Rebekah Cauthen, Sales Manager—DACH region, Critical Manufacturing10.01.21
Quality is a cornerstone in the production of medical devices. It is fundamental to the success of a business and to the patients that use the devices. Yet, traditionally, it is also one of the biggest cost overheads for manufacturers. The way we approach quality, however, is changing and modern technology offers possibilities to enhance quality, reduce costs, improve patient outcomes, and reduce the pain of regulatory compliance.
With the right context and analysis, real-time data collected during production can provide deep insights into processes through machine learning (ML) algorithms and other advanced analytics. Inspecting for quality after a process is completed is reactive and outdated. Instead, the future lies in predicting quality. For medical device manufacturers, the advantages in predictive quality are so great they simply cannot be ignored. Indeed, it is the solution they need to meet the growing and conflicting pressures of reducing costs, speeding up new product introductions, and complying with increased regulatory requirements.
Benefits of Predictive Quality
Predictive quality is a proactive and adaptive approach to build quality into processes. It combines data from throughout the shop floor and wider supply chain to determine whether a product will meet quality requirements. Using a risk-based approach, it is dynamic in nature. As new data comes in, the company can adapt its processes and decisions. Advanced analytical techniques are used to determine the level of risk and to prioritize actions needed. This offers huge benefits to manufacturers including:
Defining and Controlling Risk
Product quality is the outcome of process consistency. For predictive quality to work effectively, risks associated with all aspects of product and processes need to be defined and prioritized in order of severity, timeframe, and nature. Once these are determined, parameters such as the frequency of sampling or testing can be set to mitigate the risk of product quality.
Historically, a risk-based approach was a static process that only went as far as the design phase of a product. In modern manufacturing settings, however, a dynamic function is required that modifies in accordance with changes in products, production processes, or regulations. It needs to accommodate the variability of processes and allow for all possibilities. It requires good process control and the removal of manual processes, such as paper-based quality management systems, which add variability and are not designed to ensure timely adjustment of process controls to avoid quality issues.
To keep up with the changes in medical device technology, predictive systems require access to accurate, timely, and complete data about process conditions and outcomes. Data comes from disparate sources including equipment, sensors, automation, and software, combining operational technology (OT) and information technology (IT). To produce the consistency and stability needed for reliable product quality, this data needs to be brought together, contextualized, and feedback loops put in place to ensure everything is operating as it should.
Integration Is Essential for Data Flows
Many different factors interact to determine the quality of a final product. To see the full picture and realize a predictive quality process, data from many sources need to come together. This requires information to flow vertically and horizontally. In Industry 4.0, the flow of information between offices, plants, and equipment is often referred to as vertical integration. It incorporates data from design, manufacturing, procurement, and the plant floor. Horizontal integration is the flow of data between different manufacturing plants, and between suppliers and buyers of materials, components, assemblies, and contract manufacturing services.
But how can such levels of integration across OT and IT landscapes be achieved without spending an inordinate amount of time (and money)? The answer lies in an integrated manufacturing data platform.
The Manufacturing Data Platform
Medical device manufacturers require a manufacturing data platform that can integrate a wide variety of manufacturing data from IT and OT, create context for that data, analyze real-time data to predict issues, and drive appropriate actions to prevent issues. Some of the data sources for such a platform are applications such as manufacturing execution systems (MES), industrial internet of things (IIoT) platforms, quality management systems (QMS), manufacturing intelligence (MI), and analytics. However, none of those alone fully supports predictive quality. The platform must not only gather and analyze data to predict quality, but also show corrective actions, and then drive employees to take appropriate actions to avoid quality issues.
The quality functions within modern MES solutions move manufacturers closer to predictive quality. Unlike a QMS, the MES provides greater context about materials, specifications, equipment, work instructions, operators, and statistical process control (SPC) data. Because it can consider the full production record of a device, it is able to achieve more accurate root cause analysis (RCA) and can monitor whether preventative measures are effective. Any non-conformance is recorded with full context, enabling patterns within the full production landscape to be identified. These patterns feed effective predictive analytics algorithms to support continuous quality monitoring and improvement.
A modern manufacturing data platform integrates to ERP, PLM, SCM, and HRIS, as well as automation, IIoT, and OT equipment. Combining the plantwide information and guidance of an MES, it provides advanced analytics including predictive algorithms and models and incorporates ML and a digital twin, quality management, and maintenance and scheduling functions. It enriches and contextualizes data from throughout the web of IT and OT systems and devices.
From Data to Action
Analysis and prediction of issues are useless without timely action. Alerts may trigger a change to a process, product, material, work instruction, or equipment. For example, monitoring the condition and performance of equipment to predict failures before they happen. In this scenario, engineers can plan ahead for preventive maintenance on equipment to prevent serious consequences such as lost production time, product issues, repair costs, and lost profits. Care needs to be taken, however, that a waterfall of alerts does not prevent timely action—we have to “see the forest for the trees.” To this end, the system also needs to prioritize alerts based on the level of risk and direct them to the right people. Ultimately, the shorter the time between an event occurring and action being taken, the greater the improvement impact.
Meeting the Needs of Medical Device Manufacturers
Other factors that determine the success of a predictive quality solution include its configurability to individual site and product needs plus its readiness for validation and verification. Ideally, it should also enforce standard operating procedures (SOPs) and prohibit non-conformance across processes, products, and documentation. Successful implementation further requires a global mindset, as the full benefits lie in integrating data from all production steps across all production sites.
Conclusion
The benefits of an effective predictive quality solution are huge. Lower costs, enhanced quality, increased production efficiency, and fewer issues with compliance and customer complaints are all on the table. The challenges associated with creating such a system, however, lie in the integration and contextualization of data from sources throughout IT and OT landscapes across all manufacturing sites and the creation of correct data flows and feedback loops to turn the data into actionable information. To achieve this requires a modern, integrated manufacturing data platform with the controls and integration provided by a modern MES, all designed with a balance for configurability and validation and deep understanding of the needs of the medical device industry.
Rebekah Cauthen has over nine years of experience in manufacturing IT, having worked with various manufacturing technology companies in both the United States and Europe. In 2019, she joined Critical Manufacturing as the sales manager for the DACH region. Cauthen’s passion is supporting medical device manufacturers in their transition to manufacturing excellence and intelligence through state-of-the-art technology.
With the right context and analysis, real-time data collected during production can provide deep insights into processes through machine learning (ML) algorithms and other advanced analytics. Inspecting for quality after a process is completed is reactive and outdated. Instead, the future lies in predicting quality. For medical device manufacturers, the advantages in predictive quality are so great they simply cannot be ignored. Indeed, it is the solution they need to meet the growing and conflicting pressures of reducing costs, speeding up new product introductions, and complying with increased regulatory requirements.
Benefits of Predictive Quality
Predictive quality is a proactive and adaptive approach to build quality into processes. It combines data from throughout the shop floor and wider supply chain to determine whether a product will meet quality requirements. Using a risk-based approach, it is dynamic in nature. As new data comes in, the company can adapt its processes and decisions. Advanced analytical techniques are used to determine the level of risk and to prioritize actions needed. This offers huge benefits to manufacturers including:
- Lower costs through reduced scrap and re-work
- Reduced overheads with more targeted sampling and testing
- Increased production efficiency
- Faster new product introduction through deeper understanding of manufacturing and quality challenges
- Enhanced quality with reduced risk of non-compliance or non-conformance
- Fewer issues with shipped products and increased customer satisfaction
Defining and Controlling Risk
Product quality is the outcome of process consistency. For predictive quality to work effectively, risks associated with all aspects of product and processes need to be defined and prioritized in order of severity, timeframe, and nature. Once these are determined, parameters such as the frequency of sampling or testing can be set to mitigate the risk of product quality.
Historically, a risk-based approach was a static process that only went as far as the design phase of a product. In modern manufacturing settings, however, a dynamic function is required that modifies in accordance with changes in products, production processes, or regulations. It needs to accommodate the variability of processes and allow for all possibilities. It requires good process control and the removal of manual processes, such as paper-based quality management systems, which add variability and are not designed to ensure timely adjustment of process controls to avoid quality issues.
To keep up with the changes in medical device technology, predictive systems require access to accurate, timely, and complete data about process conditions and outcomes. Data comes from disparate sources including equipment, sensors, automation, and software, combining operational technology (OT) and information technology (IT). To produce the consistency and stability needed for reliable product quality, this data needs to be brought together, contextualized, and feedback loops put in place to ensure everything is operating as it should.
Integration Is Essential for Data Flows
Many different factors interact to determine the quality of a final product. To see the full picture and realize a predictive quality process, data from many sources need to come together. This requires information to flow vertically and horizontally. In Industry 4.0, the flow of information between offices, plants, and equipment is often referred to as vertical integration. It incorporates data from design, manufacturing, procurement, and the plant floor. Horizontal integration is the flow of data between different manufacturing plants, and between suppliers and buyers of materials, components, assemblies, and contract manufacturing services.
But how can such levels of integration across OT and IT landscapes be achieved without spending an inordinate amount of time (and money)? The answer lies in an integrated manufacturing data platform.
The Manufacturing Data Platform
Medical device manufacturers require a manufacturing data platform that can integrate a wide variety of manufacturing data from IT and OT, create context for that data, analyze real-time data to predict issues, and drive appropriate actions to prevent issues. Some of the data sources for such a platform are applications such as manufacturing execution systems (MES), industrial internet of things (IIoT) platforms, quality management systems (QMS), manufacturing intelligence (MI), and analytics. However, none of those alone fully supports predictive quality. The platform must not only gather and analyze data to predict quality, but also show corrective actions, and then drive employees to take appropriate actions to avoid quality issues.
The quality functions within modern MES solutions move manufacturers closer to predictive quality. Unlike a QMS, the MES provides greater context about materials, specifications, equipment, work instructions, operators, and statistical process control (SPC) data. Because it can consider the full production record of a device, it is able to achieve more accurate root cause analysis (RCA) and can monitor whether preventative measures are effective. Any non-conformance is recorded with full context, enabling patterns within the full production landscape to be identified. These patterns feed effective predictive analytics algorithms to support continuous quality monitoring and improvement.
A modern manufacturing data platform integrates to ERP, PLM, SCM, and HRIS, as well as automation, IIoT, and OT equipment. Combining the plantwide information and guidance of an MES, it provides advanced analytics including predictive algorithms and models and incorporates ML and a digital twin, quality management, and maintenance and scheduling functions. It enriches and contextualizes data from throughout the web of IT and OT systems and devices.
From Data to Action
Analysis and prediction of issues are useless without timely action. Alerts may trigger a change to a process, product, material, work instruction, or equipment. For example, monitoring the condition and performance of equipment to predict failures before they happen. In this scenario, engineers can plan ahead for preventive maintenance on equipment to prevent serious consequences such as lost production time, product issues, repair costs, and lost profits. Care needs to be taken, however, that a waterfall of alerts does not prevent timely action—we have to “see the forest for the trees.” To this end, the system also needs to prioritize alerts based on the level of risk and direct them to the right people. Ultimately, the shorter the time between an event occurring and action being taken, the greater the improvement impact.
Meeting the Needs of Medical Device Manufacturers
Other factors that determine the success of a predictive quality solution include its configurability to individual site and product needs plus its readiness for validation and verification. Ideally, it should also enforce standard operating procedures (SOPs) and prohibit non-conformance across processes, products, and documentation. Successful implementation further requires a global mindset, as the full benefits lie in integrating data from all production steps across all production sites.
Conclusion
The benefits of an effective predictive quality solution are huge. Lower costs, enhanced quality, increased production efficiency, and fewer issues with compliance and customer complaints are all on the table. The challenges associated with creating such a system, however, lie in the integration and contextualization of data from sources throughout IT and OT landscapes across all manufacturing sites and the creation of correct data flows and feedback loops to turn the data into actionable information. To achieve this requires a modern, integrated manufacturing data platform with the controls and integration provided by a modern MES, all designed with a balance for configurability and validation and deep understanding of the needs of the medical device industry.
Rebekah Cauthen has over nine years of experience in manufacturing IT, having worked with various manufacturing technology companies in both the United States and Europe. In 2019, she joined Critical Manufacturing as the sales manager for the DACH region. Cauthen’s passion is supporting medical device manufacturers in their transition to manufacturing excellence and intelligence through state-of-the-art technology.