MD, Dr. Sci. (Medicine), Professor
д-р мед. наук, профессор
Cancer treatment advances are clear, but screening potential remains underutilized due to deficiencies in primary care. Early diagnosis reduces mortality; however, how to integrate AI, biomarkers, and mobile apps into resource-constrained polyclinics is unclear.
This PRISMA-systematic review is the first to comprehensively analyze the efficacy, safety, and cost-effectiveness of screening for five cancers (breast, colon, lung, stomach, and skin) in primary care. We searched PubMed, eLibrary, and the Cochrane Library for 2021–2026 publications, including RCTs, systematic reviews, and cohort studies, with quality assessment using QUADAS-2 and ROBINS-I. Late diagnosis is the key barrier: in some Russian regions, up to 80% of cases are detected at stages III–IV, versus the national average of ~35% (2023–2024), with several regions substantially exceeding the average rate). The coverage of medical examinations in the regions does not exceed 26.8%. A regional study (Voronezh Oblast) found that >50% of primary care physicians had incorrect knowledge of risk factors; extrapolation to all of Russia is limited. For innovative methods, sensitivity reached up to 95% (low-dose chest CT), and specificity ranged from 84% (AI-based skin applications) to 96.6% (low-dose CT). With a 2% disease prevalence in high-risk groups, the positive predictive value is 22–25%, leaving a false-positive problem (false-positive rates in population screening can reach 75–96%, depending on the cancer site). Cascade algorithms are cost-effective, but published models ignore false-positive costs and psychological harm; moreover, conclusions on cascade effectiveness are based mainly on economic modeling, not randomized clinical trials. Pilot projects show that digital tools (AI platforms and mobile applications) increase the detection of early stages by 1%–8% and screening coverage 2.5-fold.
A conceptual personalized screening pathway model was developed and requires further validation. The authors favor systemic restructuring—integrating evidence-based algorithms, workforce strengthening, and digital infrastructure—over fragmented technology adoption. The full text provides a detailed stepwise plan for implementing cascade personalized screening in polyclinics under resource constraints.
Cancer treatment advances are clear, but screening potential remains underutilized due to deficiencies in primary care. Early diagnosis reduces mortality; however, how to integrate AI, biomarkers, and mobile apps into resource-constrained polyclinics is unclear.
This PRISMA-systematic review is the first to comprehensively analyze the efficacy, safety, and cost-effectiveness of screening for five cancers (breast, colon, lung, stomach, and skin) in primary care. We searched PubMed, eLibrary, and the Cochrane Library for 2021–2026 publications, including RCTs, systematic reviews, and cohort studies, with quality assessment using QUADAS-2 and ROBINS-I. Late diagnosis is the key barrier: in some Russian regions, up to 80% of cases are detected at stages III–IV, versus the national average of ~35% (2023–2024), with several regions substantially exceeding the average rate). The coverage of medical examinations in the regions does not exceed 26.8%. A regional study (Voronezh Oblast) found that >50% of primary care physicians had incorrect knowledge of risk factors; extrapolation to all of Russia is limited. For innovative methods, sensitivity reached up to 95% (low-dose chest CT), and specificity ranged from 84% (AI-based skin applications) to 96.6% (low-dose CT). With a 2% disease prevalence in high-risk groups, the positive predictive value is 22–25%, leaving a false-positive problem (false-positive rates in population screening can reach 75–96%, depending on the cancer site). Cascade algorithms are cost-effective, but published models ignore false-positive costs and psychological harm; moreover, conclusions on cascade effectiveness are based mainly on economic modeling, not randomized clinical trials. Pilot projects show that digital tools (AI platforms and mobile applications) increase the detection of early stages by 1%–8% and screening coverage 2.5-fold.
A conceptual personalized screening pathway model was developed and requires further validation. The authors favor systemic restructuring—integrating evidence-based algorithms, workforce strengthening, and digital infrastructure—over fragmented technology adoption. The full text provides a detailed stepwise plan for implementing cascade personalized screening in polyclinics under resource constraints.