Deciphering Chaos: A new "Fuzzy" Artificial Intelligence to predict the battle between the immune system and cancer

Understanding how a tumor evolves against the attack of the immune system is one of the greatest challenges in modern medicine. Current mathematical models are usually deterministic; that is, they assume fixed values that rarely reflect the reality of patients, where the immune response varies enormously from one person to another. To close this gap, a team of researchers from the Escuela Superior Politécnica del Litoral (ESPOL) has developed a new computational modeling framework that uses Type-3 Fuzzy Logic and neural networks, capable of simulating tumor-immune dynamics under conditions of uncertainty and chaos.

Beyond exact data: Embracing biological uncertainty

The study addresses a critical phenomenon: the "delay" or latency in the activation of cytotoxic T cells (the body's defenses). Small variations in this response time can make the difference between tumor elimination or an aggressive relapse. The proposed new model not only computes a single trajectory but generates "bands of uncertainty" that visualize multiple possible scenarios for the same treatment.

Unlike the "black boxes" of traditional AI, this approach is interpretable. It uses a logic-oriented architecture that preserves chaotic structures and bifurcations (tipping points in a patient's health), allowing physicians to understand the "why" behind a prediction. The model proved superior to conventional techniques (such as Type-2 or ANFIS) by maintaining the fidelity of the cancer's oscillatory behaviors even with incomplete or noisy data.

Risk maps for personalized medicine

One of the most promising results of the research is the creation of visual clinical risk maps. Using comprehensible linguistic rules (for example: "If the delay is high and CD8+ cells are low, the relapse risk is high"), the model classifies patients into safe or danger zones.

This tool allows treatment stratification: identifying which patients need immediate intervention to reduce immune activation time and which will respond well to standard therapy. By aligning mathematical precision with Explainable Artificial Intelligence (XAI), this work opens the door to more robust biomedical simulators that support decision-making in precision oncology, transforming abstract numbers into clear and actionable diagnoses.