Maintenance and repair for CT scanners may soon be more immediate, less frequent and more affordable following the upcoming expansion of Glassbeam Inc.’s anomaly detection technology.
The machine data analytics company elaborated on the development at the AAMI 2018 Conference and Expo in Long Beach, California, referring to it as a part of its approach for utilizing AI capabilities to detect and alert providers to changes in components of computed tomography scanners from tube temperature to waterflow. They plan to eventually include other critical imaging modalities such as MR.
“Instead of a human being saying that the temperature pressure has shot beyond portable range, the machine alerts you by looking up the historical data of the temperature reading and saying the temperature should be between this high range and this low range. That is the anomaly direction model,” Puneet Pandit, president and CEO of Glassbeam, told HCB News. “The machine will look at the historical data, create the threshold and then alert the engineers when the threshold is crossed.”
CT scanners are equipped with sensors for monitoring different variables such as water temperature, waterflow, air temperature, fan speed, and tube temperature. Though each sensor periodically records its readings to determine if tracked variables are in the normal range, the task of accurately identifying which sensor readings are in the normal range and which ones are not is complex, often leading many to use a rule of thumb to form manually-defined thresholds.
ML-based AD techniques use historical data to train a model that can be used for detecting anomalous sensor values.
With Glassbeam’s technology, providers can utilize machine learning-based AD techniques to predict anomalies from historical data sets and address issues earlier, saving millions in maintenance costs, as well as being able to plan out more efficiently strategic actions for the management of their imaging modalities.
In addition to detecting single abnormal readings, the technology may be used to detect combinations of these readings from two or more different sensors, further helping Glassbeam raise mean time between failures and machine uptime from the industry standard range of 96-97 percent to more than 99.5 percent.
The expansion is the second phase of an initiative launched in February in which machine learning was deployed to detect with high accuracy tube failure in CTs, seven to ten days prior to the actual occurrence of such events.
It also follows the recent partnerships established
with Brown's Medical Imaging, Radiographic Equipment Services, and Calamed, which along with all of Glassbeam’s other strategic partners will distribute the solution to providers.
Pandit says the introduction of these capabilities signify the direction that all players in the medical equipment industry should be looking toward as they are are necessary for strengthening the efficiency of connectivity among medical devices and the makeup of the Internet of Things.
“This is the time for somebody to come forward and tell the hospital owners that you can collect this operational data from these connected machines, and do these five use cases to become more proactive and predictive and save money, and increase revenue and reclaim revenues,” he said. “Every machine is becoming increasingly connected. Data is there. The value of analytics is a lot higher today because there’s no more extra work to be done. The work is being done in the cloud today and can be applied to any machine or hospital tomorrow.”
Support for these capabilities is backed
by the FDA in its new report on the quality, safety and effectiveness of medical devices, finding that data retrieved from maintenance and repair of devices provides “valuable insight” into how well they perform.
The expansion is expected to go live in July or August.