|The following article features coverage from the American Society of Clinical Oncology 2020 virtual meeting. Click here to read more of Oncology Nurse Advisor‘s conference coverage.
A machine-learning algorithm was able to predict the rate of financial stress on patients who underwent treatment for their breast cancer, according to a retrospective survey and algorithm-modeling study. These findings were presented during the ASCO 2020 Virtual Scientific Program.
Six hundred and eleven patients who had undergone treatment for breast cancer at University of Texas MD Anderson Cancer Center in Houston, Texas, were retrospectively surveyed. All patients were adults who underwent either mastectomy or lumpectomy. Data were collected as a FACT-COST patient-reported outcome measurements as well as other financial indicators including insurance status and income. Missing data were imputed by multiple imputation. A LASSO-regularized linear model was used to train and validate their neural network, in which instances were randomly assigned to training or validation cohorts in a 3:1 ratio.
A minority of women (48; 23.6%) reported financial burden due to their cancer treatment. The machine learning algorithm predicted financial burden with a high accuracy (83%), sensitivity (81%), and specificity (82%), and area under the receiver operating curve (0.82)
The investigators identified the 2 key predictors of financial burden as neo-adjuvant therapy (βregularized 0.12) and autologous reconstruction (βregularized 0.10).
The study authors concluded that their machine learning model could accurately predict financial difficulties due to treatment of breast cancer. These predictions may aide in the decision-making process and that with careful planning, financial distress may be avoided. As financial toxicity is associated with poorer clinical outcomes, avoiding this stressor would ultimately lead to better quality of care.
Sidey-Gibbons C, Asaad M, Pfob A, Boukovalas S, Lin YL, Offodile. Machine learning algorithms to predict financial toxicity associated with breast cancer treatment. Presented at: ASCO20 Virtual Scientific Program. J Clin Oncol. 2020;38(suppl):abstr 2047.