Sunday, November 13, 2016

When Big = Smart: Harvesting Data to Improve Health

Personalized medicine offered a promise to fields that require a nuanced approach to care. Oncology, neurology, and endocrinology are all areas of medicine where there are seemingly no one size fits all cures. Because of this, there is a great need to better understand the role of big data in making smart decisions about medical treatments.

There are many ways to collect and harvest big data to better determine the course of care for individuals suffering from cancer, neurological, or metabolic diseases. By collecting many types of data, from demographics to DNA to treatment response, scientists have the resources to evaluate the disease state and begin to match groups of people with specific treatment approaches.

There are several ways to analyze big data. Traditional approaches such as binary logistic regression and discriminant function analysis aim to classify people into treatment groups. Essentially, you can enter several variables and these techniques create a model that decides which patient belongs in which treatment group. Sensitivity (a person gets in the right group) and specificity (the wrong person doesn’t get in the wrong group) measurements allow the physician to understand how accurate the test model is, and scientists can work to improve these models, and thus, treatment outcomes (there are no perfect models).


One of the most exciting variables that can be added to these models is data from wearable technologies. Wearable electrodes, implants, and biosensors are all capable of relaying data to prediction models in real time. This can improve treatment outcomes by notifying physicians when treatments are working; and when they’re not. By creating big data repositories, we are able to make smart decisions about the future of treatment for the individual and for the entire field.

No comments:

Post a Comment