Reliable Clinical Monitoring using Wireless Sensor Networks: Experience in a Step-down

Hospital Unit

Yetta

Outline

• Introduction• Monitoring system• Clinical study• Clinical deterioration detection• Conclusion

Introduction

• Clinical deterioration detection• ICU / step-down unit / general care unit

• IEEE 802.15.4 / IEEE 802.11

• Heart rate (HR) and blood oxygenation (spO2)

Monitoring System

• TelosB / OxiLink pulse-oximeter

Monitoring System

• CTP (collection tree protocol)– Low reliability because of user mobility

• DRAP (Dynamic Relay Association Protocol)– Isolate the mobility from multi-hop routing• Single-hop to first relay• Relay to base station

node cost to root

B 2

C 2

D 1

E 0

E

D

CB

neighbor table of node A

A

Monitoring System

• Radio power management• Sensor component (OxiLink pulse-oximeter)– Control by TelosB– average over 8 sec

• Logging component– Batching flash writing

Clinical Study

• 1200m2

• 18 relays• 41 patients• Pulse and oxygenation were measured at 30-

and 60-second intervals

Reliability

• Network reliability• Sensing reliability

• Time-to-failure • Time-to-recover

Network Reliability

Mean = 22.4 min

95% <2.5 min

Sensing Reliability

• Significantly affected by patient movement, sensor disconnections, sensor placement, and nail polish

Improvement of Sensing Reliability

• Oversample

• Median reliability: 84%(30sec), 75%(60sec)

Median = 1.81 min 75% < 1 min => short burstLong-tailed => sensor disconnection

Improvement of Sensing Reliability

• Disconnection alarms

Clinical Deterioration Detection

Clinical Deterioration Detection

• CUSUM algorithm– detecting statistically significant changes in a

series of measurements– Sliding window

Conclusion

• High network reliability• System reliability dominated by sensor

reliability– Oversampling– Disconnection alarms

• Show the potential of real-time detection of clinical deterioration