Database Reference
In-Depth Information
formatics. We divide this application space into two parts: clinical and
non-clinical applications. Clinical applications are essentially clinical de-
cision support applications for both in- and out-patient scenarios. Non
clinical applications include wellness management, activity monitoring,
the use of smart environments (e.g., smart home scenarios) and reality
mining. We provide a detailed survey of the sensors, systems, analytic
techniques, and applications and challenges in these different areas, in
this chapter.
The rest of this chapter is organized as follow. In Section 2, we present
research challenges associated with the mining of sensor data in medical
informatics. In Section 3.1, we review sensor mining applications and
systems in clinical healthcare settings, while in Section 4 we describe
several applications in non-clinical settings. We conclude in Section 5.
2. Mining Sensor Data in Medical Informatics:
Scope and Challenges
Sensors measure physical attributes of the world and produce signals,
i.e. time series consisting of ordered sequences of pairs (timestamps,data
elements). For example, in intensive care, respiration rates are estimated
from measurements of the chest impedance of the patient. The resulting
time series signals are consumed either by a human or by other sensors
and computing systems. For instance, the output of the chest impedance
sensor may be consumed by an apnea detection sensor to produce a sig-
nal measuring apnea episodes. The data elements produced by sensors
range from simple scalar numerical or categorical values, to complex data
structures. Examples of simple data elements include measures such as
hourly average of temperature in a given geographical location, output
by a temperature sensor. Examples of more complex data elements in-
clude summaries of vital signs and alerts measured by a patient monitor
sensor in a medical institution. In this chapter, we focus on sensing
challenges for medical informatics applications.
2.1 Taxonomy of Sensors used in Medical
Informatics
As shown in
Figure 14.1
, we categorize sensors in medical informatics
as follows:
Physiological sensors: These sensors measure patient vital signs or
physiological statistics. They were first used to measure vitals on
astronauts before appearing in medical institutions, at the bedside
in the 1960s. Today, physiological sensors are also available out-
