US 11,989,629 B2
Systems and methods for predicting analyte concentrations via machine learning techniques
Nicole Leilani Ing, Emeryville, CA (US); and Glenn Clifford Forrester, Oakland, CA (US)
Assigned to Metre, Inc., Oakland, CA (US)
Filed by Metre, Inc., Oakland, CA (US)
Filed on Jun. 22, 2021, as Appl. No. 17/355,157.
Application 17/355,157 is a continuation of application No. 17/067,573, filed on Oct. 9, 2020, granted, now 11,068,803.
Prior Publication US 2022/0114485 A1, Apr. 14, 2022
This patent is subject to a terminal disclaimer.
Int. Cl. G06N 20/00 (2019.01); G01N 27/49 (2006.01); G01N 33/497 (2006.01)
CPC G06N 20/00 (2019.01) [G01N 27/49 (2013.01); G01N 33/497 (2013.01)] 20 Claims
OG exemplary drawing
 
1. A computer-implemented method, comprising:
receiving sensor data associated with at least one sensor;
utilizing at least a portion of the sensor data of a respective sensor as input to a machine learning network trained at least on a cross-point observation indicating a response curve location at which response curves at the respective sensor intersect, the response curves corresponding to exposure of the respective sensor to multiple training samples of a same amount of an analyte concentration at a same particular temperature, wherein a variable amount of carbon dioxide concentrations are present across the multiple training samples; and
receiving output from the machine learning network based in part on exposure of the at least one sensor to a present sample having a given concentration of the analyte and a given concentration carbon dioxide.