Telemedicine: ANALYSIS OF DEVICE AND PATIENT DATA THROUGH HIGH PERFORMANCE COMPUTING ALGORITHMS TO FIND SECURITY EVENTS
Lead Institution: University of Washington
Project Leader: Nathanael Paul (Subcontract- University of Tennessee)
Current and new medical device systems are increasingly connected to manufacturer systems. In diabetes, patient systems generate and upload data to manufacturers, and this data helps a patient achieve glucose control. Through analysis of this data, this project will address patient authentication and data authenticity in a specific chronic condition– diabetes.
Focus of the research/Market need for this project
Patients with diabetes must perform intense self-management of their disease. Much of this management involves uploading data, downloading data, and analyzing data. If a manufacturer or patient receives data, the data may not be authentic or even artificially generated. Manufacturers, caregivers, and patients need to trust data that they rely on for the patient’s care, but current systems do not provide mechanisms to support this trust.
The goal of this project is to identify specific issues in trusting diabetes system data, developing approaches to analyze this data, and to implement algorithms to give higher assurance about the analyzed data.
Key Conclusions/Significant Findings/Milestones reached/Deliverables
We began by analyzing current diabetes systems to understand data requirements and how that data may be used. After identifying potentially useful data for increasing trust, we evaluated how we could tie this trust to a patient’s system. We approached this problem by considering 1) sensors in a mobile device (e.g., a phone) and 2) prototype sensors that could easily be integrated in a diabetes system. For example, accelerometers are a potential biometric as diabetes patients may exhibit identifiable traits when experiencing hypoglycemia (we have anecdotal evidence of this in a patient’s gait). In addition to accelerometer data, we are also considering using blood glucose data and bowel sound data. In parallel, we have considered and will address power consumption as it will be a limiting factor in deployment.
To speed up our effort, we received software from another research group that is collecting accelerometer data, and we are now modifying this software to collect data on multiple sensors. We will soon be finished with our software implementation to collect biometric data for analysis.
We have discussed implementing a future study with a diabetes research hospital (Mills-Peninsula in San Mateo, CA) and an insulin pump system company.
Materials Available for Other Investigators/interested parties
A research paper about our work is under preparation.
Market entry strategies
We have described our work to one specific insulin pump system sensor manufacturer, and they have verbally agreed that they would like to proceed with a larger clinical trial where we would collect specific biometric data on patients with diabetes. We will use the software that we developed in SHARPS, and this software will collect the biometric data for our study. Later this month, we will be discussing steps to perform this work with this company. We anticipate that this project will be successful, and our work would be adopted by this company (or others).
Increasing Trust in an Artificial Pancreas
Matthew Barthe and Nathanael Paul
Under Review, 2014