A multidimensional visualization model for zero defects in a biomechanical manufacturing environment

Combinational medical devices have evolved rapidly over the last 20 years, to be minimally invasive in order to reduce patient risk and recovery times. However the Manufacturing and Quality control of these devices has not kept pace with the advances in design technology. Instead the industry relies on eighty year old, end of line, acceptance sampling methods, due in part to destructive nature of some of the tests but mostly due to hesitancy in embracing new technological concepts. The regulatory bodies have acknowledged this hesitancy and are encouraging the use of Process analytical technology (PAT) concepts to understand and control process variation at source. Two strands of research were required to meet this challenge. The first is to develop real time non destructive assessment methods and the second requirement is to develop an approach for the statistical assessment of the overall process in real time. This thesis investigates the application of this technology on two critical to quality attributes of a combinational medical device, which currently apply end of line destructive acceptance sampling for batch release. The attributes, catheter bond strength and stent drug content were identified by the senior management in Abbott Vascular. Infrared thermography was used as to develop a non‐destructive method for the real time assessment of the laser bonded catheter joint. The IR profile was assessed using critical regions to predict tensile bond strength. Drug content was assessed by developing a transferable Raman Spectroscopy Technique which quantified the coating components non‐destructively. Extensive designed experiment was deployed to establish the critical factors affecting variation and to create populations of product at extreme conditions. Prototypes for both methods developed and demonstrated successfully on site to the senior management team in Abbott Vascular. Real time statistical assessment of the process output was achieved by developing an ensemble approach which combines the strengths of both univariate and multivariate approaches while addressing the weaknesses of each. Heatmaps with associated rules were developed to display the output and the technique was applied successfully on coating process data in Abbott Vascular.