“The ultimate purpose of collecting data is to provide a basis for action or a recommendation.” W. Edwards Deming

The words of the great Deming are truer than ever in the Industry 4.0 paradigm, where data is the biggest driving force behind the large scale digital transformation of companies, to become a more efficient, agile and resilient operation. While data has always been extremely important in manufacturing operations, what has changed in the last decade is the exponentially increased availability of raw data from the manufacturing process and the extended supply chain.
Data abundance from IoT sensors, process equipment and IT applications has created immense complexity in data analytics. Manufacturing plants no longer process data in batches. Rather, data is streamed continuously from a myriad of sources, which in its raw form is useless. Unless collected in proper data lakes/repositories/cloud, the data needs to be standardized, organized, streamlined, contextualized, analyzed and presented in actionable forms to requisite stakeholders.

What is DataOps?

DataOps as a discipline has started gaining popularity, and is focused on the way data is harnessed, ingested and engineered. The final product—data analytics—is useful only after it gets deployed across organization. DataOps brings together postulates and practices from Agile application development, DevOps and Statistical Process Control, to ensure data analysis applications are developed iteratively, improved continually through feedback and collaboration and any variance in data is detected and corrected before it can affect the output of the analytical tools being developed and deployed.
DataOps essentially builds quality into the very fabric of data management.
As data passes through the data pipeline, in a DataOps environment it is continually enriched through standardization, control and contextualization to create analytical tools which deliver pointed and actionable data models. These tools keep improving themselves, iteratively based on the feedback from the system itself and users. It allows development teams to collaborate and create better solutions faster. DataOps can be distinguished from other methodologies on the basis of how data is developed–from raw form into actionable intelligence– and based on the functionality which is unique to the process.

Data Development

In a typical DataOps system, raw data streams from various sources are ingested and assimilated into the data pipeline, eliminating data silos. Data engineers and analysts work on enriching the data as it moves through the pipeline and the process, with each incremental step adding complexity with increasing co-dependence and data refinement.
DataOps essentially brings together the best practices of Lean, Agile Development and DevOps to Data Management and it is gaining popularity in regulated industry segments.
According to a post on Techrepublic.com, based on results from a survey, companies in Banking, Retail and Healthcare are more likely to have a data engineering team and thereby more likely to invest resources in data supply side. The post also highlights that 90% of the respondents believe that data quality and trust will become more important than data quantity in the next 24 months. The two findings presented point towards the inference that highly regulated industries are looking for higher quality data and are doing so by pursuing DataOps as a way of achieving this.
So, the question worth considering becomes, what does DataOps as a discipline bring to regulated industries, and in particular industries like Medical Device manufacturing?

Automated Validation and the role of DataOps building quality into the validation process

When it comes to data management, DataOps by its very nature is a practice which puts quality first. Regulated industry segments, such as medical device manufacturing, lay major emphasis on quality in all aspects of the manufacturing process. Quality oversight includes product design and manufacturing, software applications which control and orchestrate the process itself and ultimately the quality of the product itself, which pertains to its specified and actual performance for the end user.
At each and every step of the manufacturing process, from design to production and software solution development/deployment, is subject to strict validation protocols put in place by regulatory authorities like the FDA in the US and EMA in Europe.
In our post covering the benefits of Automated Validation, we explain how MES implementations benefit from automated validation practices. We discuss how software validation requirements are alleviated and implemented through a modern MES, alongside process and product validations.
DataOps forms an essential component of automated validation, and with its correct application, it is possible to increase the speed and quality of the validation phase. With DataOps in place, and the right MES vendor, it is possible to leverage existing agile methodologies (DevOps), previous systems updates data and SPC domain knowledge to incorporate and implement high quality solutions.
Once collected and normalized, the data can then be used to expedite the validation process for the MES itself. DataOps allows faster execution of the validation cycle for software deployment and allows manufacturers to focus on product and process validation challenges which arise with any new product and where risk mitigation is a priority. With DataOps in place, medical device and other regulated industry manufacturers get information-rich validation data through SPC and analytical tools from the MES, which get collected over previous iterations and implementations of the application.
When executed through tools like Azure DevOps, it makes the validation process faster and higher precision.
DataOps is a natural fit for the Automated Validation process, which reduces not only the validation and testing costs for medical device manufacturers, but also helps improve Time to Market (TTM), boost competitiveness and inherent resilience. With DataOps, quality is built into solution implementation and the validation process. The data collected and analytics solutions developed are tightly controlled and orchestrated, with continuous testing and validation through SPC, ensuring data streams are propagating the highest quality data with minimal possible erroneous data points.
It is due to this self-actuating nature of DataOps in capturing, streamlining and enriching data that it is gaining popularity in the Regulated industry segments. With the right MES, which is capable of deploying agile software development methodologies, using tools like Azure DevOps, lean code management and SPC built into its development process, DataOps and Automated Validation can become a reality, delivering high quality, validated and self-improving data management across the organization.

WP Automated Validation