Miyawaki, Isabele Ayumi

160 Background: In the United States, racial/ethnic disparities in adoption of breast cancer evidence-based practices (EBPs) persist along the cancer care continuum. This scoping review applied an implementation science lens to describe the current evidence base for interventions to reduce disparities, synthesizing evidence on effective interventions while highlighting gaps in the literature. Methods: A formalized search strategy identified USA-based interventional studies aiming to improve adoption of an EBP along the breast cancer care continuum among racially/ethnically minoritized patients. Relevant articles from 1/2010 - 8/2022 (Medline, Embase, Cochrane, and Scopus) were included. For each article, two reviewers screened and extracted data, with a third resolving discrepancies. The PRISMA-ScR checklist and JBO Manual for Evidence Synthesis framework were followed. Results: Our search strategy yielded 45,215 articles. After applying inclusion/exclusion criteria, 59 articles were retained, targeting screening (85%, 50 of 59), treatment (12%, 7 of 59), and survivorship (5%, 3 of 59) (2 targeted multiple areas). No articles targeted end-of-life care. Of these, 49% (29 of 59) were randomized controlled trials (RCTs). Among 24 screening RCTs, the most common interventions—and fraction with statistically significant improved screening uptake—were patient navigation (100%, 11 of 11) and patient education (80%, 4 of 5). Among 4 treatment RCTs, 3 targeted adjuvant chemotherapy uptake. Of these, 1 had significant improvements and used patient navigation. The 2 studies without significant improvements used a patient tracking tool/alert system or patient education. Among 2 survivorship RCTs, one used email interventions, resulting in significantly improved physical activity and dietary health. A second used a nurse-facilitated internet support group, resulting in significantly improved physical/psychological symptoms. Early to mid-stage implementation outcomes (acceptability, adoption, appropriateness, feasibility, or fidelity) were commonly measured (81%, 48 of 59). No study measured later stage implementation outcomes (sustainability or costs). Conclusions: In multiple studies, patient navigation and education improved breast cancer screening among racially/ethnically minoritized populations. In one study, patient navigation improved completion of adjuvant treatment. Despite these studies, patient navigation is not reimbursed for cancer screening, and adoption has not been widespread. Future studies should focus on evaluating and improving later stage implementation outcomes (sustainability and costs). Based on these results, insurance reimbursement mechanisms for screening-related navigation may be warranted. Future breast cancer care delivery disparities research should also be expanded to include treatment, survivorship, and end-of-life care.

Objective Deep learning (DL) has shown promising results for improving mammographic breast cancer diagnosis. However, the impact of artificial intelligence (AI) on the breast cancer screening process has not yet been fully elucidated in terms of potential workload reduction. We aim to assess if AI-based triaging of breast cancer screening mammograms could reduce the radiologist's workload with non-inferior sensitivity. Methods PubMed, EMBASE, Cochrane Central, and Web of Science databases were systematically searched for studies that evaluated AI algorithms on computer-aided triage of breast cancer screening mammograms. We extracted data from homogenous studies and performed a proportion meta-analysis with a random-effects model to examine the radiologist's workload reduction (proportion of low-risk mammograms that could be theoretically ruled out from human's assessment) and the software's sensitivity to breast cancer detection. Results Thirteen studies were selected for full review, and three studies that used the same commercially available DL algorithm were included in the meta-analysis. In the 156,852 examinations included, the threshold of 7 was identified as optimal. With these parameters, radiologist workload decreased by 68.3% (95%CI 0.655–0.711, I² = 98.76%, p < 0.001), while achieving a sensitivity of 93.1% (95%CI 0.882–0.979, I² = 83.86%, p = 0.002) and a specificity of 68.7% (95% CI 0.684–0.723, I² = 97.5%, p < 0.01). Conclusions The deployment of DL computer-aided triage of breast cancer screening mammograms reduces the radiology workload while maintaining high sensitivity. Although the implementation of AI remains complex and heterogeneous, it is a promising tool to optimize healthcare resources.

Abstract Introduction: Deep learning (DL) has shown promising results for mammographic breast cancer diagnosis. However, the impact of artificial intelligence (AI) in the screening process have not yet been fully addressed in terms of workload reduction, which has potential to decrease healthcare disparities. Radiologists are tasked with overwhelming volumes of screening mammograms, particularly in medically underserved areas. Therefore, when applied to workflow improvement, AI might be a tool to reduce healthcare disparities. The purpose of this systematic review and meta-analysis was to assess if AI-based triaging of breast cancer screening mammograms could reduce the radiologist’s workload with non-inferior sensitivity. Methods: PubMed, EMBASE, Cochrane Central and Web of Science databases were systematically searched for studies that evaluated AI algorithms on computer-aided triage of breast cancer screening mammograms. We extracted data from homogenous studies and performed a proportion meta-analysis with random-effects model to verify the radiologist’s workload reduction and the software’s sensitivity. Results: A total of 14 studies were systematically selected. Three studies using the same commercially available DL algorithm were included in the meta-analysis, with 156852 examinations evaluated at the threshold of 8. The radiologist’s workload decreased by 68.3% (95%CI 0.655-0.711, I² = 98.76%, p &lt; 0.001), with a sensitivity of 93.1% (95%CI 0.882-0.979, I² = 83.86%, p = 0.002). Conclusion: Our findings suggest that DL computer-aided triage of breast cancer screening mammograms significantly reduces the radiologist’s workload with high sensitivity. Although AI’s implementation remains complex and heterogeneous, it is a promising tool to optimize healthcare resources with a potential large impact in low resource settings that struggle with workforce shortage. Citation Format: Debora Xavier, Isabele A. Miyawaki, Carlos Alberto Campello Jorge, Matheus Jose Barbosa Moreira, Bruno M. Carvalho, Felipe Batalini. Artificial intelligence-based triaging of breast cancer screening mammograms and radiologist workload reduction: a systematic review and meta-analysis [abstract]. In: Proceedings of the 2022 San Antonio Breast Cancer Symposium; 2022 Dec 6-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2023;83(5 Suppl):Abstract nr P3-04-06.