Target knows you are pregnant before your doctor does based on algorithms about consumer shopping data. Healthcare lags behind retail services and the financial industry in its ability to collect analytics on consumers. This is in part due to concerns about patient confidentiality and in part due to funding. However, big data can save American taxpayers around $450 billion in annual healthcare costs if used correctly. The healthcare industry wastes over $2 Trillian dollars annually on inefficient procedures for curing diseases and illnesses. Analytics can help providers ameliorate these results by determining exactly is causing a patient ailment and the best way to treat the patient based on their individual characteristics. Hospitals and healthcare providers have a lot of options on which types of analytics and products to use and are often confused. Which databases and products a hospital should use is based size and type of the analysis they need to perform. Kaiser Permanente has around 10 PB of data, but your healthcare organization may have much less. It also depends on whether you are merely looking to qualify for meaningful use or if you need to setup custom analytics based on your medical specialty.
Types of Healthcare Analytics
Retrospective/Descriptive Healthcare Analytics
Most common and simplest form of Healthcare Analytics, describes what happened to patients in the past
o Can include basic information such as Allergies, Medications, Patient Visits, Family History, Patient History, and Problems
o Used for Meaningful Use quality based reimbursement incentive and pay-for-performance (P4P) programs
- PCMH-Patient Centered Medical Homes
- ACOs- Accountable Care Organizations
- PQRS-Physician Quality Reporting System
o Used in population health management to find trends in health and providers among specific groups
Diagnostic/In-Line Definitive Healthcare Analytics
Focuses on why outcomes happened
o Can help providers determine whether lifestyle, genetics, medications, diet or lack of exercise are causing the particular illness the patient is presented with.
o Used in population health management to determine what is causing patient illnesses
Integrates information from Retrospective and Diagnostic Analytics to predict future outcomes for patients
o Allows providers to better discuss and analyze treatment decisions with the patient
o Payers can better determine which treatments will provide the most value
o Can discourage pay for service model for ineffective/inefficient treatments
o Allow providers to prescribe medications based on individual characteristics instead of population averages
Takes predictive analytics one step further by using processes such as Clinical Decision Support Systems to determine what the next steps for the patient are
o Allows providers to use protocols that have proven efficacy to achieve the best patient outcomes
Currently most healthcare providers use relational databases to store and analyze information however in the coming years, big data solutions like Hadoop will need to be implemented to analyze the wealth of information coming from unstructured and structured data sources.