WORCESTER, Mass.--(BUSINESS WIRE)--Health care providers and payers face a critical need: gain control over rising costs and maintain a high quality of care. At the same time, the complexity of health care data is exploding, with adoption of Electronic Medical Record (EMR) systems. New regulatory mandates also require more reporting on clinical performance – tied to payments and penalties.
Driven by the tsunami of data, new ‘health care intelligence’ technology holds tremendous promise — to improve the efficiency, quality, and safety of care delivery. With the right tools, your data can save your company millions of dollars, help your patients avoid unnecessary care, and focus resources to improve outcomes.
The Current Approach is Broken
Current ‘predictive modeling’ tools analyze pharmacy and medical claims data to create a ‘risk’ score for each member or group. Accuracy is limited, however: R2 (r-squared) scores, which measure predictive accuracy, typically range from 0.2 to 0.3 (20% to 30%) for a standard cost prediction. They depend on claims data that are often 30+ days old, and cannot handle the wealth of clinical detail from new Electronic Medical Records (EMR), Hospital Information Systems (HIS) and Health Information Exchanges (HIE), which can be delivered and analyzed immediately.
Why is this important? Low accuracy means that we cannot predict cost or risk with enough precision to identify the patients that need more help – or target the drugs and therapies that are most effective. Risk predictions are always outdated, limiting our ability to improve care/disease management. The lack of EMR data blocks our ability to add new clinical results that are very helpful in predicting risk.
Fundamentally, the old approach is broken. We need new technology to handle the flood of new clinical data and provide more effective and efficient care.
- How do we measure the accuracy of our predictions? R2 scores measure predictive accuracy from .0 to 1.0 (or 0% to 100%). Higher scores are better, e.g. 100%. This is a standard measure for predictive analysis. For example, the Society of Actuaries used R2 scores in their tests of health care predictive modeling systems in 2002 and 2007.
Artificial Intelligence Technology = Improved Accuracy
Sophisticated ‘artificial intelligence’ (AI) technology has been used for years by the financial and defense industries to analyze complex data. UHealthSolutions is using the same approach with complex health care data, including:
- Neural networks: Software that can be trained to recognize patterns and identify groups, e.g. high risk patients.
- Genetic algorithms: An approach that mimics colonies of bacteria that change and grow to identify key factors, e.g. the drivers behind health care risk.
- Fuzzy logic: A more flexible way to analyze complex data, improving accuracy.
- Rule systems: This tool helps capture what we know about health care, add reference data and plug-in useful statistics.
This combination of AI techniques radically improves accuracy and flexibility. With a standard cost prediction, for example, we can deliver precision of .75 to .95 (75% to 95%). The latest clinical data from EMR, HIE and HIS systems can be leveraged – to provide immediate results for each patient.
Current 'predictive modeling' products cannot handle this complexity and chaos: the Age of the Simple Algorithm is over. With advanced AI technology, we can quickly identify at-risk patients to provide the care they need -- and reduce costs.
UHealthSolutions at 2012 PBMI Conference
If you are attending the PBMI conference, please drop by the UHealthSolutions exhibit at Booths 15 and 16. John Ormond, Director of Healthcare Intelligence Solutions, will be available for individual discussions during the conference. To schedule a meeting at the conference, please call 774-823-0268 or email SPigeon@UHealthSolutions.org.
About UHealthSolutions, Inc.
A not-for-profit affiliate of the University of Massachusetts Medical School, UHealthSolutions recently released three new solutions for health care: (1) the PIDA predictive analysis system, (2) the Healthcare Explorer intelligence platform and (3) the BEEs case and benefit management system. These solutions are available through UHealthSolutions for PBMs, payers, providers, hospitals, ACOs, health plans, TPA, brokers and health care software vendors – including not-for-profit and commercial organizations.
For more information, visit: www.UHealthSolutions.org.
Photos/Multimedia Gallery Available: http://www.businesswire.com/cgi-bin/mmg.cgi?eid=50172149&lang=en