Early diagnosis is critical for improving outcomes in cancer patients; however, the application of diagnostic markers derived from serum proteomic screening remains challenging. Artificial intelligence (AI), encompassing deep learning and machine learning (ML), has gained increasing prominence across various scientific disciplines. In this study, we utilized cervical cancer (CC) as a model to develop an AI-driven pipeline for the identification and validation of serum biomarkers for early cancer diagnosis, leveraging mass spectrometry-based proteomics data. By processing and normalizing serum polypeptide differential peaks from 240 patients, we employed eight distinct ML algorithms to classify and analyze these differential polypeptide peaks, subsequently constructing receiver operating characteristic (ROC) curves and confusion matrices. Key performance metrics, including accuracy, precision, recall, and F1 score, were systematically evaluated. Furthermore, by integrating feature importance values, Shapley values, and local interpretable model-agnostic explanation (LIME) values, we demonstrated that the diagnostic area under the curve (AUC) achieved by our multi-dimensional learning models approached 1, significantly outperforming the diagnostic AUC of single markers derived from the PRIDE database. These findings underscore the potential of proteomics-driven integrated machine learning as a robust strategy to enhance early cancer diagnosis, offering a promising avenue for clinical translation.

Background: Lung cancer is one of the leading causes of cancer-related mortality worldwide. Due to the lack of specific symptoms at early stages, most patients are diagnosed at advanced stages, underscoring the need for reliable non-invasive early detection strategies beyond invasive biopsy and non-specific imaging. Methods: We integrated serum proteomics with multi-algorithm machine learning to establish an artificial intelligence–based workflow for early lung cancer diagnosis. Serum samples from 774 participants were analyzed using matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS) to generate peptidomic profiles. After data preprocessing and normalization, eight machine learning algorithms were applied for model construction, and model performance was systematically evaluated using the receiver operating characteristic (ROC) curve. Results: Most machine learning models showed strong diagnostic performance in independent test sets. By combining feature importance, Shapley additive explanations (SHAP), and local interpretable model-agnostic explanations (LIME), nine lung cancer–associated peptide peaks were identified. Multidimensional models based on these peptide signatures achieved AUC values approaching 1, significantly outperforming single markers derived from the PRIDE database. Candidate proteins were further validated through integrative transcriptomic and proteomic analyses, supporting their biological relevance. Conclusions: Our findings indicate that serum proteomics combined with multi-algorithm machine learning enables accurate and non-invasive early detection of lung cancer, providing a promising framework for clinical translation.