Doctors and researchers are increasingly going through big data to find answers to medical problems. This approach ensures that a one size fits all approach isn't used for everyone's treatment plan. It then becomes critical that doctors use all the tools at their disposal to find the best treatment for their patients. Increasingly, the best tool for personalized medicine is extracting insights from big data.
An example of big data's use in radiology
Doctors are starting to see big data play a huge role in medicine. Imagine a patient who presented his primary care doctor with a small nodule on his chest. Although his doctor is sitting right across from him, it's possible that the doctor needs more information to properly diagnose whether his patient has cancer. Right now, doctors have guidelines like the Fleischner Criteria that says the patient should wait a certain number of months before coming back to get imaged again. But the patient wants to immediately know what is the likelihood that it's cancer and what he should do about it. Looking at a patient's race and family history, doctors can diagnose which patient has a greater chance of having cancer.
The challenge for health practitioners is being able to find the data and using it to make a recommendation. To do that, a doctor would need to look through tens of thousands of different medical records. That's where big data technologies come into play. By focusing only on patients that have a similar nodule size, the doctor can provide the patient some very specific recommendations. These recommendations include how they ought to be diagnosed as well as the various treatment options available.
Radiology does not easily fit in the new big data world
Big data speaks to the trend of more doctors trying to personalize their treatments. An industry challenge for radiology and big data is the fact that imaging consists predominantly of pictures. These medical images are part of the patient's electronic medical record, but they are mostly unparseable by software programs written to look through lab values. One of the clinical challenges facing radiology and imaging informatics is to figure out ways that these images can be turned into 'data.' One solution is to tag the images, while another would be to use computer vision to categorize the images algorithmically.
These images will have to be accessed through radiology information systems that have enough associated data to make them discoverable. Once this image data is computer parseable, doctors will be able to quickly find pertinent information using big data. A doctor would be able to use big data to extract insights from other patients with an abnormality similar to the one that the doctor is currently treating. By doing that, doctors would be able to personalize their treatments.
Future challenges for big data and radiology
Healthcare computer technology is already able to use many of the concepts employed with big data. Many hospitals already employ Hadoop to glean actionable insights from patient records. For radiology data to stay relevant in the age of big data, practitioners must make sure that this imaging data is discoverable. Doctors are going to need clean data to properly inform data-driven decisions. This might include matching the right study for the right patient. And as medical records make the transition from paper to digital, this will present new challenges for hospital IT staff. Diagnostic imaging data is so complex that it makes it harder to extract meaning from the images themselves. Putting these images into a database (and making them machine-searchable) is going to be the future of radiology and imaging technology.