Figure 5. Specificity of miRNA signature for ovarian cancer compared to other diagnoses.

The neural network 14 miRNA signature did not separate any other diagnoses from the control group in the published dataset by Keller, et al ¹³. The study also included 70 healthy controls. The number of subjects (n) denotes the number of cases of the given diagnosis in the Keller, et al dataset. (a) Pancreatic ductal cancer (n = 45); (b) Prostate cancer (n = 23); (c) Stomach cancer (n = 13); (d) Other pancreatic cancers (n = 48); (e) Melanoma (n = 35); (f) Lung cancer (n = 32); (g) Periodontitis (n = 18); (h) Pancreatitis (n = 38); (i) Multiple sclerosis (n = 23); (j) Acute MI (n = 20); (k) Chronic obstructive pulmonary disease (n = 24); (l) Sarcoidosis (n = 45). (m) Overall, neural network was highly specific for ovarian cancer cases against all other diagnoses (i.e. healthy controls or other cancers).