Predictive Analytics Technology

SBM is a foundational technology that is revolutionizing the way we derive actionable intelligence from large data streams. At SmartSignal, SBM has already established a proven commercial track record in predictive equipment health monitoring. SBM has a wide range of valuable applications to human-generated signals, including health and vital sign data; proteomic and metabolomic data; and speech, profiling and security biometrics. In the life sciences, VG-BIO is exploring SBM as an enabler of vigilant critical care monitoring, remote patient management and advanced cardio-diagnostics. These health-care applications could have tremendous impact on meeting the demands of an aging population and an overextended health-care system.

SBM technology works by forming a personalized model representative of how any complex system (for example, a human patient) behaves normally from available data streams or data patterns. The model is entirely empirical, embodying the relationships in the data without requiring explicit medical knowledge of them (i.e., it does not require a "blueprint" of the human body).

SBM improves detection of incipient problems by providing a dynamic baseline that acts like a set of moving alert limits around the sensor data. As a result, problems can be detected long before the sensor signal drifts outside conventional fixed alert limits.

When deployed in a vigilant monitoring role, SBM can be used to provide a dynamic baseline of normal system behavior which can be used to gauge whether current biosignals or vital signs are deviating. In contrast to standard methods of asking whether a measurement is simply "out of normal range", the SBM baseline enables residual-based analysis, which asks the more pertinent question "Is this measurement different from what it should be?" (even if it is in range). This provides automated advanced warning with increased lead time for medical intervention. When deployed in a diagnostic role, SBM personalizes to the patient's characteristics and provides visbility into subtle patterns thast reveal disease states.

This patented technology is currently developed in the MatLab analytic platform and can be rapidly prototyped for your application using our tool kit. If you are interested in a demonstration of efficacy and design, please contact us.