<![CDATA[Background: Corruption in public procurement remains a systemic challenge, particularly in the health sector, where irregularities can undermine access to services and their quality. Although previous studies have examined corruption risks and anomaly detection methods, limited empirical evidence exists on how data-driven approaches can be systematically applied to identify atypical contractual patterns in developing countries using large-scale administrative data. Objective: This study aims to propose a methodological framework for identifying potential signals of corruption in the public procurement of health services in Colombia, using administrative data from the SECOP II platform. Methods: A quantitative correlational research design was employed. Data from the SECOP II public procurement platform were analyzed to examine the relationship between contract duration and contract value in medical service contracts. Exploratory analytics techniques and linear regression analysis were used to detect atypical contractual patterns across municipalities and departments. Results: The analysis identified territorial variations in contractual behavior. Three departments concentrated contracts with the highest values relative to their duration. Bogotá exhibited the most pronounced increase in awarded contract values, while the Atlántico department presented the highest volume of contracts. Regression results indicated a statistically significant positive relationship between contract duration and contract value, with a Pearson correlation coefficient of 0.77. Conclusion: The findings demonstrate the potential of data-driven analytical approaches to identify atypical procurement patterns that may serve as warning signals for oversight bodies. The proposed framework contributes to the development of technological tools that support transparency, accountability, and evidence-based monitoring of public procurement processes in the health sector. Keywords: analytics; corruption; public procurement; health sector; anomaly detection]]>
