What makes Galileo Cosmos™ unique?
A client recently asked for a simple graphic that represented their core target patient population. Easy, you say; but unfortunately this particular patient population is made up of complex cohorts. These cohorts are based on a complex mixture of diagnosis, previous procedures, treatments for the target condition and other therapies for co-morbid conditions. Our job was to explore their data to identify the characteristics of those cohorts and then put them into a graphic that would be easy to understand. We had to explore first and then visualize – all on a tight timeline and with a fixed dataset.
Visual data exploration involved here was a very different concept than the data visualization. In visual data analytics the goal is to provide the analyst with an interactive visual representation of their data based on an exploratory approach. Data visualization displays results from queries based on a known or defined hypothesis. For this project we used Galileo Cosmos™- a visual exploration and analysis tool.
Galileo Cosmos™ is written in a language that allows us to manage and analyze data in ways that conventional tools do not. Rather than a conventional relational data structure, Cosmos has a flattened data structure using a language that is designed to deal with vast arrays of data in real time. This gives Cosmos many benefits over conventional data structures:
- Speed of operation – the language we use is designed for real-time analysis of huge datasets and so working with a few million patient claims records, or the entire Hospital Compare dataset is a straightforward task.
- Layered data integration – by flattening the layers within the data, Cosmos allows us to integrate data from multiple sources into a single dataset for analysis. For example combining clinical trial data with claims data, primary market research with EHR data etc.
- Non-relational structure – analysis of the data is not restricted by conventions built into the data structure. Analysts can identify and build their own structures without being concerned about existing structures. For example, building co-morbidity variables based on a combination of demographics and test procedures as well as diagnosis.
- Low machine overhead – the core language of Cosmos is a fraction of the size of conventional data structure approaches, which means that it runs more efficiently (and faster) as well as having all the parallel processing capabilities necessary to run large data sets in real time.
From the founding of the company, Galileo Analytics has been working with our sister company, Optima Systems from the UK. Their 30-year experience with the core languages and data handling techniques used in Cosmos have made it the powerful tool it is today – as further described in the linked case study below.