Managing supply chains; analyzing economic data; controlling the spread of disease; using ChatGPT to write letters; using your face to unlock your phone. One thing we've all learned in recent years is a healthy appreciation for mathematical modeling, whether we call it AI, predictive analytics, computer algorithms or one of its other many pseudonyms.
In medicine, of course, mathematical modeling has long been integrated into the field: in academic research to assess statistical significance, in basic science to understand disease pathophysiology and in clinics to predict risk.
A recent publication from Zhihui Wang, PhD, a computational biologist at Houston Methodist, and his colleagues suggests that these applications are only the tip of the iceberg. They predict that personalized mathematical models identifying the best treatment strategies for individual patients will soon be an indispensable part of cancer care.
Advances in modeling techniques and computational power, biological understanding and biomarker libraries have enabled mathematical models to grow in complexity.
No longer content to sit on the sidelines as a helpful analytical tool, mathematical models incorporating patients' unique biological and physical tumor-immune interaction mechanisms are being integrated into clinical workflows to aid in decision-making.
"Many modeling teams are now transitioning their work away from descriptions of tumor–immune systems for basic science purposes, and instead are moving towards immunotherapy models for clinical applications," explained Dr. Wang. "We expect that engineered, computation-based immunotherapy treatment strategies will become a critical part of the next-generation therapies by facilitating clinical translation of new drugs and optimizing personalized treatment strategies for maximized therapeutic success."
Dr. Wang's review article, published in the journal Nature Computational Science, focuses specifically on cancer immunotherapy, a field at the forefront of the movement due to the complexity of treatment pathways and the high stakes nature of the disease.
Typical cancer immunotherapy methodologies include immune checkpoint inhibitor therapy, adoptive cell transfer therapy, vaccination and exogenous cytokine therapy. Immunotherapy can also be used in combination with traditional cancer therapeutics such as chemotherapy and radiation therapy. Combination therapies often lead to better treatment outcomes since multiple key pathways are targeted synergistically.
There are currently more than 600 cancer drugs, including at least 30 immunotherapy agents that are approved by the Food and Drug Administration (FDA). That makes for a high number of possible drug combinations with unknown treatment outcomes.
"It is not possible to systematically assess each combination via clinical trials alone," Dr. Wang stated. "Mathematical modeling combined with artificial intelligence may prove to be indispensable to efficiently and effectively identify optimal drug combinations as well as predicting treatment outcomes."
In addition to predicting the efficacy of personalized treatment combinations, in silico — or computer modeled — dosing studies can predict the maximal patient response at the lowest therapeutic dose/frequency and can even estimate the toxicity of new drugs before testing on human beings has occurred.
"Such modeling platforms not only provide valuable foundations to study the mechanisms underlying the delicate balance between treatment success and failure but also provide quantitative tools to transition the treatment optimization process from approaches based on trial and error to engineered design," stated Dr. Wang and his colleagues in the published report.
For more information about the study or the predicted applications of mathematical modeling in cancer immunotherapy, click here to read the full article in our sister publication Methodology.