Applying Neural Networks to Analyse Inflammatory, Sociodemographic, and Psychological Factors in Non-Melanoma Skin Cancer and Colon Cancer: A Statistical and Artificial Intelligence Approach

• Published in: Diagnostics (Basel) 2024-12, Vol.14 (23), p.2759
• Main Authors: Mercuţ, Răzvan, Ciurea, Marius Eugen, Traşcă, Emil Tiberius, Ionescu, Mihaela, Mercuţ, Maria Filoftea, Rădulescu, Patricia Mihaela, Călăraşu, Cristina, Streba, Liliana, Ionescu, Alin Gabriel, Rădulescu, Dumitru
• Format: Article
• Language: English
• Subjects: Algorithms; Artificial intelligence; Cancer; Care and treatment; Colon cancer; Colorectal cancer; Comparative analysis; Demographic aspects; Development and progression; Health aspects; Inflammation; inflammatory factors; Medical colleges; Mortality; Neural networks; non-melanoma skin cancer; Oncology; Oncology, Experimental; Physiological aspects; psychological factors; Skin cancer; Sociodemographics; Socioeconomic factors; Stress (Psychology)
• ISSN: 2075-4418; 2075-4418
• DOI: 10.3390/diagnostics14232759

Chronic inflammation and psychosocial factors significantly influence cancer progression and patient behavior in seeking medical care. Understanding their interplay is essential for enhancing early detection and developing personalized treatment strategies. This study aims to develop a comprehensive patient profiling model by comparing non-melanoma skin cancer (NMSC) and colorectal cancer (CRC). The goal is to identify common and distinct patterns in inflammation and psychosocial factors that affect disease progression and clinical presentation. We conducted a comparative analysis of patients diagnosed with NMSC and CRC, integrating clinical data with sociodemographic and psychological assessments. Advanced neural network algorithms were employed to detect subtle patterns and interactions among these factors. Based on the analysis, a cancer risk assessment questionnaire was developed to stratify patients into low-, moderate-, and high-risk categories. Patients with low systemic inflammation and adequate vagal tone, supported by a stable family environment, demonstrated heightened sensitivity to subclinical symptoms, enabling earlier diagnosis and timely intervention. Conversely, patients with high systemic inflammation and reduced vagal tone, often influenced by chronic stress and unstable family environments, presented at more advanced disease stages. The developed risk assessment tool effectively classified patients into distinct risk categories, facilitating targeted preventive measures and personalized therapeutic strategies. Neural network profiling revealed significant interactions between biological and psychosocial factors, enhancing our understanding of their combined impact on cancer progression. The integrated profiling approach and the newly developed risk assessment questionnaire have the potential to transform cancer management by improving early detection, personalizing treatment strategies, and addressing psychosocial factors. This model not only enhances clinical outcomes and patient quality of life but also offers a framework adaptable to other cancer types, promoting a holistic and patient-centered approach in oncology.