NOTE: This post highlights early testing data from the development of the Nima Gluten Sensor. The final validation data can be reviewed on our science page.
One of our frequently asked questions from buyers and potential buyers is around Nima’s accuracy. Since we’re still in pre-production and working toward shipping our pre-order units this fall, we are doing on-going testing of the units that come off the manufacturing line to make sure that Nima is functioning at the level we originally targeted. At the launch of our pre-sale last October, our goal was to have Nima’s accuracy at 99.5 percent to detect 20 parts per million of gluten in a food sample. We conducted a study off our latest round of test units to demonstrate its ability to consistently detect 20 ppm of gluten in a food.
For this study, 200 cornbread samples were tested, each individually prepared to contain 20 ppm of gluten. Bob’s Red Mill gluten-free cornbread mix was chosen as the starting ingredient. The gluten-free cornbread mix was prepared and baked and then tested with Nima’s chemistry to ensure the mix was gluten-free. Once the mix was verified to be gluten-free, the next step was to prepare 200 samples with the addition of gluten to achieve a final concentration of 20 ppm.
Six Nima devices were used for testing, with the 200 samples evenly spread between the units. Each of the samples were weighed and then added to each capsule. The capsules were then run on the devices and the results recorded.
In the 200 tests performed, 195 tests on Nima accurately reported a ‘gluten found’ result (gluten symbol). Four tests reported an indeterminate result (exclamation point), while one test reported an incorrect gluten-free result (smile symbol).
The majority of tests performed as designed. Only one test in 200 incorrectly displayed a gluten-free smile result, representing a 99.5 percent accuracy rate. In this case, the erroneous result was caused by an internal failure.
There were four tests that resulted in indeterminate results. These tests were evaluated and, of the four, three were found to be caused by a mechanical failure. For the remaining indeterminate test result, insufficient mixing was reported by the device.
We are highly encouraged that we are able achieve the goal we initially set for Nima’s accuracy, but we know that our work is not done yet. In all five cases (four indeterminate and one incorrect result), the underlying failures will be addressed in the next round of manufacturing to ensure a higher level of performance. We are actively working out the kinks with improved manufacturing process controls to decrease the mechanical failures of the device before we ship.
In the near future, testing will also be done in partnership with a third party to validate Nima’s accuracy. As always, results will be posted here and on the website as soon as we have them.
Read the full whitepaper/report with more details on methodology below or here, and as always, reach out to us with any questions at firstname.lastname@example.org.