Research Article
Validation of a Machine Learning Approach for Venous Thromboembolism Risk Prediction in Oncology
Table 1
Clinical and laboratory attributes of the patient dataset (
).
| Demographics | | Age, mean ± SD (range) | 63 ± 12 (18–88) | Sex | | Males | 293 (48%) | Females | 315 (52%) | BMI, mean ± SD | 25.2 ± 4.4 |
| Primary tumor | | Colorectal | 155 (25%) | Gastric | 28 (5%) | Esophageal | 10 (2%) | Pancreatic | 21 (3%) | Biliary | 4 (1%) | Lung | | Non-small cell | 81 (13%) | Small cell | 15 (3%) | Breast | 149 (24%) | Prostate | 31 (5%) | Ovarian | 16 (3%) | Genitourinary | 42 (7%) | Head-neck | 23 (4%) | Sarcoma | 7 (1%) | Unknown | 7 (1%) | Other | 19 (3%) | Stage of disease | | Primary | 253 (42%) | Relapsing/metastatic | 355 (58%) |
| Anticancer drugs | | Platinum compounds | 290 (48%) | Fluoropyrimidine | 213 (35%) | Anthracycline | 87 (14%) | Taxanes | 87 (14%) | Paclitaxel | 58 (10%) | Bevacizumab | 80 (13.2%) | Gemcitabine | 68 (11%) | Irinotecane | 79 (13%) | Pemetrexed | 38 (6%) | Herceptin | 36 (6%) | Antityrosine kinase | 16 (3%) | Aromatase inhibitors | 60 (10%) | Supportive drugs, N (%) | Erythropoiesis stimulating agents | 11 (2%) | Prophylactic myeloid growth factors | 18 (3%) | Corticosteroids | 109 (18%) |
| ECOG-PS, N (%) | | 0 | 431 (71%) | 1 | 158 (26%) | 2 | 19 (3%) |
| Hematology and biochemical attributes | | Blood cell counts | | Red blood cells | 4.5 ± 0.8 | Hematocrit | 36.6 ± 7.6 | Hemoglobin | 12.5 ± 1.9 | White blood cells | 7.7 ± 3.5 | Neutrophils | 5.2 ± 3.1 | Lymphocytes | 1.8 ± 1.0 | Platelets | 261 ± 102 | Mean platelet volume | 8.6 ± 1.0 | Neutrophil-lymphocyte ratio | 4.0 ± 4.4 | Platelet-lymphocyte ratio | 185 ± 145 | Routine blood chemistry | | Blood urea nitrogen | 38 ± 17 | Creatinine | 0.9 ± 0.3 | eGFR | 89.8 ± 28.4 | Glucose | 110 ± 39 | Insulin | 28 ± 26 | HbA1c | 6.0 ± 0.9 | Total bilirubin | 0.6 ± 0.5 | Alanine transaminase | 24.0 ± 20.0 | Aspartate transaminase | 25.4 ± 23.1 | γ-Glutamyl transferase | 69 ± 143 | Triglycerides | 139 ± 82 | Total cholesterol | 197 ± 52 | High-density lipoproteins | 48.0 ± 14.1 | Low-density lipoproteins | 123.1 ± 42.1 |
| Venous thromboembolism | | Pulmonary embolism | 11 (1.8%) | Deep venous thrombosis | 32 (5.3%) | Median time-to-event (months) | 2.5 months |
|
|
BMI: body mass index; ECOG-PS: Eastern Cooperative Oncology Group Performance Status. eGFR: estimated glomerular filtration rate. Including mesothelioma (), melanoma (), neuroendocrine tumors (), glioblastoma (), small intestine (), liver (), and one skin cancer. 11% neoadjuvant, 32% adjuvant, and 57% metastatic treatments.
|