HOW CAN A LARGE DATABASE ASSIST IN MEDICAL CARE?
For most diseases doctors prescribe therapies based on their experience and best guess. They do not have precise information about what treatments lead to what outcomes—standard treatment guidelines exist for only a limited number of diseases.
The advantage of having large sums of shared data available leads to an entirely new medical paradigm. Information can be accessed in real time and used to manage an individual’s health in a much more precise and quantitative fashion (Figure 1). If a particular person becomes ill with a particular disease, for instance, a cancer associated with a particular genetic mutation and molecular profile, then in principle, algorithms can and should exist that search all other patients with the same disease and mutation (and ideally a complete genetic and molecular characterization), how they were treated, and what the outcome was. In this fashion treatments can be driven by data, not hunches, thus resulting in a new era of “data-driven medicine.” Moreover, each individual who is treated will exhibit a specific response, which can be added to the “living database,” thereby providing information that will improve accuracy for future patients. This process requires a common format and sharing of information, topics that are discussed in more detail later.
Figure 1. Databases can be searched to find cases similar to the one of interest (light individuals). By examining how they are treated and the outcome (poor outcome outlined in black) a physician can determine the best mode of treatment.
It is important to note that this model of a “living database,” in which observational data is collected and utilized clinically, is very different from the manner in which drugs are currently vetted for clinical use (Figure 2). Presently, randomized trials are run in which one set of patients are provided a drug and the other set are provided a placebo; the results are collected and analyzed after a certain time span. These trials are expensive and slow, generally costing hundreds of millions of dollars and taking years to perform. In contrast, collecting pooled data that already exists can be much less expensive and can be performed in a matter of days or weeks. It also can be much more precise; for example, it can focus on individuals of the same ethnic or genetic background and in the same geographical locations. One needs to correct for any potential biases, but overall the concept of patients also contributing research information, which, in turn, helps the next patient, is paradigm shifting and will transform the manner in which medicine is practiced.
Figure 2. Treatments are currently evaluated in randomized trials in which treatments are given to some patients and not others. This is a slow and costly process. By searching databases for cases similar to the one of interest, outcomes of people receiving treatments can be rapidly deduced and clinical decisions made based on existing evidence.