Predictive Quality for Medical Devices

Based on the e-book, Predictive Quality for Medical Devices: Comprehensive Data to Move Risk-Based Approaches into Production (Download here)

The medical device industry faces a myriad of challenges when it comes to managing product quality. As batch sizes reduce, product mix increases and electronic components in the product become more complex. Data abundance from in-process IIoT sensors and devices add to the potential noise. Manufacturers need to transition from a traditional quality management philosophy to one more attuned with Industry 4.0 to deliver quality which relies on assessment, prioritization and subsequent mitigation of risk.

Quality as perceived in an Industry 4.0 world extends beyond the factory floor. Quality is expected to be monitored, controlled and harnessed throughout the product’s lifecycle. A manufacturing strategy based on quality leads to better products, which are launched faster. It also mitigates risk while being produced and in use.

Effective quality management depends on accurate and contextualized data. The abundance of data created through automation, equipment-level integration, supply chain management and from product use in the field becomes a major challenge to utilize if it is not contextualized and presented in forms that facilitate better decision making.     

Traditional quality management is broken

• Quality-related data, when collected through traditional forms, leads to reactive decision making. It can’t effectively be used to implement process control. The more the time taken to detect, correct and prevent an issue, the larger the potential loss. Modern, agile processes rely on fast decision making which can translate to immediate actions which may prevent loss. Most traditional quality management tools are incapable of delivering this much-needed speed.

• Quality is a function of multiple factors which translates to data points which are often beyond the factory floor. Many quality events have repercussions which extend beyond a company or a site to further down the supply chain. With traditional QMS systems, more often than not granularity of quality data is not representative of its impact. Data exists in silos and are unable to impact changes needed in a supplier or contract manufacturer’s process, in time to reduce impact of events happening at the customer end of the supply spectrum.
• Traditional quality management does not leverage modern tools which can facilitate decision making. Integration between IT and OT, IIoT, AI and ML keeps the entire process in perspective, avoiding myopic results.  
• For multisite operations, quality management processes, production flows, IT infrastructure, product variants, personnel culture/language, and set standards may vary significantly, leading to varied quality results within different manufacturing locations of a single company. Traditional quality tools tend to be localized. They lack the overarching quality infrastructure to bring all production sites under one uniform system.
• Traditional quality management systems rely on in-process sampling and testing, and post-production reviews. It doesn’t focus on risk mitigation, which is a priority for regulators.
• Feedback loops from in-process quality management and product lifecycle data to R&D, suppliers and contract manufacturers are almost non-existent with traditional quality systems.  If feedback is collected and shared, it is nowhere near real-time.

With a flawed quality management system, not suitable for Industry 4.0, medical device manufacturers need to look towards the future. They need to modify the very definition of what quality and its management means for them. This transformation from an archaic quality management paradigm to a modern system is easier said than done.

So how do you change from a reactive, restrictive and redundant quality management practice to a more risk-based approach, where quality is embedded in the DNA of the process and product?

The answer: predictive quality.

Predictive quality explained

Predictive quality is a risk-based approach to quality. The goal is to minimize risks as efficiently as possible by adapting activities to the size or magnitude of each risk. For Quality, the processes, as well as the products, need risk specifications to determine the impact of each event.

• Predictive Quality Management overcomes the shortcomings of traditional quality management systems by allowing possible quality events to be predicted. It prompts proactive actions to prevent the infraction from ever happening.
• Risk-based analysis begins with categorization of risks, followed by definitions of likelihood and severity. It permits process activities to be modelled in a way that mitigates risks based on their potential threat level. This approach covers the entire product lifecycle, ensuring end-to-end quality management. Critical to quality (CTQ) parameters monitored from design to production and any incident which goes beyond the established process envelope leading to immediate automated alerts and prompted actions are clearly defined through self-improving SOPs (standard operating procedures). 

• Predictive quality management works on feedback loops which enables data flow to remain real-time and in context as it gets collected and subsequently passed onto design, contract manufacturers and supply chain partners.
• Predictive quality creates a ‘virtual process envelope’ which encompasses assessed and prioritized risks with complete documentation at every step.
• Predictive quality incorporates the three qualification levels: installation, operational and performance (IQ, OQ and PQ), forming the basis of stable, predictable processes.
• Predictive quality is a derivative of IT and OT data convergence, which relies on integration for accuracy and timeliness. Data from vertical integration (integration between Automation, MES, QMS and ERP) within the plant and from horizontal integration (data collected from several plants across the company)  is leveraged to create a self-regulating process performance envelope, with just the right amount of standardization and site-based customization. 
• Predictive quality needs more than a MES or a QMS. While it can be argued that MES and QMS are better together, the tool which unleashes predictive quality is a modern Manufacturing Data Platform, which extends beyond the levels of connectivity and insight provided by any standalone system.

A Manufacturing Data Platform for Predictive Quality

A manufacturing data platform is a modern, overarching solution which covers the entire supply chain. It collects data from IIoT sensors, factory automation, MES, QMS, ERP and other sources from supply chain and support systems. Then it applies AI/ML and advanced analytics to compare historical data with real-time data collected. Finally it creates the insights for the basis of predictive quality.

With the right MES data platform, trends, insights and predictions can be used to control actions on the shop floor. A performance envelope is deployed which lets the process function smoothly.  This performance envelope leads to the automated batch delivery of products, where production continues uninterrupted unless the system detects and reports an issue.   

As the modern MES platform contains quality management within its functionality, SPC (statistical process control), RCA (root cause analysis), CAPA (corrective/preventative action) and NC (non-conformance) reporting becomes much more efficient.

As is data collected, it is refined, contextualized and presented in a manner so that personnel can understand why an event occurred, how it can be corrected, if it presents future risk and if it be mitigated permanently through changes in process or design.  Employees are empowered to focus more on what needs to be done, eliminating issues as prescribed by the Platform.

Predictive quality final thoughts

Predictive quality brings together advanced analytics and process control through the identification, quantification and elimination of risks. Its scope spans from the medical device design and manufacturing process through the extended supply chain. It provides enterprise-wide insight on possible quality events and helps improve the design of process and products, so that quality is incorporated throughout the product’s lifecycle.

Predictive quality management, through the application of advanced analytical capabilities, is a journey. It begins with harnessing data and contextualizing it via a single, comprehensive data platform.

Learn more about how predictive quality management can change the way you operate and how it can help reduce quality related costs, while ensuring compliance, increasing production efficiency, incorporate automated product delivery and making quality a part of product design DNA. Read our new e-book by clicking below.