Breast cancer remains the most commonly identified malignant condition worldwide and represents the principal cause of death among women. To address this issue, population-based screening initiatives—primarily mammographic exams—began to appear in the 20th century. Their introduction has substantially lowered mortality and enhanced outcomes for individuals diagnosed with this illness. Even so, reading mammograms involves a certain level of inconsistency and depends heavily on the expertise and background of the clinicians who interpret them. Seeking to assist the assessment of mammographic images and strengthen the diagnostic workflow, this study introduces the conception, development, and proof of concept of an innovative intelligent clinical decision support system based on two predictive strategies executed in parallel. The first strategy uses several expert modules driven by fuzzy inference mechanisms designed to process the attributes linked to the chief mammographic findings. From this, a collection of indicators—termed Symbolic Risks—can be derived, each tied to the likelihood of breast cancer based on the detected features. The second strategy relies on a machine-learning classification model that, using both mammography-related descriptors and general patient data, computes another metric, called Statistical Risk, likewise associated with the probability of developing breast cancer. These metrics are then integrated to generate a combined measure, the Global Risk. This value may subsequently be adjusted using a weighting element derived from the BI-RADS category assigned by the medical specialists. The resulting Corrected Global Risk can then be interpreted to determine a patient’s status and issue tailored recommendations. The proof-of-concept evaluation and software implementation were performed using a dataset of 130 patients obtained from a database at the University of Wisconsin–Madison School of Medicine and Public Health. The initial outcomes were promising, suggesting the system’s potential utility, though thorough clinical testing in real practice settings is still required. Its integration into existing clinical information systems may further streamline diagnostic procedures and enhance patient outcomes.
