Search

Brachytherapy is a critical component of the standard-of-care curative radiotherapy regimen for women with locally advanced cervical cancer (LACC). However, existing literature suggests that many patients will not receive the brachytherapy boost. We used machine learning (ML) and explainable artificial intelligence to characterize this disparity. Patients with LACC diagnosed from 2004 to 2020 who received definitive radiation were identified in the National Cancer Database. Five ML models were trained to predict if a patient received a brachytherapy boost. The best-performing model was explained using SHapley Additive exPlanation (SHAP) values. To identify trends that may be attributable to the coronavirus disease 2019 (COVID-19) pandemic, the previous analysis was repeated and limited to 2019 to 2020. A total of 37,564 patients with LACC were identified; 5799 were diagnosed from 2019 to 2020 (COVID cohort). Of these patients, 59.3% received a brachytherapy boost, with 76.4% of patients diagnosed in 2019 to 2020 receiving a boost. The random forest model achieved the best performance for both the overall and COVID cohorts. In the overall cohort, the most important predictive features were the year of diagnosis, stage, age, and insurance status. In the COVID cohort, the most important predictive features were FIGO stage, age, insurance status, and hospital type. Of the 26 patients who tested positive for COVID-19 during their course of radiotherapy, 19 (73.1%) received a brachytherapy boost. A gradual increase in brachytherapy boost utilization has been noted, which did not seem to be significantly impacted by the onset of the COVID-19 pandemic. ML could be considered to identify patient populations where brachytherapy is underutilized, which can provide actionable feedback for improving access. [ABSTRACT FROM AUTHOR]

Simple Summary: Lung cancer management continues to evolve with improvements in survival across all stages. The review highlights the current data supporting screening and discusses barriers to screening and potential opportunities to improve screening access. Further, the review discusses the current challenges in diagnosis and biomarker testing in early stage lung cancer and ways in which these can be improved upon. Management of lung cancer has transformed over the past decade and is no longer considered a singular disease as it now has multiple sub-classifications based on molecular markers. The current treatment paradigm requires a multidisciplinary approach. One of the most important facets of lung cancer outcomes however relies on early detection. Early detection has become crucial, and recent effects have shown success in lung cancer screening programs and early detection. In this narrative review, we evaluate low-dose computed tomography (LDCT) screening and how this screening modality may be underutilized. The barriers to broader implementation of LDCT screening is also explored as well as approaches to address these barriers. Current developments in diagnosis, biomarkers, and molecular testing in early-stage lung cancer are evaluated as well. Improving approaches to screening and early detection can ultimately lead to improved outcomes for patients with lung cancer. [ABSTRACT FROM AUTHOR]

The complexity of cancer care requires integrated and continuous support to deliver appropriate care. An expert network with complementary expertise and the capability of multidisciplinary care is an integral part of contemporary oncology care. Appropriate infrastructure is necessary to empower this network to deliver personalized precision care to their patients. Providing decision support as cancer care becomes exponentially more complex with new diagnostic and therapeutic choices remains challenging. City of Hope has developed a Pyramidal Decision Support Framework to address these challenges, which were exacerbated by the COVID pandemic, health plan restrictions, and growing geographic site diversity. Optimizing efficient and targeted decision support backed by multidisciplinary cancer expertise can improve individual patient treatment plans to achieve improved care and survival wherever patients are treated. [ABSTRACT FROM AUTHOR]

Brachytherapy (BT) is a critical component of the standard-of-care curative regimen for women with locally advanced cervical cancer (LACC). However, existing literature suggests a sizeable proportion of patients will not receive the BT boost. Although predictive models such as logistic regression have been able to help identify characteristics of patients who are most at-risk for omission of BT boost, its predictive performance could be improved. Therefore, we utilized machine learning (ML) combined with explainable artificial intelligence to characterize this treatment disparity, with an emphasis on more recent trends that might be attributable to the COVID-19 pandemic. Patients with LACC diagnosed from 2004-2020 were identified in the National Cancer Database. The query was subsequently limited to patients who received definitive radiation. Five ML models were trained to predict if a patient received a BT boost. Model performance, measured using area under the curve (AUC), was compared to multivariable logistic regression. The best-performing model was explained using SHapley Additive exPlanation (SHAP) values. To permit more traditional statistical understanding, SHAP values were converted into odds ratios (OR) and used to identify how model features influenced predictions. To identify trends that may be attributable to the COVID-19 pandemic, the previous analysis was repeated and limited to the years 2019-2020. A total of 54,896 patients with LACC were identified; 8298 were diagnosed from 2019-2020. Of these patients, 54.8% received a BT boost, with 70.5% of patients diagnosed in 2019-2020 receiving a boost. The majority of the increase occurred from 2018 onward, with rates every year prior being a maximum of 57.0%. The Extreme Gradient Boosting model outperformed all other models with an AUC of 0.61 (95%CI: 0.57-0.65) compared to 0.56 (95%CI: 0.52-0.60) for logistic regression. SHAP values > zero represent greater probability of BT boost. SHAP summary plots of the ML models from 2004-2020 and 2019-2020 are shown in Figure 1A and 1F, respectively. SHAP dependence plots of the ML models from 2004-2020 and 2019-2020 are shown in Figure 1B-E and 1G-J, respectively. Overall, a more recent year of diagnosis (OR: 1.08; 95%CI: 1.07-1.08; p<0.001) and lower Charlson Deyo score (2+ vs 0; OR: 1.27; 95%CI: 1.17-1.38; p<0.001) were strong predictors of receiving a BT boost. Other factors associated with BT boost receipt included younger age, private insurance, and white (vs black) race. When examining the years 2019-2020, year of diagnosis (OR: 0.98; 95%CI: 0.92-1.01; p=0.31) and Charlson Deyo score (2+ vs 0; OR: 1.07; 95%CI: 0.99-1.22; p=0.19) were no longer significant. It does not appear that a COVID-19 diagnosis during radiation adversely impacted rates of BT utilization, with 24 of 34 (70.5%) patients who tested positive for COVID-19 during their radiation course receiving a boost. There were slight delays reflected in total treatment duration relative to national recommendations of 8 weeks, lasting a median of 10.7 weeks. This study illustrates how ML can be used to examine practice patterns and identify areas of underutilization. ML approaches yielded improved performance, relative to logistic regression, at predicting receipt of BT boost. To our knowledge, this is the first time SHAP values have been converted to OR, CI, and p-values to describe such models. An increase in BT boost utilization has been noted since 2018, which fortunately did not seem to be significantly impacted by the onset of the COVID-19 pandemic. ML should be considered to better identify health disparities across patient populations where BT is underutilized, which can provide actionable feedback for improving access. [ABSTRACT FROM AUTHOR]