Machine Learning
Don’t wait for quarterly reports to spot trends. We deal in real-time data analysis and solutions.
The vast majority of data generated across the globe is unstructured and never analyzed.
Natural Language Processing (NLP) is the key to transforming text data from an IT nuisance into a valuable asset. With streaming NLP, we are able to answer relevant questions like: “How is the perception of my brand changing in real-time with this product release?” Combining streaming data frameworks such as Apache Storm with advanced machine learning NLP algorithms allows us to extract brand sentiment from social media posts and detect changes in real-time.
A single patient’s electronic health record contains thousands of meaningful data points. Supplement a health record with demographic and geospatial data, aggregate over a whole population of patients, and the number of data points grows exponentially. Healthcare organizations need to extract critical insights from this data, but they cannot wait for analysts to generate reports.
For instance, prior to discharging a patient following a heart attack, a hospital needs to know the risk that the patient will have a secondary event leading to a 30 day all-cause readmission. If the proper information is available, the hospital can ensure appropriate referrals are made and post-discharge care is delivered. We employ adaptive machine learning ensemble models that can quickly churn through the masses of retrospective clinical, demographic, and geospatial data at each discharge to accurately gauge risk. These ensembles do not grow cold when clinical protocols are implemented and trends abate — like us they learn and adapt.
Did someone leave the lights on?
No matter the industry, from restaurant chains to manufacturing lines, outliers are the enemy. Actions as innocuous as leaving a light on after close are easily detectable in today’s world of connected devices and the Internet of Things (IoT). Along with the cloud infrastructure to monitor IoT connected devices, machine learning is needed to detect outliers. We build advanced machine learning algorithms that understand what “normal” is, allowing them to detect outliers and transmit alerts in real-time when a deviation is detected. We actually monitor our own electricity consumption in this way, so we always know when someone leaves the lights on.