About this Research Topic
Biomarker readings provide short term, and sometimes long term, information on the state of health of a patient. Providing actionable meaningful decisions on such data on an on-going basis, while the patient is in either critical care or in hospital, remains a challenge yet to be overcome. Predictions on the evolving state of health, both in the short term (hours or days) to longer term (typically weeks) require a wide range of statistical and mathematical techniques.
Both frequentist and Bayesian inferential approaches prove helpful, but challanges remain in handling the widely dispersed readings in certain cytokines and the fact that they cannot generally be rescaled to look Gaussian by the usual choices of intuitive transformations. Successful modeling of such biomarker data, using both statistical and dynamical systems tools, remains for now a still elusive goal.
The over-arching aim of this research volume is to advance new techniques, be they theoretical, computational or experimental, that further the understanding, analysis and modeling of biomarker data. They should become efficient new predictive tools for various outcomes that take such biomarker data as input. The novelty should be in either specializing known theories to the biomarker predictive setup, or developing new tools that are particularly effective in handling the idiosyncrasies of the biomarker data for prediction. These developments may vary with the field of application. Our primary interest is to seek and develop such new methodologies for trauma, traumatic brain injury and oncology, but impact on other areas of pharmacology or medicine are welcome as well.
Generally one measures the state of health of a patient as a function of the biomarker and genetic data. Whereas such measures may form a a random vector, in most cases we model a univariate response. The response variable is typically a multinomial, often simply binomial, although is certain settings it can also be continuous. Biomarker data is a time series in each of the biomarkes for every patient.
We welcome new ideas in clustering techniques that may extend the use of the Mahalanobis distance in a lognormal case, spectral clustering or more novel manifold clustering methods.
Classical multinomial logistic, lasso models and machine learning approaches would also be useful to study within the special context of genetic and biomarker data.
Bayesian technology, parametric or nonparametric, can be brought in as well, as it already has in biomarker studies in oncology.
This Research Topic welcomes contributions to imuno- and gene therapy for cancer through the identification of biomarkers derived from genomics and proteomics to predict response to drug treatment.
In particular, gains in understanding genetic interactions in cancer and its effects across tumor types are of special interest. It is also of interest to consider mechanistic models, using systems of ordinary differential equations, that may highlight an interplay of the biomarker cascades over time and the impact this has on the response variables. Most importantly, novel ideas in handling biomarker and genomic data are particularly welcome, especially when successfully applied to specific medical settings such as oncology or trauma.
Keywords: cytokines, chemokines, genetic markers, inflamation, data, modeling
Important Note: All contributions to this Research Topic must be within the scope of the section and journal to which they are submitted, as defined in their mission statements. Frontiers reserves the right to guide an out-of-scope manuscript to a more suitable section or journal at any stage of peer review.