Power tools – biomedical informatics

Medical informatics can transform healthcare and pharmaceuticals in a great number of ways. The combination of large, cloud-based databases and advanced crunching tools can enhance understandings of a patient's risk profile, helping to decide the most effective course of treatment. For example, if you can search through a database of hundreds of thousands of genes and find which combinations represent a high risk for a particular disease, you can take effective pre-emptive action.

The use of computers to store and analyse large amounts of data can also help accelerate the drug development process by easing the transfer of knowledge from research teams to clinicians (and vice versa) - such as by allowing researchers to access data collected by medics on handheld devices when a patient first checks in to hospital. This can also allow pharmaceutical companies to more accurately target participants for clinical trials and predict trial outcomes.

There are a few major obstacles to overcome, however. As ever, when large amounts of personal data are involved, stringent, standardised security measures need to be devised to prevent personal information being used for unethical or illegal purposes. There is also the need for a sturdy regulatory framework that can ensure compliance with various interested government agencies. In a broader sense, some believe that the technology available has developed at a faster rate than clinicians have been able to keep up with, and that medics would benefit from more technical knowledge.

This idea underpins the new master of science in biomedical informatics, announced in February by the Graham School - a part of the University of Chicago that specialises in liberal and professional studies. The course director is Dr Samuel Volchenboum, a paediatric oncologist, associate professor in paediatrics and director of the University of Chicago's Center for Research Informatics. He speaks to World Pharmaceutical Frontiers about the aims of the course, the thrust of his own research and the potential for open-source data to change the culture of secrecy in the medical research and pharmaceutical industries.

World Pharmaceutical Frontiers: Could you tell us a little about the course and how it came about?

Dr Samuel Volchenboum: The need for the programme stems from a growing realisation that we have many positions within healthcare where there is need for proficiency in technology as well as medical background, and that people were falling into these positions with training in one but not the other.

We identified that if we provided a background in technology, computer programming and computer science, as well as a background in hospital medicine and other aspects of healthcare, we could help train future leaders to provide the informatics support that's needed to push healthcare forward in a modern academic medical centre.

Generally speaking, what is the current state of informatics as it is applied hospitals and research centres in the US?

I think there are many tools for analysing data that are progressing very rapidly. The deficiency is in the lack of people trained to use the tools in the correct way. There are multitudes of informatics tools available with which people can load data and run analyses very quickly, at the touch of the button - it's a very powerful thing. But if you're not familiar with the mathematics behind generating those answers, you could produce results that are not reliable.

I don't think we fully appreciate the extent to which informatics could help medicine, but at the same time, we don't fully appreciate the extent to which it could be dangerous if not used right. I think the state of informatics right now is what I would expect it to be, but we need to do a better job training people to harness these tools.

Is it difficult to get medical staff to take an interest in the technological side of informatics? Do they perhaps view it as extraneous to their main responsibilities?

I was recently talking to the dean of the medical school about informatics training for medical students, and observed that many people who end up in informatics positions get there through a very serendipitous process. They have a medical background and find an opportunity to lead an informatics group, and they basically learn on the job.

I think there's an important opportunity there to train physicians early on, when they are medical students and residents, so they have a better outlook on the potential of going into these positions deliberately rather than falling into them.

There's been a lot of talk about bringing down the walls between clinical trials and research, which would allow data to be shared, and potentially accelerate drug discovery. Are we seeing much progress on this front?

Clinical trials are a huge interest of mine. I've spent a lot of effort on tools to enable them; and having an understanding of the medical side (having written and performed clinical trials) as well as the technological side (having implemented systems to enable them) means I'm in a fairly unique position when it comes to trying to enable clinical trials. Drawing out the kinds of people who have experience in both areas is really going to help move clinical research forward, in hospitals and in the pharmaceutical setting, too.

Hospital systems are there to take care of patients and deal with the transactions around taking care of patients: they are not really designed to enrol patients on trials and collect data. Those kinds of activities require a dedicated informatics group that can create compliant systems for storing data, and is able to negotiate the data-usage agreements and contracts involved in sharing data with investigators and outside interests. So I do think it requires specialised training and services, whether those are handled in-house, or by consultant or professional services companies. Each group handles it differently.

What aspects of informatics are of particular interest to your research at the moment?

I'm interested in looking at the ability to take large samples of medical records and use them to study risks to patients. I think there's a lot of power when you have that much data, and if you can analyse it in the right way and standardise it, you can make a lot of interesting predictions about risks to patients, and try to improve processes from there.

That must be high on the agenda for pharmaceutical companies and insurers, too?

Yes, it's very important. But at the same time, most people don't have the kind of clean data that is needed for this type of research. We've been very fortunate at the University of Chicago; I have a very good data warehouse team, completely dedicated to producing good data, cleaning the data and standardising it so we can use it for research. Bad data just leads to bad results.

There's a lot of data wrangling involved in getting good results: it's all about making sure that when the weight in the field is 10kg it's actually marked as 10kg, not 10lb, for instance. There's a lot of work to be done in verifying the fields and understanding where weaknesses lie in the system.

Do you see any opportunities for the use of open-source data in research and drug development? Or is the culture of privacy too entrenched?

I'm very passionate about our ability to share data, and I think things will get better. But I think many of these changes will come from the consumers themselves. I think people want the data to be liberated and democratised, and used for research. A lot of the laws and regulations are there for a good reason, but once the right protections are put in place, there's no reason why we can't share the data and make it available to a wider group for research.

There has to be accountability from those who use data - assurances that they are not going to try to reidentify people. There is no such thing as non-reidentifiable data: any data can be reidentified if there is the proper nefarious intent. So I think there has to be the expectation that people aren't going to do that.

I think there also has to be protection against discrimination, depending on what kind of medical information may be available about people. This protection is starting to be put in place more strongly, but it's going to take some time.

Do you feel the industry is working together to create the right conditions for open-source to thrive?

More or less, but obviously when funding is difficult to get, it's not in every group's best interests to just share their data without having some sort of embargo around it first, so that they can publish the results and try to get further funding. So the financial climate itself creates this desire to keep data private. Until the funding gets easier, it's going to remain like that.