Big data and predictive analytics
are having a moment in healthcare. In fact, a 2013 report by the Institute for Health Technology Transformation (iHT2) noted that by 2011, U.S. healthcare organizations had generated 150 exabytes-that's 150 billion gigabytes-of data. Having aided in efforts for care coordination and population health management, these technology enablers are now paving the way for new breakthroughs in personalized medicine. Though still an evolving field, advances in genomic research over the last decade have contributed to this form of medicine becoming a real possibility.
Personalized medicine is about tailoring medical treatment to the individual characteristics of each patient, classifying individuals into groups that differ in their susceptibility to a particular disease or their response to a specific treatment. It relies on biomarkers that present signs of normal or abnormal processes. With the amount of data being mined and analyzed, it will be easier to identify genetic correlations, identify patterns in patient and population data, identify patient specific patterns and predict physiological conditions and provide better patient self-management for enhanced clinical outcomes.
While there is no shortage of data, we are still waiting on the supportive technologies to catch up and put that information to use. Before personalized medicine can become a reality, the industry will need to focus on the following four areas:
1. Relevant Big Data Technologies
For it to be successful, healthcare organizations need to manage and analyze large, complex data sets that would include both structured and unstructured data types. Traditional warehouse structures and analytics tools are not equipped to store and manage this size and complexity of this information, which is where technologies like HDFS (Hadoop Distributed File System) and Columnar Databases come into play. A framework like Map Reduce allows powerful processing and analytics by running complex algorithms across large sets containing structured and unstructured data.
Organizations also need to adopt analytics and visualization capabilities that can combine clinical and genomic information with unstructured, non-clinical, or even analog data to provide rich, actionable insights to healthcare decision makers. Such an example would be mapping lifestyle trends and genetic patterns within a specified age group, and their correlation to incidence of diabetes in a given population.
2. Data Aggregation and Integration
The healthcare data needed for personalized medicine spans across multiple types of sources within and outside the healthcare organization, such as genomic, radiology, biometric and patient data from lab systems, HIEs, RIS/PACs and patient portals. The challenge here is to effectively aggregate and normalize the data from a wide range of applications, with huge variations in structures and formats. We are seeing the emergence of ETL technologies that can extract this information from multiple sources, process it and populate big data storage in a performance efficient way.
3. Data Quality Management
There is no question that vast amounts of data exist. The issue is with the quality of healthcare data itself. One of the most difficult and widespread challenges for healthcare organizations is achieving a consistent level of quality across a wide variety of source systems, making the right information accessible to the right decision makers, at the right time. Key aspects of data quality include standardizing data and terminologies, semantic mapping and effective patient matching algorithms.
4. Data Privacy & Security
Data privacy and security need to be closely monitored as a large amount of diverse patient information would be accessible to multiple stakeholders across the care continuum. Systems that store and transact protected health information (PHI) need to be compliant with data security guidelines under HIPAA and the HITECH Act. Organizations need to address security challenges across multiple dimensions, including de-identification of patient data, rule-based processes for tight access control, security measures implemented for data in transit and risk identification processes and mitigation plans.
The Road Ahead
The road ahead for personalized medicine will become clearer as the industry moves towards creating rich repositories for genomic information as well as clinical data, partly through research and development efforts as well as mandates driven by federal and commercial incentive programs that focus on quality and cost of healthcare.
Considering the enormous amounts of patient data that would be made available across distributed computing environments, there would be some serious concerns about security and privacy. As organizations require huge amounts of storage and processing power to manage and process big data, they need to leverage cloud technologies for scalable and cost-effective infrastructure.
While a significant amount of work has been done by the healthcare industry as well as the technology vendor community, there is still work to be done before we can achieve a truly seamless, standards-based and interoperable ecosystem. It will be a while before we have the massive confluence of interoperability and other tools that are needed to make personalized medicine work. In spite of this, big data and advanced analytics
technologies available today are the best catalysts for personalized medicine to become a reality.