For decades, the basis of nearly every medical discovery and research has been about the collection and analysis of data: who gets sick, why do they get sick and why. However, big data has enabled the potential for enormous breakthroughs in this sector.
A recent example of this is, is a study conducted by computer scientists of the Free University in Amsterdam in collaboration with the University Medical Centre Utrecht, which found a new predictor for intestine cancer. The computer scientists looked for distinctive patterns in 263.000 (anonymised) electronic patient records, of which 1292 received the diagnosis of intestine cancer. They compared both groups (cancer versus no cancer) by allowing software to search for the differences in the run-up to the diagnosis. The research has reconfirmed previously known predictors (iron deficiency, age, and constipation) and found one new one: metabolic syndrome, which is a metabolic disorder.
The technique used in this research will, in the near future allow a risk calculation at the desk of the doctor’s office, based on routine patient information of an individual with intestinal complaints. This can help in the decision whether to have the patient examined further or not.
The strength of this new predictive model is that the software uses the electronic patient records data without any bias. It is not programmed in a way to search for specific predictors, as is usually the case in medical research based on data analysis. This causes the model to be applicable on all kinds of data, and therefore it has enormous potential to enhance disease prediction in other diseases, making early detection and intervention possible.
Big data has been known to be beneficial in many industries, often allowing companies to achieve substantial competitive advantage. However, its potential and importance in the healthcare industry exceeds those of other industries since it can actually save lives.
References:
Kop, R., Hoogendoorn, M., Teije, A. T., Büchner, F. L., Slottje, P., Moons, L. M., & Numans, M. E. (2016). Predictive modeling of colorectal cancer using a dedicated pre-processing pipeline on routine electronic medical records. Computers in Biology and Medicine, 76, 30-38. doi:10.1016/j.compbiomed.2016.06.019