At BioMire, we have integrated these new data analytics practices into our nomad platform. The system links microbial test results to process parameters through a paperless, real-time process that delivers insighful dashboards to manage risks and monitor progress.
La prévention des contaminations microbiennes dans l'industrie par le renforcement des opérations et le développement du capital de connaissances
Today's Data is Tomorrow's Raw Material When we face hard-to-solve problems, a change in perspective can sometimes be amazing. The Swiss cheese model of accident causation is one that can turn a wall of complexity into a manageable challenge. In short, the model represents a #riskmanagement system as layers of cheese with randomly placed and sized holes in each slice. Each hole represents a weakness in defenses. For the system to fail, several holes (or planets) need to align. It differentiates active and latent failures, the first encompasses human errors and the later includes contributory factors that may lie dormant until they contribute to an accident. For example, an error in incoming QC control is an active failure and a weakness in the Forward Process Flow is a latent one. Individually, each failure should not cause an accident, but combined they might. Analogy with HACCP and HARPC The 2008 PCA public health disaster was essentially caused by the coincidence of contaminated raw material and poor cleaning practices. The HACCP approach focusses on controlling a number of Critical Control Points, which we can imagine as fixed points on the cheese layer stack. In the PCA example, some test results indicated the presence of contaminants in the products alerting to an alignment of holes. But since testing consistently showed that products met microbial specifications and re-tests were negative for contamination, the alert was disregarded, … and the system later crashed. In reaction, the FDA introduced the HARPC system to emphasize proactive risk identification in #FoodSafety That’s like measuring the holes in each cheese layer, analyzing how they might align and work on maintaining a misalignment (including closing holes). Measuring holes in each layer means separating and analyzing each layer. The layers are generally known: raw material control, process control, environmental hygiene, product testing, training and procedures. Measuring holes requires that test results can be associated with contextual information on the cheese layers. For example, a test shows equipment E is sanitary, …when cleaned on Monday with protocol P, using disinfectant D, left closed overnight and used for production on Tuesday. If test results are higher when all other parameters being identical, E is cleaned on Friday and put in production on Monday, this indicates a weakness in equipment standby preservation. #DataDrivenDecisions means a lot of data Yes, but most of it already exists. The tools to aggregate the test results in a way that generates insights is the issue. Some insightful data is missing Yes but there are now tools to define, collect and combine that information with existing data history. The promise of HARPC or knowing your cheese, is the simultaneous reduction of accidents and of the cost of controls, because testing with purpose builds understanding, understanding contributes to process efficiencies.