Big Data is Coming to Big Pharma
A 2016 study investigated the potential impact of big data and data modeling on the health care industry, and, while the paper may be a bit older, the outlooks can still yield great benefit to us today. Some of the things mentioned in the study may even be things that are being implemented as you read this!
Introduction
As of the writing of this research, the Information Technology field has had a slow start into the medical field.
As powerful as IT is, integration into the medical field is all but necessary, and the advancement of Big Data as an industry is extraordinarily promising. Especially as storage methods become more efficient and better in general.
Since IT is so prevalent in other, similar fields, it seems odd that the medical adoption rate was lagging so far behind.
It stands to reason that it would be more than worth it to study other fields for their adoption methods and efficacy in terms of advancing technology.
Background
Big Data, Past and Present
In terms of terms, “Big Data” has been a phrase since 1997, when it was first used by Michael Cox and David Ellsworth in a paper presented at IEEE, explaining the visualization of data and the challenges that computer systems may encounter because of it.
From 2001 to 2008, Big Data as an industry started to take shape. Revolving around the 3Vs (Volume, Velocity, and Variety), it was accompanied by advancements in methods for storing and using data - things like XML and Hadoop were still new on the scenes.
Additionally, this is when hospitals started to digitize their patient records.
2009 saw a shift toward unstructured data as health data climbed to a staggering 150 exabytes - or 150,000,000,000 GIGABYTES of data.
Across industries, cloud environments were starting to gain traction. Banks, commerce, and a lot of other industries started seeing process improvements, but the potential of it all was still very young. Predictive models and simulation capabilities were still over the horizon and hadn’t been fully developed.
By 2016, the time of the research paper’s writing, cloud computing started to reach a point of normalcy.
Research Architecture
In order to make any progress, the research had to have a structure for its approach. Researchers worked with various IT practitioners and academics to design a layered process to determine best practices.
1. Data
Data is required for operations support. Studies can’t be done if there isn’t data. At this layer, data is split into structured, unstructured, and somewhere in-between. This data was collected from sources both internal and external and then was immediately stored.
2. Aggregation
Layer 2 is for processing and generally handling the data from before, effectively cleaning and converting the data for use in later layers.
3. Analytics
The third layer is where the bulk of the processing is handled. Additionally, this is for real-time and parallel processing. Basically, any kind of analysis is done at this point.
4. Information Exploration
The exploration layer is what we would now call EDA, or exploration data analysis. This is where certain aspects of the data would be visualized and where real-time data is monitored, as well as a few other pieces of explanatory processing.
5. Data Governance
Data governance describes the overall management of the data at hand, including how policies are applied to both the input and the output data. Things like regulation would be observed and managed here.
The lifecycle of data is also looked at at this point. How to manage business information, how to archive, maintain, and test everything, it all falls under data governance. Related to this is deletion, which is covered in much the same way.
This layer is the most concerned with ethical, legal, and regulatory status. It fully encompasses data security and privacy.
Analytics Capability of Big Data
Big Data is a concept designed around a lot of data. Conveniently, a lot of data is what the medical industry is working with. The field of data analysis is set up to be extremely efficient at working with this sort of thing, and therefore has the exact kind of analytical capability that we’re looking for.
Conceptualizing benefits
There is a lot of potential for benefits when adopting Big Data methodologies across several pieces of the industry puzzle.
IT infrastructure
- Flexibility in business needs
- Reduction in costs
- Greater general capabilities
Operations
- Reduced costs
- Reduced cycle times
- Greater productivity
- Higher quality
- Improved customer service
Management
- Better resource management
- More informed decision making
- Improved planning and performance
Strategic Benefits
- Stronger business growth support
- Business alliance support
- Better cost-leadership
- Greater product differentiation
- More robust external linkages
Organizational Benefits
- Work pattern changes
- Facilitation of organizational learning
- Organizational empowerment
Reasearch Methods
Research done took a quantitative approach to using multi-case analysis, as described in the following sections.
Case collection
All cases used were drawn from then-current and past projects that meet two specific criteria:
- Representative of actual big data implementation
- Clearly describes the benefits and implementation of software
There were 26 cases that were related to the health care industry in some way.
- 14 were from vendor or company materials
- 2 were from academic journal databases
- 10 were from print publications
By region, 17 were from North America, 7 were from Europe, and the remaining 2 were from the Asia-Pacific region.
Results
Capability profile
There were 5 general categories of application for Big Data principles uncovered by the study, as outlined here:
| Category | Number of Cases |
|---|---|
| Analytical/ Pattern of Care | 43 |
| Unstructured | 32 |
| Decision Support | 23 |
| Predictive | 21 |
| Traceability | 17 |
Patterns of care
Big data analysis methods provide a broader view for clinical practice that is evidence-based given the sheer quantity of medical data that exists. Over massive amounts of data, seeing and recognizing consistent needs for care.