By Bruce Schatz and Richard Berlin

People are different, so different lifestyles have different effects. Some doughnut eaters get diabetes, some do not. Some smokers get heart disease in their thirties, some never do. Some women with BRCA gene markers get breast cancer, but some never do. The fundamental approach to viable healthcare is to give each person only what works for them and nothing else. This would maximize the quality, but minimize the cost.

Personalized Medicine is being realized by the technologies of genome biology, where a profile is measured for each individual then used to choose treatments targeted to the individual person. Current treatments are for whole populations without individual variation, since there is no effective method for determining population stratification. Personalized medicine pushes protein profiles for each individual. But current technology can only screen a population of a hundred at this level. Much like current clinical trials, it costs $100M to screen 100 persons. It is known that this number is far too small for effective stratification, since the population helped by a drug cannot be distinguished from the population hurt by the drug, and many patients take medications without documentation of any significant clinical benefit. This phenomenon produces medical errors and unnecessary deaths, as happened with blockbuster drugs such as Vioxx or Resulin. There is no practical way to examine the protein and gene profiles of the entire population in a periodic and verifiable manner at the present time.

Population Health deals with large numbers of persons, to measure their health and assess their risk. The measurement is necessarily done across all activities of daily living, because health risk has major effects from both genes and environment. Thus, diet and exercise or stress and sociality must be considered equally with genetics and inheritance. Current health measurement records tens of features, to assess risk for acute conditions. Future health measurement must record thousands of features, to assess risk for chronic conditions, which must be managed rather than cured. A comprehensive national health survey would record daily values of thousands of features for millions of persons. Current commercial systems for social networking, such as FaceBook or MySpace, already record features for millions of persons. Current research systems already record thousands of features, which are explicitly extracted from persistent conversations like health messages and implicitly extracted from ubiquitous sensors like home motes. Thus information systems at the requisite scale will soon be technically feasible.

Scalable Personalized Medicine will be realized through revolutionary technologies that can effectively screen an entire population of 100M persons. Information technology for health measurement can cover social and environmental effects, while integration with electronic medical records covers demographics and diagnosis. Such a system is scalable since each screen costs one dollar instead of one million dollars, yet personalized due to the number of features considered. A scalable technology for population measurement performs an inexpensive screen with computers for whole populations, using this to place persons into cohorts, then performs the expensive genetic and proteomic profile only for representatives from each cohort at highest risk for a particular condition. The coarse comprehensive population survey will filter out most people for any condition, making the genomic testing economically feasible. Only a few persons should be tested for cystic fibrosis, only mature women at significant risk should be tested for breast cancer to optimize BRCA screens, only susceptible persons living in detrimental environments should be tested for asthma. With many more measured features of personal health, there will be many finer categories of disease risks appropriate for finer testing, well beyond these simple examples.

We need to implement scalable technologies for personalized health across whole populations now. This is a rare instance of biomedical research into basic science that would be immediately practical as clinical application. There is common ground between genome medicine and healthcare infrastructure. The integration of social and environmental cohorts with genetic and proteomic profiles must be implemented at a large scale, while leveraging the national efforts towards electronic medical records. The American Recovery and Reinvestment Act just signed into law allocates stimulus monies of $1B specifically for comparing the effectiveness of different treatments for the same illness and $10B specifically for new biomedical research relevant to practical healthcare. It also includes $20B specifically for national development of electronic medical records. To save the economy by creating viable healthcare, we urge NIH and other appropriate government agencies to use these funds to put major funding into practical research on Scalable Personalized Medicine, towards individual variation for whole populations.

Bruce Schatz is the Director of the Bioinformatics Laboratory in the Institute for Genomic Biology and the Head of the Department of Medical Information Science in the College of Medicine at the University of Illinois at Urbana-Champaign.

Richard Berlin is a general surgeon at a local hospital in south central Illinois and former Medical Director of the regional HMO. They are writing a book on healthcare infrastructure for the 21st century.