Know Labs, Inc. (NYSE American: KNW) today announced results from a new study titled, “Novel data preprocessing techniques in an expanded dataset improve machine learning model accuracy for a non-invasive blood glucose monitor.” The study demonstrates that continued algorithm refinement and more high-quality data improved the accuracy of Know Labs’ proprietary Bio-RFID sensor technology, resulting in an overall Mean Absolute Relative Difference (MARD) of 11.3%.
As with all Know Labs’ previous research, this study was designed to assess the ability of the Bio-RFID sensor to non-invasively and continuously quantify blood glucose, using the Dexcom G6® continuous glucose monitor (CGM) as a reference device. In this new study where data collection was completed in May of 2023, Know Labs applied novel data preprocessing techniques and trained a Light Gradient-Boosting Machine (lightGBM) model to predict blood glucose values using 3,311 observations – or reference device values – from over 330 hours of data collected from 13 healthy participants. With this method, Know Labs was able to predict blood glucose in the test set – the dataset that provides a blind evaluation of model performance – with a MARD of 11.3%.
Comparatively, Know Labs released study results in May 2023 that analyzed data from five participants of a similar demographic using 1,555 observations from 130 hours of data collection, and the first application of the lightGBM ML model, which resulted in an overall MARD of 12.9%.
In June 2023, Know Labs announced the completed build of its Gen 1 prototype, which incorporates the Bio-RFID sensor that Know Labs has been using to conduct clinical research in a lab environment for the last two years, and has published results of its proven stability, into a portable device. Testing with the Gen 1 device is underway, optimizing the sensor configuration for data collection, including new environmental and human factors.
The Company’s focus is on collecting more high-quality, high-resolution data across a diverse participant population representing different glycemic ranges and testing scenarios, to refine its algorithms based on this new data, and to optimize its sensor in preparation for scale. To support this work, the Company is continuing to test with its Gen 1 device every day in parallel with ongoing clinical research with its stationary lab system. Gen 1 is expected to generate tens of billions of data observations to process which will be critical to helping validate algorithm performance across the real-world scenarios in which Know Labs’ glucose monitoring device may be used. This is a key component of realizing the Company’s vision for bringing an FDA-cleared product to the market.
For 2023, Know Labs remains focused on refinement and external validation of its technology and contributing to its growing body of peer-reviewed evidence, which can be found at www.knowlabs.co/research-and-validation.