sensor networks for medical care authors: victor shnayder, borrong chen, konrad lorincz, thaddeus r....
TRANSCRIPT
Sensor Networks for Medical Care
Authors: Victor Shnayder, Borrong Chen, Konrad Lorincz, Thaddeus R. F. Fulford Jones, Matt Welsh
Presenter: Velin Dimitrov
Introduction
• Why do we need WSNs in medicine Continuously monitor patients long
term Emergency/Disaster Scenario
• “Active Triage Tag”
Immediate life-critical notifications Augment/replace existing wired
telemetry systems Improve overall care of patients
Motivation
• Real time continuous patient monitoring In-Hospital setting
• Home Monitoring – Elderly/Chronic Continuous data Long term care/trend analysis
• Collection of clinical data
Adapted from Matt Welsh’s presentation on CodeBlue at UCSD
Medical Applications
• Stationary nodes with low data rates to central station
• Improvements High Data Rates Reliable communications Multiple receivers
• In-network aggregation cannot be used
Current Implementations
• Wireless medical monitors EKG Pulse Oximeters Fetal Heart Rate Maternal Uterine
• “Cut the cord” implementations• Bluetooth, WMTS, Wi-Fi• Systems do not scale well
Current Implementations
Develop tiny, wearable, wireless sensors for medical care and disaster response
• Scalable, robust wireless communication protocols
• Integrate real-time sensor data into medical care
• Explore a range of clinical applications
Adapted from Matt Welsh’s presentation on CodeBlue at UCSD
CodeBlue Goals
• Sensor modules compatible with Mica2, MicaZ, and Telos mote designs
• Pulse Oximeter• Two lead electrocardiogram (EKG)• Motion analysis sensor• SFF Telos design for wearable use• telosb_datasheet_rev 20111109 (1).
CodeBlue – Hardware
• Device Discovery• Publish/Subscribe multi-• hop routing• Query interface – simplicity• RF-based localization
Low power Bluetooth and 802.15.4 (WPAN)
CodeBlue - Software
• Wearable Sensor Platform “…large batter packs and protruding
antennas are suboptimal for medical use.”
Small, Lightweight, Wearable sensors
• Reliable Communications Data Availability How much packet loss is acceptable? Sample rates vary 1Hz to 10’s kHz
Requirements
• Multiple Receivers Multicast capabilities
• Device Mobility Multi-hop routing Device Discovery
• Security Health Insurance Portability and
Accountability Act
Requirements
• A review of the implementation of the HIPAA Privacy Rule by the U.S. Government Accountability Office found that health care providers were "uncertain about their legal privacy responsibilities and often responded with an overly guarded approach to disclosing information...than necessary to ensure compliance with the Privacy rule".
Wilson J (2006). "Health Insurance Portability and Accountability Act Privacy rule causes ongoing concerns among clinicians and researchers". Ann Intern Med
HIPAA
• Disaster Response Research Funded US National Library of
Medicine SMART AID-N WiiSARD
• Centralized systems Reliability and Scalability Concerns
Related Work
Wireless Medical Sensors
• Mature technology (1970s)• Measures heart rate and SpO2• Catch hypoxemia before visible
symptoms• Array of Infrared LEDs• Array of IR detectors
650nm and 805nm
• BCI Medical Micro-Power Pulse Oximeter
Pulse Oximeter
• Two different types• Standard EKG
30 sec of data from 12-15 probes Diagnose wide range of cardiac
arrythmias
• Continuous EKG 2-3 probes Diagnose intermittent problems
EKG
• Single pair of electrodes• INA321 CMOS instrumentation
amplifier• 94dB CMRR• Gain of 5• TinyOS samples at 120Hz
Mote-Based EKG
• Parkinson's Disease and Stroke• Wired systems with many wires
carried in a waist harness• Sensors are placed on limbs of
interest• Accelerometers• Gyroscopes• EMG
Motion Capture Systems
• Wireless• 3 Axis Acc – STMicroelectronics• 1 Axis Gyro – Analog Devices• 1 EMG unit – Motion Lab Systems• One mote is placed on each
segment of interest
Mercury Motion Analysis Board
• Mica2Dot No 802.15.4
• Pluto TI MSP430 ChipCon CC2420 radio 120 mAh Li-Ion battery Mini USB
Pluto Mote
CodeBlue Software Architecture
• Sensors publish data to relevant channel
• Requirements Requested data rates/local filters to
limit bandwidth Multi-hop routing Mobility of senders/receivers in
calculation of routing paths
Publish/Subscribe Routing Layer
• Adaptive Demand-Driven Multicast Routing
interface PubSub {command result_t publish(uint16_t chan);command result_t subscribe(uint16_t chan);command result_t leave(uint16_t chan);command result_t send(uint16_t channel,
uint8_t length, TOS_Msg* msg);event result_t sendDone(TOS_MsgPtr msg,
result_t success);event TOS_MsgPtr receive(TOS_MsgPtr m,
uint16_t channel, uint16_t srcAddr);}
ADMR
• Implemented by forwarders• Rebroadcasts messages to a given
channel with duplicate suppression• Route discovery
Node table indexed by Publisher ID Each entry has path cost and previous
hop to the publisher Path costs are updated continuously
Multi-Hop/Multicasting
• PDR – path delivery ratio• CC2420 radio provide Link Quality
Information (LQI)• LQI mapped to Link Delivery Ratio
(LDR)• Summed for the path making PDR• Link Cost is 1-PDR or Path Loss
Ratio• PDR is held and updated in the
header of the ADMR messages
Calculating Path Cost
• Very simple• Periodically all nodes broadcast
metadata Node ID Sensor types
• Receivers subscribe to the broadcast channel
Discovery Protocol
• CBQ query are supplied by the GUI• Instructs CB to publish data to a
specific channel that meets query conditions
• S – Node IDs• Tau – Sensor Type• Rho – Sampling Rate• Chan – Channel to publish to• C – Total Number of Samples
Query Interface
• ADMR and CBQ are separate• Simplifies CBQ protocol• Inefficiencies arise that ADMR and
CBQ both flood the network with broadcast requests
• CBQ may not be sufficient in all situations but is for most
Inefficiencies
• Generic interface for sensors• getData()• dataReady()
Sensor Interface
RF Based Location Tracking
RF Based Location Tracking
MoteTrack, Konrad Lorincz, Harvard University
User Interface
Evaluation
Scalability - Location
Scalability – 1 Reciever
Scalability – 3 Receivers
Fairness
Packet Jitter
Mobility
Multiple Transmit Packets
• Sharing bandwidth across sensors Data priority
• Security Private Key encryption Public Key protocol
• Integration outside of a hospital setting
Future Work
Progress
• From 2009• http://www.ted.com/talks/eric_topol
_the_wireless_future_of_medicine.html
• 2:45 minutes
Questions/Comments/Discussion
• What characteristic make this a CPS?
• What potential challenges exist in the successful real-world implementation of CodeBlue?
• How does the boom in smartphones and ubiquitous computing help CodeBlue or does it make it obsolete?
• Is current encryption technology adequate to secure the system?
Questions