|IBM Watson™ made its debut in dramatic fashion as a contestant on the television game show Jeopardy!, where it showcased its amazing ability to digest and index astonishing volumes of information and process natural language queries in an instant. Watson used these advances to arrive at a correct answer—phrased in the form of a question, of course. The results were impressive, but the game show was not even playing to Watson’s strength as a cognitive computing engine.||
|Editor’s note: The author, Dr. Alex Philp, spoke with Ginny Lee, vice president, IBM US Federal Government, and Dr. Guruduth Banavar, vice president, IBM Research, for the background material to write this article.|
The Watson cognitive computing engine is rapidly evolving to recognize geometry and geography, enabling it to understand complex spatial relationships. As Watson combines spatial with temporal analytics and natural language processing, it is expected to derive associations and correlations among streaming data sets evolving from the Internet of Things. Once these associations and correlations occur, Watson can combine the where, what, and when cognitive dimensions into causal explanations about why. Putting the where into Watson represents a key evolutionary milestone into understanding cause-effect relationships and extracting meaningful insights about natural and synthetic global phenomena.
Employing cognitive processes
Until now, computing has relied heavily on tabular analysis to examine relationships between geospatial and temporal data. The where and when are correlated to understand what has occurred. While it is valuable for businesses and government organizations to answer the what questions about their operations, the far more important explanation of why something has occurred has remained elusive.
Put simply, cognitive computing means that Watson has the ability to reason and learn over time. It grows smarter. Just as humans do throughout their lives, Watson develops hypotheses, learns from outcomes, and is prepared with better future responses than past ones through continuous feedback loops—aka cognitive processes. In deriving answers, Watson can relate the question to its vast storehouse of knowledge, the wisdom it has accumulated from the experience of generating massive hypotheses and testing them, or a combination of the two.
This process allows Watson to understand and know things it didn’t know before—answers that were not previously entered into its internal database. For example, Watson can determine, with stated degrees of probability, why an event has taken place. This type of cognitive intelligence is referred to as the third wave of computing.
As the paradigm evolves, Watson continues to learn by recognizing shapes and spatial relationships—proximity, distance, boundaries, and patterns—that are the geometry of geography. In other words, the where is being added to Watson’s cognitive capabilities. By combining spatial and temporal components of data, Watson will derive meaningful cause-and-effect relationships from incredibly diverse data sets comprising networked sensors and human inputs numbering in the billions. And it will accomplish these relationships in real time. Imagine when Watson becomes directly connected to these networks of sensors and is able to parse the spatial-temporal components from the trillions of observations it makes by the second.
Deciphering complex associations from disparate data sets, especially those streaming at high velocity, is becoming increasingly crucial as the era of the Internet of Things dawns—the natural evolution from the wired world web in which we now live. Consider mobile devices. Smartphones contain an average of 15 sensors, which are the future eyes and ears and sensing media for Watson. Moreover, the array of sensors and devices that abound today—computers, cameras, satellites, and drones—are all collecting and transmitting data, and combined with mobile devices open the floodgates for torrents of potentially valuable data slipping past us.
As these devices and sensors become interconnected—and they will—extracting intelligence from these waves of big data in motion, and doing it in real time, will become the exclusive domain of the cognitive computing platform. This platform links streaming analytics to Watson in the same way sensory organs link to the human brain. Each piece of data will have a location- and time-stamped component. Each component must be analyzed immediately in its proper context to draw insights that can fundamentally change the way vital sectors of the economy operate such as healthcare, energy, retail, and many other areas.
Although the list of industries and markets that can benefit from Watson’s unique ability to process big data is long, massive growth in medical, energy, and retail demonstrates the potential for enormous immediate impact on those areas. Of these, healthcare may be where the impact is most dramatic. I personally sit on the board of the Providence St. Patrick Hospital in Missoula, Montana, where I have witnessed up close the challenges facing the medical field right now. Many of them are related to data overload.
Medicine has traditionally focused on treating sick people to make them well, and their business model has been fee-for-service–based. Patients pay a doctor to treat their ailments. But as anyone who pays attention to the news knows, healthcare costs have skyrocketed. Among the many changes this situation has necessitated is a gradual shift in emphasis to a concept called population wellness. The idea is to cut costs tomorrow by enhancing the health of the population today.
Getting people to quit smoking, eat healthy foods, and exercise more often are obvious contributors to improving overall health. But wellness very quickly gets much more complicated than changing just one factor. For example, medical professionals have long noticed that some diseases are more prevalent among people in specific geographic areas. Is the cause environmental or behavioral? Or do the people in a given region have a similar genetic composition that predisposes them to a disease? ZIP code and genetic code can play big roles in investigating why people get sick.
Each of these possible explanations can be broken down into thousands of genetic, environmental, and behavioral variables. The geography of medicine has been lacking the right technology to answer the questions that require correlating thousands, if not millions, of seemingly unrelated real-time and historical data points—until now. Watson has the ability to analyze disease incidents, genetic codes, and geographic boundaries to help determine the factors contributing to an illness and suggest ways to prevent it. Watson can consume and combine billions of data points representing the convergence of the genetic code (individual human), bar code (electronic medical record), and ZIP code (the geography of the patient population) and derive new insights. These data points combine the what, when, and where into the why.
The ultimate goal is to improve health by preventing disease before it even occurs through advance treatment and education. The concept is hardly new, but cognitive computing will take it to a new level. If doctors know what disease has occurred along with where and when, Watson may be able to tell them why. By comparing the genetic makeup of an ethnic group that is predisposed to a certain affliction, Watson may identify the environmental toxins or lifestyle choices that serve as triggers for that particular sickness. The cognitive computer has the ability to suggest changes that would alter the predicted outcome for the better.
Analyzing enormous volumes of medical data in existing databases is a daunting enough challenge that has huge potential payoffs. Geographic Communication Systems (GCS)—a geoanalytics company delivering leading-edge solutions through cloud, mobile, and analytical services—is focused on that endeavor now. Looking down the cognitive computing road toward the near future, the Internet of Things will enable Watson to monitor real-time heart monitor feeds from hospitals, over-the-counter drug sales at pharmacies, and population moods reflected in social media. Using this data, Watson can predict where the next outbreak of a disease is about to happen, which will allow medical authorities to take appropriate action. Not only will this capability curb healthcare costs, but it will save lives.
Watson’s cognitive computing capabilities are lighting the fuse on what can be called an analytics-as-a-service paradigm for big data. Some large organizations will invest in building their own cognitive computing solution, but most are expected to be interested in the answers that can be derived from the Watson-enabled Internet of Things. And those organizations may pay handsomely to service providers that can extract specific insights in real time and deliver them in packages that can be easily and immediately ingested by the Watson-enabled global business enterprise.
Please share any thoughts or questions in the comments.
Nature of analytics video: Dances with Rhinos
Case study: Find out how to build a customer-focused business
Case study video: Discover how IBM Entity Analytics can prevent fraud
Case study video: Learn how IBM expertise minimized traffic congestion
Case study: See how an advanced framework can enhance experimentation
Big data in a minute: Transforming unstructured data into enhanced healthcare