A pan-European consortium is drafting a proposal to develop infrastructure to create virtual models that will help physicians tailor treatments to individual patients' genetic and physiological characteristics.
The group, called the Information Technology Future of Medicine consortium, is seeking funding from the European Commission to support a 10-year project aimed at creating "virtual patients” that will incorporate health records and genetic information.
ITFoM comprises more than 25 members including academic institutions such as the Max Planck Institute for Molecular Genetics, University College London, the European Molecular Biology Laboratory, the Wellcome Trust Sanger Institute; and vendors such as Illumina, Roche, Oxford Nanopore, and IBM.
These and other partners plan to create new techniques to acquire and evaluate patient data, as well as tools for storage and processing of real-time patient data in mathematical models, and new systems that can learn, predict, and inform.
To house and manage this data, the group also expects to develop new computing, storage, networking, and modeling technologies.
“We model complex processes from the weather to the behavior of cars to galaxies using lots of data and complex computer models. We have to do the same thing in medicine,” Hans Lehrach, director of the vertebrate genomics group at Max Plank Institute for Molecular Genetics and ITFoM coordinator, told BioInform.
The models are "a good way to generate individually optimized treatments because you can take all the information you collect on an individual and put it into the model,” Lehrach said, adding that the approach would go further than stratified medicine approaches that group patients into treatment categories based on broad similarities.
The group received a one-year grant of €1.5 million ($2.2 million) in preliminary funding from the European Commission in May. It is one of six shortlisted projects that are competing to receive €1 billion ($1.4 billion) over 10 years under the European Future and Emerging Technologies scheme, which will select two projects in the second half of 2012.
If ITFoM is funded it could result in significant savings in healthcare costs worldwide while simultaneously increasing the quality of the care patients receive from their physicians, Lehrach said.
The team is using the initial funding to flesh out a proposal for the 10-year effort.
"If at the end of these ten years, what we would like to have is a working virtual patient .... what are the processes which need to be put in place to make that happen ... or to be well on the way to that happening?," Daniel Jameson, a project manager in the Manchester Center for Integrative Systems Biology and an ITFoM participant, told BioInform.
Currently, the group is considering the infrastructure it will need and evaluating available resources. For example, a virtual liver cell being developed by Germany's HepatoSys project could be incorporated into the system, Jameson said.
He also said the team will be considering the most efficient methods of creating computer models and applying them to clinical scenarios.
For its part, U of Manchester will work on integrating medical and analytical data into models of cells, organs, and eventually whole virtual patients, Jameson said.
In past research projects, the Manchester team has worked on developing low-level mathematical models of yeast metabolism as well as larger-scale reconstructions of yeast metabolism, he said, adding that he and his colleagues are currently working on a collaborative reconstruction of human metabolism.
Meanwhile the Wellcome Trust Sanger Institute expects data from several large-scale genome projects to “naturally feed” into the ITFoM project, Tim Hubbard, who heads up Sanger's vertebrate genome analysis group, told BioInform.
In particular, he cited the UK 10K project, a three-year initiative that aims to sequence 10,000 individuals in order to discover rare genetic variants that are important in disease, and the Deciphering Developmental Disorders project, which aims at analyzing the DNA of around 12,000 children with undiagnosed developmental diseases.
However, he noted that in these early days, the ITFoM consortium will focus on “putting the right framework in place,” including tools to integrate data and infrastructure from different groups as well as actual patient data.
“I think that there will be publicly available databases … [containing] collective knowledge about things like variation … and then you can have different services built on top of that which actually integrate with patient data,” he said. “In terms of actually having simulations, I think you can have a virtual machine model … populated by the individual's data, which can then run on cloud machines.”
An IT First
Hans Westerhoff, director of the Manchester Center for Integrative Systems Biology, who is heading the University of Manchester's part of the project, noted in a statement that this project would mark the first time that an IT system looking at individual care will be combined with genomics and medical needs.
Under the system, doctors would have instant and in-depth knowledge of an individual patient’s health needs and medical history at their fingertips. In addition, physicians would have access to patients' genomic, metabolomic, and proteomic data among other types of information, including the computational models that would tie this data together.
“The models will be there to help diagnose a particular problem and provide solutions," he said. "Obviously this would need to be done in conjunction with a person’s [general practitioner] depending on the gravity of the situation."
Potential uses for the ITFoM infrastructure would be to make predictions about certain health-related scenarios, such as if a patient changes medications or makes a lifestyle change such as beginning to run three times a week.
Through genome sequencing and clinical information gathered from the patient, the general model could be adapted to suit the particular health demands of any individual, including such issues as allergies, congenital defects, and current treatment.
Lehrach cautioned, however, that some diseases will be easier to incorporate in the model than others, simply because scientists currently don’t have a good understanding of the molecular mechanisms behind certain complex conditions, such as schizophrenia.
The ITFoM proposal represents a "paradigm shift" in bioinformatics and genomics towards translational bioinformatics, said Alvis Brazma, senior team leader of the functional genomics team at the European Bioinformatics Institute.
Speaking with BioInform, he noted that the emphasis for bioinformatics is moving away from building reference datasets, which was the goal 10 years ago, to working with individual genomes.
"The impact of that on patient treatment has already started," he said. "The importance of knowing the genotype or other molecular profiles [is] already important in healthcare and in ten years it will become the norm."
In terms of ITFoM's vision and mission, "we need to achieve a situation [where] on one hand a molecular profile … is properly recorded and linked to the electronic health record for the patient from which these samples come and made available to the entire research community in some secure and properly ethical way," he said. "On the other hand, we also want to make sure that the results from this research are made available to the doctors and have an impact on healthcare in a very prompt way."
Although hardware and software technologies have grown significantly since the early days of genomics, the necessary infrastructure for this vision to become reality isn't in place.
For starters, the group will have to come up with novel storage and compression methodologies to handle the flood of data that will be generated as genomic tools are incorporated into standard medical practice, Brazma said.
Additionally, issues related to data security need to be addressed. A third requirement, according to Brazma, would be technology for "controlled federation" or a hub-and-spoke architecture, where, for instance, EBI could serve as a hub providing some archiving capabilities that are linked in a federated fashion to patient electronic health records.
Manchester's Jameson noted that the hardware requirements for a project of this size are still unclear, pointing out that a "huge amount of computation" would be required to run a single patient's model in real time in addition to storage requirements and data security.
"All we can do at the moment is to say these are the things that we need to take place in order for this to start to happen [and] these are the sort of services we need," he said.
"I think we are probably going to have to think about new ways of actually representing the models and the data in order to fit it into what we've got at the moment. It will drive technology in both hardware and probably in the way we think about programming these [models]."
Since it is still early days, it also isn't clear whether systems would be installed locally in hospitals or if there would be a central system that physicians and patients can link to. Brazma expects it would be a "bit of both."
"I think the electronic health records will always stay in the hospitals but perhaps we need some kind of index like Google index which would allow [users] to find which record and which data are in which hospitals," he said. "Obviously there is a balance needed between central archiving for all kinds of cancer datasets, for instance, and specific patient records in the hospitals and it all has to work seamlessly."
Have topics you'd like to see covered in BioInform? Contact the editor at uthomas [at] genomeweb [.] com.