Fighting Infection with Missile Defense Algorithms
Healthcare Powered by Data Analytics
The early identification of threats around the globe is a basic tenant in Lockheed Martin’s approach to missile defense, seeking to deliver technologies that predict and defend against attacks. Today, that same innovative technology is being integrated into healthcare technology. Using the same algorithm that tracks missile trajectories, Lockheed Martin’s Sia™ platform is configured to detect sepsis in large patient datasets.
Sepsis: Unremarkable Symptoms with Extraordinary Impact
Sepsis is the 10th leading cause of death in the United States. An estimated 750,000 cases are identified annually, and the mortality rate is nearly 40 percent. Severe sepsis is also one of the most common causes of death in hospital critical care units, according to research from Penn Medicine. Often associated only with those critically ill or the elderly, sepsis does not discriminate on basis of age or condition.
Caught early, sepsis can be treated with strong antibiotics. However, the timeline for a sepsis infection to turn life-threatening is often measured in hours. In fact, for every hour of delay in treatment, the mortality rate increases by 7 percent, underscoring the need for a sophisticated early warning system that follows changes in a patient’s condition.
The symptoms of a sepsis infection are common and often confused with the flu, resulting in a large number of false diagnoses. As a result, testing and treatment for sepsis are often initiated unnecessarily, putting a huge burden on hospital staff who must provide the around-the-clock care. Add to that the impact of costs. A recent study labels sepsis as the most expensive condition treated in hospitals today — accounting for more than $20 billion in annual costs to the U.S. healthcare system.
Insight Gained from Missile Detection
Automated sepsis detection systems aren't new, but the differences between Lockheed Martin’s Sia™ algorithmic platform and the conventional sepsis detection systems are timing and accuracy. The current Sepsis/Systemic Inflammatory Response Syndrome (SIRS) method of detecting sepsis accurately flags sepsis patients 69 percent of the time; however, it incorrectly flags 65 percent of uninfected patients as well. This means almost all patients are flagged as septic at one time or another.
In contrast, Lockheed Martin's approach continually monitors changes in patient vital signs and blood work and uses sophisticated data analytics to flag sepsis cases. Initial findings in a trial that included more than 4,500 patients disclosed Lockheed Martin’s solution correctly identified cases of sepsis more than 90 percent of the time and, on average, 14 to 16 hours before the conventional approach reached diagnosis. Less than one percent of patients were incorrectly flagged as potentially septic.
“In missile detection, Lockheed Martin treats data as continuously changing signals, and we thought there was a strong analogy to look at vitals and lab data the same way," said Melanie Lang, Lockheed Martin business development lead within Information Systems & Global Solutions. "We took our insights from real-time, streaming sensor data, detecting missiles at mach speeds, and reapplied those same techniques to real-time patient data.”
Delivering Accuracy in Time for Recovery
“Our current sepsis algorithm is targeted for the general population,” says Mike Draugelis, chief data scientist at Lockheed Martin Information Systems & Global Solutions. “But with additional research, we can focus on categories of patients: the elderly, the very young, cancer patients and heart disease patients. By focusing on similar patient groups, we should be able to improve our accuracy even further to drive an even earlier diagnosis of sepsis.”
Industry stakeholders see a broad potential to use Sia™ for early detection and high accuracy across many conditions. The same data analytic techniques used to diagnose sepsis may also be able to predict myocardial infarction or the onset of diabetes and blood clots, among others.
“Lockheed Martin’s solution leverages the best in science, clinical protocols and data processing to achieve unparalleled levels of detection accuracy" said Heather Lavoie, President of Geneia, a Capital Blue Cross subsidiary that focuses on health care innovations. "We see broad potential to use this solution for early detection and high accuracy across numerous conditions to cost-effectively improve the health of our customers.”
All of this leads to better accuracy, lower mortality rates, higher recovery rates, and cost-effective improvements in health — a new level of healthcare defense for patients everywhere.
February 26, 2014
- Sepsis is the 10th leading cause of death in the United States. An estimated 750,000 cases are identified annually, and the mortality rate is nearly 40 percent.
- Using the same algorithm that tracks missile trajectories, Lockheed Martin’s Sia™ platform is configured to detect sepsis in large patient datasets.
- Lockheed Martin’s solution correctly identified cases of sepsis more than 90 percent of the time and, on average, 14 to 16 hours before the conventional approach reached diagnosis.