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.

Ovarian cancer poses a significant clinical challenge due to its asymptomatic onset and poor prognosis, highlighting the critical need for effective early detection strategies. This study developed a framework that integrates serum proteomic profiling with machine learning algorithms. Serum samples from 188 patients and 208 healthy controls were analysed via Matrix-Assisted Laser Desorption/Ionization Time - of - Flight (MALDI - TOF) mass spectrometry, revealing 43 differentially expressed peptides (17 upregulated, 26 downregulated). An ensemble pipeline incorporating eight machine learning algorithms showed favorable discriminatory ability in the study cohort, with the area under the receiver operating characteristic curve (AUC) of the integrated models approaching 1.00 for distinguishing ovarian cancer samples from healthy controls. We identified three consensus biomarkers [mass-to-charge ratio (m/z) = 4211.41, 2881.50, 2662.15] through feature importance analysis, and their diagnostic reliability was supported by receiver operating characteristic curves optimization and decision curve analysis. Interpretability approaches integrating Shapley values and LIME indicated that the model’s high performance (AUC ≈ 1) was driven by robust multi-dimensional feature contributions. Cross-referencing with existing datasets suggested the PDE11A as a potential diagnostic biomarker. Collectively, this ensemble machine learning algorithms leveraged on serum proteomics shows promising potential for early detection of ovarian cancer, offering a strategy to mitigate the limitations of single-analyte biomarkers.