Embracing AI to Enhance Quality Control

Quality control (QC) is an essential part of any life sciences business that fixes product failures, reduces risk of recalls, mitigates regulatory violations and, ultimately, prevents patient harm. In addition to being essential, it can also be costly. By some estimates, quality-related costs range as high as 15-20% of sales revenue. QC processes, and especially when they catch failures, generate a significant portion of these costs through appraisals, waste, rework/repairs, recalls, and corrective and preventive actions (CAPAs). Fortunately, new applications of artificial intelligence (AI) in quality management and QC promise to increase inspection accuracy and extract new operations insights, while reducing the costs of both. Specifically, AI diagnoses defects, digitizes paper document data, and, combined with Internet of Things (IoT), discovers new parameters to augment quality control and decrease cost.

The Role of AI in Quality Control

Quality control inspections are often highly manual processes, requiring someone’s time and effort to check for product defects. Yet despite good efforts, inaccuracies persistently slip through; causing rework, batch loss, recalls, and potentially patient adverse events. To reduce this, life sciences companies are applying computer vision and machine learning to improve inspection accuracy and reducing external failure costs. Take LandingLens, which integrates deep learning software into a machine vision system to inspect medical device products. Trained on images from its manufacturing clients, it continuously improves accuracy as it analyzes more products and trainers add new defect images to its inventory. Moreover, more life sciences solutions providers apply AI in quality control to inspect an ever-expanding number of products to realize cost and quality benefits.

Another emerging use case is applying AI to digitize quality data currently documented on paper. For many life sciences companies, especially small- and mid-size ones, paper documents remain the primary tool to ensure and record that tests or production processes were performed in accordance with quality standards. These documents will contain process data and parameters; some of which are deemed critical to product outputs and part of existing QC checks. By digitizing written content through highly accurate AI-powered optical character recognition (OCR) and interpreting it through natural language processing (NLP), life sciences companies can take advantage of their troves of quality data and accelerate their digital transformation in quality management.

With this newly digitized dataset, AI-powered data science can analyze quality data to improve QC parameters, and subsequently overall product outcomes. Specifically, AI can find correlations between current processes parameters and rates of deviations, identifying parameter adjustments that optimize yield. Furthermore, by applying data science not only to existing critical parameters, but also to other data contained in the documents, there are opportunities to discover new factors which can be controlled for optimal outcomes going forward. For example, Aizon.ai applies AI to analyze data from an expanded set of process parameters for root cause analysis and process optimization. From this, it applies AI again to estimate and adjust QC parameter ranges to optimize batch yield. The capacity of AI in quality control to augment or discover critical parameters facilitates continuous improvement and supports regulatory objectives.

Extending AI’s Reach Through IoT

In addition to applying AI to newly digitized datasets, life sciences companies can enhance AI with IoT sensors to analyze previously uncollected data and further improve quality. Collecting data at more points of the process discovers new data types to analyze and enables intermediate QC checks. As with digitized quality documents, AI and IoT can be applied to identify new critical parameters, predict optimal parameter ranges, and monitor products or processes to further reduce quality deviations. In addition, this quality control granularity with AI monitoring can accelerate root cause analysis and CAPA management through precise identification, prompted by AI, of where a process deviated from standards.

Conclusion

The integration of artificial intelligence into quality control within life sciences is proving to have significant potential. QC, while vital, can generate significant expense. However, AI technologies can moderate this expense and improve outcomes through performing QC inspection, document digitization, AI-powered data analysis, and new parameter discovery tasks. With reduced defects and more insights, quality employees can reallocate their efforts to higher-value activities, such as optimizing standard operating procedures and rapidly resolving CAPAs. As more companies integrate these technologies into their QC operations, AI will transition from emerging to standard technology, with those lagging in adoption left behind. From the examples above, to the new applications coming out, AI is set to have substantial impact in quality control.