Grid-based Medical Devices for Everyday Health Mobile Medical Monitoring Presented by David De Roure Overview of talk • • • • • • Partners Scenario Grid software Demonstration Current activity Closing thoughts Technical innovation in physical and digital life Henk Muller (Bristol), Matthew Chalmers (Glasgow), Adrian Friday, Hans Gellerson (Lancaster), Steve Benford, Tom Rodden (Nottingham), Bill Gaver (RCA), David De Roure (Southampton), Geraldine Fitzpatrick (Sussex), Anthony Steed (UCL) University of Nottingham Tom Rodden Chris Greenhalgh Alastair Hampshire Jan Humble John Crowe Barry Hayes-Gill Carl Barratt Ben Palethorpe Mark Sumner University of Southampton David De Roure Don Cruickshank University of Oxford Lionel Tarassenko William R. Cobern Oliver J. Gibson University of Lancaster Adrian Friday Oliver Storz Nigel Davies University of Glasgow Matthew Chalmers University of Bristol Henk Muller Chris Setchell Scenario • Patients are remotely monitored using a series of small mobile and wearable devices constructed from an arrangement of existing sensors • Information collected from these remote devices is made available using Grid technology • Medical professionals have tools to analyse online medical information and are able to access these through remote interfaces. Grid Research Agenda • Making remote data available to the Grid in order that a wider scientific community can access scientific data as quickly as possible, often across variable bandwidth communication services • Making Grid facilities available to remote users when these need to be delivered across lower bandwidth communication using devices with significant display and processor limitations The Maturing eScience “Grid” 1998 Broadening Computation Research Focus Additional Challenges New Uses New Scientists Resources Security Management Modelling Simulation 2001 2003 Knowledge Information Sensors Mobility Semantic Modelling Access Structure Metadata Knowledge Discovery and Recording Physics Pharmacy Medical Chemistry Astronomy BioInformatics Engineering 2005 Devices Activity Ubiquitous Autonomic Behaviour Remote Sensing Remote Access Architectures (e.g P2P, Ad-hoc networs) Environmental Monitoring Capturing Activity and Process Activity and Lab Monitoring Field Scientists Environmental Scientists “Wet” Lab Scientists MIAS - Devices • Exploring the development of mobile medical technologies that can be remotely connected onto a distributed grid infrastructure – Continuous monitoring of multiple signals via wearable devices – Periodic monitoring using Java phones and blood glucose measures • All signals available to a broad community and can be processed using standard Grid Services Java Phone + Blood Monitor Standard Grid Service for feature detection Proxy Buffers Material for sending on Grid based Storage Services Patients Visualisation Services Proxy Wearable Devices Converts Signals to database record Asynchronous Mobile World Display Clinicians Grid Services Wearable Device Sensor bus • • • • Easy Plug and Play of Sensors Wireless connection using 802.11 Positioning information from GPS Nine wire sensor bus running through wearable to allow new sensors GPS aerial Range of different sensors • ECG • Oxygen saturation • Body movement – Accelerometers – GPS • All plug and play to standard bus • Changes reported to the underlying infrastructure Blood Glucose Monitoring • Exploring medical devices that rely on self-reporting • Extends web based system developed by Oxford University and e-San Ltd • Off-the-shelf GPRS (General Packet Radio Service) mobile phone • Blood Glucose meter Self Reporting • Patient takes measurement • Measurement sent via mobile phone to remote infrastructure • Series of lifestyle questions asked as part of the clinical trial • Users promoted for compliance. • Current trial involves 100+ patients Deploying on the Grid Register new device Generic device proxy factory(s) Blood sugar meter PAR sensor JavaPhone proxy JavaPhone Data logger Iridium Other sensors ECG sensor accelerometer 802.11 S/w module S Data logger proxy D PAR sensor proxy S Multicast beacon 9-wire bus (pluggable) Device Proxy Management Client Device S Sensor Trial manager Add sensor to trial database Other sensor proxy S Cyberjacket proxy D ECG sensor proxy S S Other sensor proxy New GRID Port Types: D D Blood sugar meter proxy GPS receiver DF DeviceProx yFactory GPRS PAR sensor Cyberjacket (Bitsy) DF New device configuration S Sensor data-pump Sensor data-pump Sensor data-pump Sensor Database Service RDBMS Live monitoring display Sensor and device status display GPS live map Scrolling sensor charts Ōelipse of normalityÕ visualisation Dataflow user interface Data chooser/ fetcher Table views Graph views Putting devices on the Grid • Make devices and sensors available as if they were first class Grid Services • Two new application-independent port types: – a generic sensor, – a generic device (assumed to host a number of sensors) • Currently our devices require a proxy to match between these definitions and the sensor • Project was an early GT3 adopter for prototype – Grid Service model worked – concerns about security Sensor port type: self-description Name # Modify ?False Description 1 Mutabilit yConstant IdentifiedAs Description 1 Mutable False Expanded description, e.g. placement, accuracy, etc. MeasurementTemplate 1 Constant False The format in which measurements are reported MeasurementDiscardPolicyExtensibility MeasurementPublicationPolicyExtensibility ConfigurationExtensibility 1.. * 1.. * 1.. * 1 Constant False Constant False Constant False Acceptable XML schema types for the measurementDiscardPolicy SDE Acceptable XML Schema types for the measurementPublishingPolicy SDE Acceptable XML Schema types for sensor configuration SDE Mutable False Current status, e.g. in contact with proxy or disconnected ProxyStatus Sensor ID, names and type Sensor port type: Externally modifiable configuration Name # Mutabili ty Modify ? Description MeasurementDiscardPolicy 1 Mutable True The conditions under which the sensor should discard historical measurements MeasurementPublishin gPolicy configuration 1 Mutable True The conditions under which the sensor (proxy) should make a new measurement public 0.. * Mutable True Sensor-specific configuration information, e.g. sample rate Sensor port type: measurement Name # Mutability Modify ? Description Measurement 1 Mutable False The most recent measurement made by the sensor MeasurementCounter 1 Mutable False A running counter of measurements made MeasurementHistory 1 Mutable False The complete known history of measurements Demo at All Hands Meeting in Nottingham, 2003 Related activities Advanced Grid Interfaces for Environmental e-Science in the Lab and in the Field • The Antarctic Lake Carbon Cycling project • The Urban Pollution Monitoring Project See demonstrations or www.equator.ac.uk Live clinical record • Readings appear as a live database • Standard queries and interfaces can be used to manipulate the data • On-line services used to process the data • Exploits existing grid standards for reliability • Presents a range of different interfaces for clinicians • Provides range of feedback to patients. Portal for Information Access • Interactive access to live and stored information (e.g. visualised, excel) collected from wearable devices – For use by clinicians – Could be used by patients – Also needed by “pervasive support desk” • Accessible via pervasive devices, e.g. phone • Based on spatial model Patient Proxy of Mobile clinician Location ontology is-adjacent-to is-part-of AbstractSpace is-owned-by akt:Organisation has-usual-occupant subClassOf permits-access-to akt:Person is-part-of PartitionedSpace EnclosedSpace permits-access-to akt:PostalAddress subClassOf subClassOf has-postal-address Building is-part-of FloorTraversingSpace subClassOf is-part-of subClassOf subClassOf Floor is-part-of subClassOf subClassOf Corridor is-part-of Stairs Lift Room Ian Millard Semantic Pervasive Grid Fundamentally about Interoperability and inference Grid and Pervasive share issues in large scale distributed systems. e.g. service description, discovery, composition; autonomic computing. These can be aided with semantics. Pervasive applications need the Grid, e.g. Sensor Networks Grid applications need Pervasive Computing e.g. Smart Laboratory http://ubigrid.lancs.ac.uk/ Conclusion • We have demonstrated the collection of medical and contextual data from wearable devices using Grid infrastructure • We have demonstrated a means of access to that data by a variety of users including use of pervasive devices • We have provided an illustration of the important relationship between Grid and Pervasive computing www.equator.ac.uk