CONVOLUTIONAL NEURAL NETWORK SYSTEMS AND METHODS FOR DATA CLASSIFICATION
Classification of cancer condition, in a plurality of different cancer conditions, for a species, is provided in which, for each training subject in a plurality of training subjects, there is obtained a cancer condition and a genotypic data construct including genotypic information for the respective training subject. Genotypic constructs are formatted into corresponding vector sets comprising one or more vectors. Vector sets are provided to a network architecture including a convolutional neural network path comprising at least a first convolutional layer associated with a first filter that comprise a first set of filter weights and a scorer. Scores, corresponding to the input of vector sets into the network architecture, are obtained from the scorer. Comparison of respective scores to the corresponding cancer condition of the corresponding training subjects is used to adjust the filter weights thereby training the network architecture to classify cancer condition.
1 - 20 . (canceled)
21 . A computer-implemented method for diagnosing a disease state in a test subject, the computer-implemented 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;
generating, using one or more processors, bin-level methylation data from the sequencing data;
identifying, using the one or more processors, one or more signals in the bin-level methylation data attributable to white blood cells (WBCs);
masking, using the one or more processors, the identified one or more signals from the bin-level methylation data to generate a WBC-masked input data set;
formatting, using the one or more processors, the WBC-masked input data set into one or more vector sets;
applying, using the one or more processors, the one or more vector sets to a trained machine learning model; and
generating, using the one or more processors, an output from the trained machine learning model indicating the disease state of the test subject.
22 . The computer-implemented method of claim 21 , wherein the identifying the one or more signals comprises utilizing a reference methylation profile derived from isolated WBCs.
23 . The computer-implemented method of claim 21 , wherein the masking comprises excluding, from the WBC-masked input data set, bins associated with the identified one or more signals.
24 . The computer-implemented method of claim 21 , wherein the masking comprises applying a weighting function to one or more bins associated with the identified one or more signals, wherein the weighting function reduces an influence of the one or more bins in the WBC-masked input data set.
25 . The computer-implemented method of claim 21 , wherein each of the one or more vector sets comprises a two-dimensional matrix, wherein a first dimension of the two-dimensional matrix corresponds to genomic regions and a second dimension of the two-dimensional matrix corresponds to methylation features.
26 . The computer-implemented method of claim 21 , wherein the trained machine learning model comprises a multi-path convolutional neural network (CNN).
27 . The computer-implemented method of claim 26 , wherein each path of the multi-path CNN is configured to process a designated input data type selected from the group consisting of: copy number variation data, methylation data, or WBC-masked methylation data.
28 . The computer-implemented method of claim 21 , wherein the classification output comprises a probabilistic score associated with each of a plurality of possible disease states.
29 . The computer-implemented method of claim 21 , further comprising outputting a diagnostic result for the test subject based on the classification output.
30 . The computer-implemented method of claim 21 , further comprising generating, based on the output, a graph that compares a performance of the trained machine learning model processing the WBC-masked input data set and an unmasked input data set.
31 . The computer system for diagnosing a disease state in a test subject by using a trained artificial neural network, the computer system comprising:
at least one processor;
a graphical processing unit having a graphical processing memory configured to store a network architecture; and
a memory, the memory storing at least one program for execution by the at least one processor, the at least one program comprising instructions for:
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;
generating bin-level methylation data from the sequencing data;
identifying one or more signals in the bin-level methylation data attributable to white blood cells (WBCs);
masking the identified one or more signals from the bin-level methylation data to generate a WBC-masked input data set;
formatting the WBC-masked input data set into one or more vector sets;
applying the one or more vector sets to a trained machine learning model; and
generating an output from the trained machine learning model indicating the disease state of the test subject.
32 . The computer system of claim 31 , wherein the instructions for identifying the one or more signals comprise instructions for utilizing a reference methylation profile derived from the isolated WBCs.
33 . The computer system of claim 31 , wherein the instructions for masking comprise instructions for excluding, from the WBC-masked input data set, bins associated with the identified one or more signals.
34 . The computer system of claim 31 , wherein the instructions for masking comprise instructions for applying a weighted function to one or more bins associated with the identified one or more signals, wherein the weighting function reduces an influence of the one or more bins in the WBC-masked input data set.
35 . The computer system of claim 31 , wherein each of the one or more vector sets comprises a two-dimensional matrix, wherein a first dimension of the two-dimensional matrix corresponds to genomic regions and a second dimension of the two-dimensional matrix corresponds to methylation features.
36 . The computer system of claim 31 , wherein the trained machine learning model comprises a multi-path convolutional neural network (CNN).
37 . The computer system of claim 36 , wherein each path of the multi-path CNN is configured to process a designated input data type selected from the group consisting of: copy number variation data, methylation data, or WBC-masked methylation data.
38 . The computer system of claim 31 , wherein the classification output comprises a probabilistic score associated with each of a plurality of possible disease states.
39 . The computer system of claim 31 , wherein the instructions further comprise:
outputting a diagnostic result for the test subject based on the classification output.
40 . A non-transitory computer-readable storage medium storing computer-executable instructions which, when executed by a processor, cause the processor to perform operations 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;
generating bin-level methylation data from the sequencing data;
identifying one or more signals in the bin-level methylation data attributable to white blood cells (WBCs);
masking the identified one or more signals from the bin-level methylation data to generate a WBC-masked input data set;
formatting the WBC-masked input data set into one or more vector sets;
applying the one or more vector sets to a trained machine learning model; and
generating an output from the trained machine learning model indicating the disease state of the test subject.