Machine Learning to Predict Tamoxifen Nonadherence Among US Commercially Insured Patients With Metastatic Breast Cancer

Abstract

PURPOSE Adherence to tamoxifen citrate among women diagnosed with metastatic breast cancer can improve survival and minimize recurrence. This study aimed to use real-world data and machine learning (ML) methods to classify tamoxifen nonadherence. METHODS A cohort of women diagnosed with metastatic breast cancer from 2012 to 2017 were identified from IBM MarketScan Commercial Claims and Encounters and Medicare claims databases. Patients with < 80% proportion of days coverage in the year following treatment initiation were classified as nonadherent. Training and internal validation cohorts were randomly generated (4:1 ratio). Clinical procedures, comorbidity, treatment, and health care encounter features in the year before tamoxifen initiation were used to train logistic regression, boosted logistic regression, random forest, and feedforward neural network models and were internally validated on the basis of area under receiver operating characteristic curve. The most predictive ML approach was evaluated to assess feature importance. RESULTS A total of 3,022 patients were included with 40% classified as nonadherent. All models had moderate predictive accuracy. Logistic regression (area under receiver operating characteristic 0.64) was interpreted with 94% sensitivity (95% CI, 89 to 92) and 0.31 specificity (95% CI, 29 to 33). The model accurately classified adherence (negative predictive value 89%) but was nondiscriminate for nonadherence (positive predictive value 48%). Variable importance identified top predictive factors, including age ≥ 55 years and pretreatment procedures (lymphatic nuclear medicine, radiation oncology, and arterial surgery). CONCLUSION ML using baseline administrative data predicts tamoxifen nonadherence. Screening at treatment initiation may support personalized care, improve health outcomes, and minimize cost. Baseline claims may not be sufficient to discriminate adherence. Further validation with enriched longitudinal data may improve model performance.

A Machine Learning Model to Predict Patients’ Adherence Behavior and a Decision Support System for Patients With Metastatic Breast Cancer: Protocol for a Randomized Controlled Trial

“…The 5-year survival rate of patients with MBC is around 38% [2]. Notwithstanding, several studies [3][4][5][6] have shown that adherence to anticancer treatments is a critical issue in the disease trajectory of patients with breast cancer, especially regarding oral anticancer treatments (OATs), which are intensely demanding because patients are responsible for assuming medications according to the medical prescriptions, thereby increasing the risk of not appropriately taking the therapy [7].…”

“…Accruing evidence have highlighted that patients with advanced cancer might report significant levels of nonadherence because in case of some cancers such as MBC, the patients have to change the type of treatments or dosage frequently and they experience a high level of fear of cancer spread [3]. For example, Yerrapragada and colleagues [6] reported nonadherence to tamoxifen in patients with MBC, ranging between 30% and 85% and further reducing over time. In addition, patients with MBC reported several barriers to the daily management of OAT, such as emotional (eg, worry, depression) and physical distress related to the side effects (eg, fatigue, weakness, sleep disturbance, emotional burden, pain) and lack of knowledge about their disease.…”

“…In addition, patients with MBC reported several barriers to the daily management of OAT, such as emotional (eg, worry, depression) and physical distress related to the side effects (eg, fatigue, weakness, sleep disturbance, emotional burden, pain) and lack of knowledge about their disease. Further, patients might experience a lack of control and a lack of perceived benefits during the disease pathway, and they may experience difficulty in managing therapy [3,6]. Other studies have shown that patients with MBC experience modest QoL due to the related treatment side effects, financial burden affecting therapy discontinuation, and a significant level of nonadherence [4,7].…”

Machine Learning to Predict the Need for Postmastectomy Radiotherapy after Immediate Breast Reconstruction

“…21 ML modeling has been applied for predicting breast cancer outcomes, including risk of diagnosis, 22,23 prognosis, 24 treatment adherence, 25 and treatment morbidity. 26 In this present study, we sought to create an ML model to predict the probability of requiring PMRT after IBR. Secondarily, we aimed to compare the performance of ML to traditional regression methods for prediction modeling in this clinical setting.…”

Use of Recommended Real-World Methods for Electronic Health Record Data Analysis Has Not Improved Over 10 Years

“…[21,22] During phase 2, the analytic phase (the statistical modeling and analysis of the data), many Analytic Real World Methods (ARWM) have been developed to analyze RWD, from the statistical, epidemiological, and computer-science communities. [19,[23][24][25] During phase 3, the post-analytic phase, sensitivity analyses establish the robustness of conclusions to reasonable uncertainty in assumptions. [24][25][26][27] In the current paper, we call the entire set of methods in these 3 phases of research "recommended Real World Methods (recommended RWM).…”