Research

Overview

Treating successfully any disease has as its cornerstone an early and accurate diagnosis. The mission of my research is the development of new statistical methods towards diagnostic testing and early detection of (potentially any, but in particular) fatal diseases.

My primary research focus lies on the development of methods related to (bio)marker discovery, (bio)marker evaluation, modeling, and comparisons. I am mainly interested in the mathematical aspects and different metrics that are involved in the receiver operating characteristic (ROC) space. Different metrics within the ROC space might be appropriate for different settings, depending on the severity or stage of the disease, as well as other factors.

My statistical methodology has been driven by real-world problems, related to a variety of clinical questions. These refer to early detecting/diagnosis of pancreatic cancer, lung cancer, prostate cancer, colorectal cancer, liver cancer, late-onset sepsis in neonates, Alzheimer's disease, Parkinson's disease, dementia, as well as other very different frameworks (e.g. entomology). During these projects I have been collaborating closely as the lead statistician with the departments of Clinical Cancer Prevention and Translational Molecular Pathology (The University of Texas MD Anderson Cancer Center (MDACC)), Departments of Radiation Oncology and Surgical Oncology (MDACC), International Agency for Research of Cancer-World Health Organization (WHO, Lyon, France), Institute of Evolution (University of Haifa, Haifa, Israel).

Two highlights of my clinical collaborations related to lung and pancreatic cancer are the following:

- A four-protein biomarker blood test that improves lung cancer risk assessment. This is the first blood-based biomarker study which relied on pre-diagnostic samples from one cohort (the US CARET cohort) to develop a marker panel that was subsequently subjected to a blinded validation using independent pre-diagnostic samples from ten European countries. Our results initiated consultations with the FDA. (Guida et al. 2018, JAMA oncology)

-In 2017, we presented a protein panel that outperforms CA19.9 for detecting early stage of pancreatic cancer which was also independently validated on a European cohort (Capello et al. 2017, JNCI). In 2018, we present a metabolite panel that significantly improves further our aforementioned protein panel (Fahrmann et al. (2018, JNCI). This new combination/equation is also validated on independent European samples by the International Agency for Research of Cancer (WHO).