For decades, it has been standard practice to treat patients based on broadly-defined groups that they fit into based on sex, racial group or other outward-facing factors. Unfortunately, this has led to drugs' often being prescribed to patients based on a relatively small amount of information known about them, with little emphasis on how the patient’s own genetic or phenotypic status may affect the outcomes.

This is the case for diabetes, a metabolic disorder that affects millions of people in the U.S. and around the world. For type I diabetes, the most common prescribed medication is insulin. Patients who take insulin often give themselves injections multiple times per day, and monitor their blood sugar levels. Any relief of burden on diabetes patients from adverse events or from the prescribed medications themselves would be significant.

How Might Translator Help?

Translator offers the opportunity to recommend a drug of choice for a given patient by answering questions about drug-genotype or drug-phenotype interactions quickly and efficiently. In this case, the search also yielded an unexpected interaction that warrants further investigation. When this unexpected result was investigated in more depth, it was noted that a Google search for the terms produced too much noise to easily allow for the connection to be made. A PubMed search yielded 17 hits but only the 17th was relevant and that paper was from 1968 and written in Spanish, making it essentially inaccessible to a large number of researchers. Our researcher hypothesized that either (1) physicians are implicitly aware of different subcategories of diabetes and that these subcategories are reflected in the medications they prescribe, or (2) that different diabetes medications cause different phenotypes.

Can Translator use clinical and research data sets to help clinical researchers find patterns in adverse events and explain the mechanism behind them?

Translator Insight

Translator returned a connection between insulin and atelectasis, with atelectasis not showing up in the adverse events lists of eight other common classes of diabetes medications. Insulin is an extremely common treatment for diabetes patients, with most type 1 diabetes patients requiring insulin, along with approximately 20% of type 2 diabetes patients.

Atelectasis is a partial or sometimes total collapse of the lung occurring when the alveoli become deflated or filled with fluid, which can prolong hospitalizations and be potentially life-threatening. A connection between insulin and atelectasis could have broad implications in the treatment of diabetes. Translator identified and explained this surprising connection through the synthesis of multiple, disparate data sources. Furthermore, because we are using an open health database that was created as a part of Translator, we were able to make connections between real patients’ medication history and their later diagnoses with seemingly unrelated phenotypes. This made it possible to find commonalities between them not necessarily codified in other data sources.