538
B. Farahani et al.
• Learning-based: IoT e-Health should be designed based on suitable data-driven
learning techniques to handle the varying dynamics of cyber and physical
components.
• Time-aware: Spatial and temporal variations in the dynamics of the cyber and
physical components of an IoT e-Health must also be addressed.
11.6.8 Data Management
In the healthcare sector, IoT e-Health faces many of the same data management
challenges as in other fields. One distinguishing factor, however, is the fact that eHealth data originates from medical sensors worn by human subjects, and the human
body is a constantly changing system. Thus, from an IoT e-Health perspective,
an ongoing flux of data will continually flow inward from edge sensors via fog
computing nodes. On a positive note, sensors and computing are both declining in
cost, making Big Data more cost-effective to be collected in a brief timeframe. IoT
e-Health has evolved to deal with the complicated nature of these data, even as their
variety, volume, and velocity have continued to increase. At the same time, IoT faces
a challenge almost unknown 10 years ago: that of data variety. Dozens of healthcare
applications targeted at end users use their own data format; for example, ECG data
is often encoded in XML, while camera-based IoT devices typically record data in
a variety of image formats. Meanwhile, various edge computer manufacturers use
their own data formats, which can also vary by customer. Data models on the cloud
also vary widely, creating a desperate need for standardization. Difficulties related
to data volume and velocity, on the other hand, are more related to the ability of
the fog node hardware to acquire, analyze, store, and transmit data from medical
devices (which could be located at hospitals or clinics or carried with the patient)
at high fidelity and resolution. This creates a clear demand for fog administrators
capable of supervising the data flux between computing in the cloud and the fog
[1, 9].
11.6.9 Context Awareness
A complete, detailed picture of a particular patient’s context enables a system to
determine its actual need for assistance more appropriately. In the area of IoT
healthcare, events and data must be accurately interpreted in order to obtain a
reasonable, holistic understanding of contextual data. For example, based on a
patient’s condition and context, not every detected fall indicates a critical case
or requires a reaction by a caregiver or health provider. In general, the creation
of context-aware healthcare applications is challenging due to the issues of data
acquisition (e.g., determining which data is important enough to capture) and data
analysis methods, including presentation of context-based services and information.
B. Farahani et al.
• Learning-based: IoT e-Health should be designed based on suitable data-driven
learning techniques to handle the varying dynamics of cyber and physical
components.
• Time-aware: Spatial and temporal variations in the dynamics of the cyber and
physical components of an IoT e-Health must also be addressed.
11.6.8 Data Management
In the healthcare sector, IoT e-Health faces many of the same data management
challenges as in other fields. One distinguishing factor, however, is the fact that eHealth data originates from medical sensors worn by human subjects, and the human
body is a constantly changing system. Thus, from an IoT e-Health perspective,
an ongoing flux of data will continually flow inward from edge sensors via fog
computing nodes. On a positive note, sensors and computing are both declining in
cost, making Big Data more cost-effective to be collected in a brief timeframe. IoT
e-Health has evolved to deal with the complicated nature of these data, even as their
variety, volume, and velocity have continued to increase. At the same time, IoT faces
a challenge almost unknown 10 years ago: that of data variety. Dozens of healthcare
applications targeted at end users use their own data format; for example, ECG data
is often encoded in XML, while camera-based IoT devices typically record data in
a variety of image formats. Meanwhile, various edge computer manufacturers use
their own data formats, which can also vary by customer. Data models on the cloud
also vary widely, creating a desperate need for standardization. Difficulties related
to data volume and velocity, on the other hand, are more related to the ability of
the fog node hardware to acquire, analyze, store, and transmit data from medical
devices (which could be located at hospitals or clinics or carried with the patient)
at high fidelity and resolution. This creates a clear demand for fog administrators
capable of supervising the data flux between computing in the cloud and the fog
[1, 9].
11.6.9 Context Awareness
A complete, detailed picture of a particular patient’s context enables a system to
determine its actual need for assistance more appropriately. In the area of IoT
healthcare, events and data must be accurately interpreted in order to obtain a
reasonable, holistic understanding of contextual data. For example, based on a
patient’s condition and context, not every detected fall indicates a critical case
or requires a reaction by a caregiver or health provider. In general, the creation
of context-aware healthcare applications is challenging due to the issues of data
acquisition (e.g., determining which data is important enough to capture) and data
analysis methods, including presentation of context-based services and information.