The earlier doctors find diseases, the better (typically) one's prognosis. Looking for biomarkers -- the biochemical signatures of certain diseases that are found in tissue, blood, and urine -- is one way to diagnose diseases earlier, even before a patient is symptomatic.
The good news is that scientists are identifying new biomarkers in labs on a regular basis; the bad news is that it can take years and even decades to study them in clinical trials, and it can be expensive to conduct those trials. So scientists are working more and more closely with engineers to build computational models that evaluate those biomarkers far more efficiently.
Now, a multidisciplinary team of urologists, pathologists, and engineers at the University of Michigan is closing in on a faster and more affordable computational model they say will improve prostate cancer diagnosis. Using large data sets with information on biomarkers, biopsy test results, and treatment history, they are looking for the most promising biomarkers (as well as biomarker combinations, frequency of patient testing, etc.) to get a prostate cancer diagnosis as early as possible.
"Practical applications could include a number of possibilities," says Brian Denton, an engineer at the University of Michigan who specializes in "optimization under uncertainty" as it applies to the detection, treatment, and prevention of chronic diseases. "One would be for companies trying to develop biomarkers to evaluate and understand the benefits relative to the costs at the population level, so from a population health perspective. And on the other end of the spectrum, for existing biomarkers being utilized, this can help doctors make decisions about whether to do another biomarker test on a patient. So you might get a test every year, but there's an open question about whether it should be every two or three years instead."
Denton adds that the software and analysis his team is developing have the potential to translate directly to other cancer types and even other chronic diseases -- the data sets would be what change. He hopes that within the next year or two his team will have made substantial progress optimizing the computational model so that companies and docs alike can begin reaping the benefits.