US 12,380,964 B2
Convolutional neural network systems and methods for data classification
Virgil Nicula, Cupertino, CA (US); Anton Valouev, Palo Alto, CA (US); Darya Filippova, Sunnyvale, CA (US); Matthew H. Larson, San Francisco, CA (US); M. Cyrus Maher, San Mateo, CA (US); Monica Portela dos Santos Pimentel, San Jose, CA (US); Robert Abe Paine Calef, Redwood City, CA (US); and Collin Melton, Menlo Park, CA (US)
Assigned to GRAIL, Inc., Menlo Park, CA (US)
Filed by GRAIL, Inc., Menlo Park, CA (US)
Filed on Aug. 31, 2023, as Appl. No. 18/240,489.
Application 18/240,489 is a continuation of application No. 17/936,529, filed on Sep. 29, 2022, granted, now 11,783,915.
Application 17/936,529 is a continuation of application No. 16/428,575, filed on May 31, 2019, granted, now 11,482,303, issued on Oct. 25, 2022.
Claims priority of provisional application 62/679,746, filed on Jun. 1, 2018.
Prior Publication US 2024/0062849 A1, Feb. 22, 2024
This patent is subject to a terminal disclaimer.
Int. Cl. G06N 3/04 (2023.01); G06N 3/084 (2023.01); G16B 30/10 (2019.01); G16B 40/20 (2019.01); G16B 40/30 (2019.01); G16H 50/20 (2018.01)
CPC G16B 30/10 (2019.02) [G06N 3/04 (2013.01); G06N 3/084 (2013.01); G16B 40/20 (2019.02); G16B 40/30 (2019.02); G16H 50/20 (2018.01)] 20 Claims
OG exemplary drawing
 
1. A method of diagnosing a disease state in a test subject by using a trained artificial neural network, the method comprising:
obtaining, from a biological sample associated with the test subject, sequencing data derived from a methylation sequencing assay of cell-free nucleic acids in the biological sample;
applying, subsequent to the obtaining, the sequencing data to the trained artificial neural network;
determining, subsequent to the applying and by processing the sequencing data using a plurality of filter sets resident within a plurality of convolutional layers of the trained artificial neural network, whether a methylation profile of the sequencing data is detected that is indicative of a disease state, wherein:
a first filter of the plurality of filter sets is configured to identify a methylation pattern at a single methylation site in a genomic region;
a second filter of the plurality of filter sets is configured to identify a relationship between each of the single methylation sites in the genomic region; and
wherein the trained artificial neural network is configured to integrate outputs from the first filter and outputs from the second filter to generate the methylation profile; and
providing, on a display screen of a computing device and based on the determining, a diagnosis for the test subject with respect to the disease state.