Examining the role of Artificial Intelligence in Age-appropriate Cancer screening: A Scoping review

Background: Artificial intelligence (AI) tools are playing an increasingly important role in improving healthcare across various specialties. This review focuses on the application of AI in age-appropriate cancer screening tests.

Study design: We conducted a scoping review of various AI tools aimed at enhancing screening rates and cancer detection. A comprehensive search of Google Scholar and PubMed identified 26,090 articles. After applying exclusion criteria, 56 studies were selected for detailed review.

Results: Our review identified limited research (1 study) on using AI to directly improve screening rates. However, AI applications showed promise in interpreting mammograms (3 studies), lung cancer screenings (3 studies), pap smears (40 studies), colonoscopies (3 studies), and prostate cancer screenings (6 studies). The virtual patient navigator ‘My Eleanor’ holds promise for increasing colonoscopy screening rates. AI-assisted analysis of mammograms and low-dose chest CT scans can detect more lesions, but this can also lead to a higher rate of false positives. Radiologist review remains crucial. The centaur model, where radiologists collaborate with AI, has the potential to improve cancer detection while reducing false positives. Artificial neural networks (ANNs) have shown promise in reducing false positives for prostate cancer detection by analyzing various other predictive factors for prostate cancer factors alongside free PSA levels. AI plays a significant role in pap smear analysis, with numerous AI tools available. These tools have demonstrated improved sensitivity and specificity in detecting cervical lesions, while also reducing interpretation time compared to manual analysis. Despite its potential, we did not identify any AI tool designed for improving Shared Decision-Making process for age-appropriate cancer screening.

Conclusion: AI has the potential to revolutionize all aspects of age-appropriate cancer screening, from patient education to result interpretation. However, widespread adoption is hindered by the need for larger studies, validation to ensure accuracy and fairness, mitigating algorithmic biases, addressing data privacy and cybersecurity concerns, and overcoming developmental challenges.