Artificial intelligence in early diagnosis: integration of pre-nosological screening and personalized prevention of chronic non-communicable diseases

Abstract

Introduction. Chronic non-communicable diseases (NCDs) account for 75% of global mortality, while traditional treatment paradigm demonstrates inability to contain epidemiological burden. Artificial intelligence (AI) technologies combined with telemedicine enable healthcare transformation: from reactive treatment to proactive health management through personalized prevention. Russian school of pre-nosological diagnostics, focused on identifying pre-pathological states through assessment of body’s functional reserves, creates methodological foundation for personalized approach that can be significantly enhanced by modern machine learning methods. Objective: to develop methodology for remote questionnaire-based screening of NCDs using AI with integration of holistic approach to pre-nosological diagnostics, providing generation of personalized prevention recommendations, and evaluate its effectiveness in young adults. Material and methods. Study included 3,155 university students from St. Petersburg (mean age 19.6±1.5 years) from 83 regions of Russian Federation. AI-based technology for remote screening was developed using holistic approach. System verifies risk factors by five pathology profiles (cardiology, gastroenterology, pulmonology, endocrinology, oncology). Questionnaire contains 198 information requests. Decision rules system (1,098 rules) was applied. Systematic literature review in PubMed, Scopus, Web of Science, eLibrary for 2020–2025 was conducted; RCTs, systematic reviews, WHO and Food and Drug Administration regulatory documents, methodological guidelines were analyzed. Results. Low NCD risk detected in 57.4%, moderate in 30.9%, high in 11.7% of examined individuals. Most frequent complaints related to endocrine (28.9%), digestive (21.8%), respiratory (21.1%), and cardiovascular systems (20.1%). More than 75% showed signs of polymorbidity. Statistical analysis confirmed significant consistency between system and physician assessments (p < 0.001). Cohen’s kappa showed substantial agreement for cardiology and pulmonology profiles, moderate for gastroenterology and endocrinology. System generates personalized recommendations considering age, gender, anthropometric data, harmful habits, and psychological state. Physician time savings reached 20%. User satisfaction – 96.6%, healthcare workers – 91.7%. Conclusion. Developed methodology for remote questionnaire-based AI screening with holistic approach showed high effectiveness for early risk factor detection in young adults. Integration of Russian pre-nosological diagnostics experience through pathology profiles with modern machine learning technologies creates conditions for transition to personalized prevention focused on correction of body’s functional reserves. System demonstrates significant social and economic effectiveness.

Ähnliche Arbeiten