Accenture debuts data-driven methodology for pediatric AML
Accenture’s Applied Intelligence arm has introduced a data mining, aggregation and assessment methodology designed to help physicians and researchers more precisely target treatment regimens – and improve outcomes – for pediatric acute myeloid leukemia (AML) patients.
Working in partnership with the Target Pediatric AML research project and the Fred Hutchinson Cancer Research Center, Accenture data scientists developed data and analytics tools that integrate genetic and clinical trial data from more than 2,000 children with AML into a user-friendly format. Clinicians can then use the information to solve challenging care problems.
These include identifying patients at high risk for conventional treatment failures so that regimens can be altered to avoid setbacks; predicting the likelihood of treatment success and the emergence of side effects at the diagnostic stage; developing customized treatments for individuals and patient groups with similar disease profiles; and making voluminous amounts of data more readily available to researchers seeking insights into what treatments may work best for any given patient.
The development is encouraging news for pediatric oncologists, given the challenges that come with treating AML.
“The treatment is intense and too often a one-size-fits-all approach can result in negative side effects and poor outcomes,” said Joe Depa, managing director of Accenture Applied Intelligence. “The hope is that ability to synchronize and make sense of not only the amount, but also the variability of each patient’s clinical data, will give researchers new clues to progress towards better and more individualized cancer treatments.”
To develop the methodology, the Accenture team took a copious amount of information and applied both data science and engineering mechanisms and machine-learning libraries to it. In doing so, it created a code base that will allow clinicians to model, learn and forecast how effectively certain treatments might work in patients.
“By building what we call data pipelines, this tool ingests raw clinical and genomic data, and then cleans, transforms and produces a data set that is optimized for further analyses,” Depa explained. “This reduces time for data preparation, which is a consistent issue in this area.”
In tandem with this data and analytics approach, Target Pediatric project investigators compiled and standardized a range of essential data points. They included RNA (one of the most critical indicators of treatment effectiveness, and a very difficult one to analyze), patient demographics, prognosis and clinical treatment arm.
Taken collectively, the data will help researchers compare patient profiles and outcomes at appropriate scale and, hopefully, give doctors more perspective on precision medicine treatments better suited to an individual patient’s medical history and genetic characteristics. For instance, physicians could determine ahead of time how a high-risk patient would respond to a standard therapy and, perhaps, opt for bone-marrow transplantation or more targeted therapies.