By Jane Clabby, Clabby Analytics
Clabby Analytics has been following the cognitive computing market for several years, including the activity from IBM with their Watson product line and other related solutions. However, I was recently contacted by Saffron Technology, a cognitive computing, big data and analytics company that got its start working with national security customers, and after reviewing their website, I was eager to hear more.
In a briefing with Gayle Shepard, chairman and CEO of Saffron Technology, I learned that the company has closed a $7 million Series B investment round which will be used to further accelerate business growth in commercial markets, including opening a new global headquarters in Silicon Valley. I also found out more about their impressive “Natural Intelligence Platform” and how it is being used in several different industries and applications. Use cases are always my favorite part of any briefing because they really bring to light the value of cognitive computing solutions that extend from saving time and money to, in the case of Saffron Technologies, saving human lives.
Saffron Technology has its roots in IBM– the founders left IBM Research’s Intelligent Agents Lab in 1999 to form the company. After 9/11, Saffron focused on national security, using their technology to connect and assess seemingly unrelated data, events and people for potential risks. Since then, they have extended these efforts into similar areas, including defense and global risk assessment, as well as energy utility management, healthcare and manufacturing, using their platform for real-time operational risk assessment and decision support.
Saffron’s Natural Intelligence Platform
Saffron’s Natural Intelligence Platform is designed to work similarly to the human brain, but with computing power for scale, performance and accuracy. It uses a patented associative memory approach that can analyze both structured and unstructured data in real-time to find connections across many sources. The platform employs a continuous, incremental learning approach that looks at and absorbs new data, actions and actual results.
With its own technology as well as through open API connectors, hybrid data including information from RDBMS, web and “Deep Web”, data streams such as Twitter, and, text, email and voice can be collected and correlated. Saffron also looks at data using connections and counts, semantic analysis, statistical analysis and clustering and patterns. By finding emerging and converging patterns based on experience, similarity and prior outcomes, potential threats and opportunities can be identified based on a real-time understanding of past and future experience. Plus, with this knowledge, future predictions can be made without using models or rules.
Let’s consider a few examples that illustrate the capabilities of the Saffron Natural Intelligence Platform.
- Cocaine trafficking – 45% of the world’s cocaine enters the U.S. through one pipeline
The goal in this example was to find the people involved, the drugs themselves and any associated weapons. Pursuing a community known for aliases and false signals, Saffron had to look at all the data and adapt as it changed, searching for connections that would lead to the people and organizations behind the data. By unifying over 80 sources of dynamic data and learning through patterns in the data, law enforcement agencies were able to eliminate knowledge siloes, and respond to new events in real-time (2 hours) instead of up to 3 weeks. They were also able to significantly reduce the training time for new analysts.
- Global aeronautics manufacturer- wants to eliminate operational efficiencies and waste
A helicopter manufacturer was looking to reduce maintenance costs without sacrificing quality or impacting safety. With Saffron, they were able to unify over 40 difference data sources including pilots’ intuition and sensory recall, complete maintenance records and mechanics knowledge and experience to learn about each helicopter and anticipate when they required maintenance. Taking into account things like where a particular helicopter was flown, weather conditions in the area, previous repairs and continuous hours flown, maintenance schedules for each individual helicopter were developed based on this knowledge. This replaced a schedule that provided certain types of maintenance for all helicopters (for example, those that have been flown for 6 months). The notion here was to learn from “one” and apply to all based on conditional similarity. By using Saffron the manufacturer was able to improve its accuracy in predicting required maintenance from 66% to 100% and cut false alarms from 16% to 1%, drastically reducing maintenance costs without compromising on safety.
- Clinical Decision Support – Cardiology diagnoses
In this example, Saffron was used to distinguish between two cardiac diseases that present similar symptoms; restrictive cardiomyopathy and constrictive pericarditis. Both conditions require very different treatment plans so it is extremely important to get a correct diagnosis. In 24 % of cases, cardiologists get the diagnosis wrong so the goal was to improve diagnostic accuracy. Saffron enabled the analysis and correlation of data from 10,000 attributes per heartbeat/per patient with 90 different metrics in 6 locations of the heart collected 20 times per one heartbeat. This approach improved the accuracy of diagnoses to 90% (from the previous 76%).
- Predictive Analytics – Large global foundation
Saffron’s predictive analytics capabilities provide a 360 view of known risks and identify potential risks. Using a threat scoring system, analytics synthesize thousands of threats in incoming correspondence providing real-time diagnostic intelligence to assist a global risk assessment team in identifying the few very serious threats. By unifying a variety of data sources, scoring and ranking threat patterns in real-time and learning about changing behavior patterns, Saffron can identify emerging risks, predict threats and determine root cause. If a threat is classified as high-risk, it is automatically routed to the foundation’s intelligence specialists.
In addition to these applications, Saffron risk intelligence technology can be used in healthcare for fraud detection, personalized medicine and treatment, experience-based decision support and lifetime wellness. In defense, security and global manufacturing, use cases for Saffron include alias detection, opportunity and threat scoring, supply-chain efficiency, and lifetime consumer experience. Cable and mobile providers can use Saffron to better understand individual customers and their preferences in order to reduce customer churn, increase customer acquisition and to create “upselling” opportunities.
I like the notion of “personalization at scale” enabled by Saffron – taking the experience of many and applying it to one. A wide range of benefits can be realized by collecting vast amounts of big data from a wide range of sources, and applying the resulting analysis and predictions to an individual entity for a specific purpose.
In healthcare, we will see individualized treatment plans that will both lower healthcare costs and improve outcomes. For asset maintenance, on anything from cars to equipment on the manufacturing floor, costs and efficiencies will be improved as well as safety. In national security, thousands of potential threats can be distilled down to identify the one that really matters.
Cognitive computing solutions are designed to work like the human brain but can scale and consider far more information in much less time than the average person. Without the insights offered by analytics and cognitive computing, big data is just big data. With its new round of funding, robust natural intelligence platform and strategic shift to commercial risk intelligence and healthcare risk, Saffron Technology is well positioned to gain a strong foothold in this growing market.