Biostatistics, Epidemiology, and Research Design (BERD) Trailblazer Award

By Frontiers , Clinical and Translational Science Institute

Oct 08, 2024

Study Title: Combinations and cutoff estimation of non-monotone markers

While many clinicians and researchers are highly specialized in one or two areas of human health, Leonidas Bantis takes a different approach. Bantis is a biostatistician who has focused his career on identification, validation, and reproducibility of biomarkers. According to him, a biomarker is “anything that can be objectively measured in the human body that may help identify individuals who are in a disease group or disease state.” Think, for instance, of a fever, where a temperature that is too high (e.g., over 100.4°) means a patient is fighting an infection and may need treatment by a physician. Other biomarkers where the level is too low or too high can also point to disease or risk for disease. For example, specific proteins might indicate unhealthy states related to cancer at both extreme lows and extreme highs; the healthy state is somewhere in the middle. Importantly, once identified, biomarkers can be crucial in identifying diseases or disorders, determining risk for development of disease, and selecting subsequent treatment plans.

Biostatisticians, such as Bantis, have developed algorithms and software to help establish the strength of certain biomarkers in identifying disease states and disease risk. However, as Bantis says “investigators and statisticians have traditionally neglected situations where a biomarker might exhibit very high or very low levels for the disease group,” which means we may be missing valuable information by focusing only on biomarkers that have a single binary cut-off (e.g., normal vs. diseased). In the past, biomarkers with multiple cut-off points, have not been included in statistical modeling. Bantis and his graduate research assistants plan to develop methods to help research teams figure out cut-off points for disease classification using biomarkers that may have two cut-off points as well as establishing what are the optimal cutoff points for optimal decision making. This new direction could be highly informative for physicians and researchers who are developing interventions and treatments.

The BERD Trailblazer award will go entirely to supporting Bantis’ graduate students as they work with him on this project. Bantis hopes that his team will be able to develop the biostatistical method, evaluate and prove that the method works, and then finally apply it to real-world datasets to figure out what biomarkers are strong indicators of disease or risk for disease. This method would be able to be applied to any type of biomarker(s) for any disease and would be freely available to other researchers. This method of identifying important biomarkers could be applied to any number of health problems, from obesity to esophageal cancer, intellectual disability to asthma, to make an impact on human health.