Artificial intelligence for early cancer detection

A method comprising:

• subjecting a blood sample of a canine subject to hypotonic lysis;
• obtaining a plurality of cells from the blood sample after the lysis, the plurality of cells comprising at least one of circulating tumor cells or circulating tumor-associated cells, wherein the circulating tumor cells comprise hemangiosarcoma (HSA) cells, and wherein the circulating tumor-associated cells comprise HSA-associated cells further comprising at least one of activated endothelial cells, platelet-coated leukocytes, or cells mobilized from a pathological vascular niche;
• obtaining a plurality of data features for the plurality of cells from the blood sample of the canine subject;
• applying each trained machine learning model of at least three a plurality of machine learning models to at least a respective subset of a corresponding plurality of respective subsets of the plurality of data features for the plurality of cells from the blood sample of the canine subject,
• wherein at least two respective subsets of the corresponding plurality of respective subsets of the plurality of data features are different, and
• wherein at least two respective subsets of the corresponding plurality of respective subsets of the plurality of data features are the same;
• generating, based on the application of each trained machine learning model of the plurality of at least three machine learning models to at least the respective subset of the corresponding plurality of respective subsets of the plurality of data features, a plurality of classifications, each classification corresponding to a respective trained machine learning model of the plurality of at least three trained machine learning models;
• and
• outputting, based on a determination that a threshold quantity of trained machine learning models resulted in a same one classification for the blood sample, the same one classification for the blood sample, wherein the classification is selected from a plurality of classification options comprising at least a high risk of hemangiosarcoma HSA classification and a low risk of HSA classification.