a shared baseline or an initial amount of antibiotics given to patients who had undergone an appendectomy. The suggested care plan helps physicians to review or adjust the antibiotic prescriptions to better cater to patients with unique sets of circumstances.
Last month, Health Catalyst announced the launch of Healthcare.AI, a suite of AI-driven products and services to augment the performance of existing tools for making accurate predictions. The company hopes to provide a more seamless, one-stop service to improve the health of patients as well as the population wherever they are, from the hospital bedside, to the boardroom – and even at their kitchen tables.