A Google search for ‘Gaussian Process Regression’ returns some intimidating material for a non-statistician. After filtering away the obscure stuff I’ll never understand and digging around within the code that makes GPR happen, I’m proud to say that I feel I’ve gotten my arms around the basics ...

When testing is costly or resource intensive, it’s not uncommon for management to ask an engineer “what are the chances that we pass?”. The answer will depend on the team’s collection of domain knowledge around the product and requirement but also in how the question is interpreted. It ...

Despite our best efforts, nitinol implants fracture and fail. Sometimes we want them to fail (on the bench, to learn). Other times they fail unexpectedly and we need find out why. When the failure is a fractured nitinol structural element, high mag...

Frequentist statistical methods, despite their flaws, are generally serviceable for a large suite of practical problems faced by engineers during product development of medical devices. But even in domains where simple models usually do the trick, there remain instances where a Bayesian approach is the best (and perhaps only logical) ...

The current industry protocol for durability testing of vascular stents and frames involves testing many implants simultaneously at a range of different stimulus magnitudes (typically strain or stress). The test levels are spread out like a grid across the stimulus range of interest. Each implant is tested to failure or ...

In this post we will try to trick linear regression into thinking that a redundant variable is statistically significant. By redundant, I mean a candidate predictor variable that in reality is just noise (no effect on the outcome) but that we might include in an experiment because we don’t ...

In this post I’ll be taking a break from my normal explorations in the medical device domain to talk about Sliced. Sliced is a 2-hour data science competition streamed on Twitch and hosted by Meg Risdal and Nick Wan. Four competitors tackle a prediction problem in real time using ...

Tolerance intervals are used to specify coverage of a population of data. In the frequentist framework, the width of the interval is dependent on the desired coverage proportion and the specified confidence level. They are widely used in the medical device industry because they can be compared directly vs. product ...

I decided to do something a little bit different with this post and show how R can be used in tandem with a traditional engineering CAD program. Together they comprise a streamlined and repeatable workflow that I’ve tried to leverage on my job when it makes sense to do ...

Today we have been assigned the task of identifying boundary conditions for a benchtop durability test of an implantable, artificial heart valve. In other words, we need to identify credible parameters for a physical test such that our test engineers can challenge the device under severe but realistic geometries and ...

I've heard it said that common statistical tests are just linear models.1 It turns out that Gage R&R, a commonly used measurement system analysis (MSA), is no different. In this post I'll attempt to provide some background on Gage R&R, describe the underlying model, and then walk through ...

Whether you are building bridges, baseball bats, or medical devices, one of the most basic rules of engineering is that the thing you build must be strong enough to survive its service environment. Although a simple concept in principle, variation in use conditions, material properties, and geometric tolerances all introduce ...

This post is my good-faith effort to create a simple linear model using the Bayesian framework and workflow described by Richard McElreath in his Statistical Rethinking book.1 As always - please view this post through the lens of the eager student and not the learned master. I did my best ...

Like many engineers, my first models were based on Designed Experiments in the tradition of Cox and Montgomery. I hadn’t seen anything like a causal diagram until I picked the The Book of Why which explores all sorts of experimental relationships and structures I never imagined.1 Colliders, confounders, causal ...

In this post I’ll be attempting to leverage the parsnip package in R to run through some straightforward predictive analytics/machine learning. Parsnip provides a flexible and consistent interface to apply common regression and classification algorithms in R. I’ll be working with the Cleveland Clinic Heart Disease dataset ...

I’ve never really worked much with Poisson data and wanted to get my hands dirty. I thought that for this project I might combine a Poisson data set with the simple Bayesian methods that I’ve explored before since it turns out the Poisson rate parameter lambda also has ...

When doing comparative testing it can be tempting to stop when we see the result that we hoped for. In the case of null hypothesis significance testing (NHST), the desired outcome is often a p-value of __ .05. In the medical device industry, bench top testing can cost a lot of money. ...

Suppose our team is preparing to freeze a new implant design. In order to move into the next phase of the PDP, it is common to perform a suite of formal “Design Freeze” testing. If the results of the Design Freeze testing are acceptable, the project can advance from Design ...

As engineers, it is not uncommon to be asked to determine whether or not two different configurations of a product perform the same. Perhaps we are asked to compare the durability of a next-generation prototype to the current generation. Sometimes we are testing the flexibility of our device versus a ...