Page 147 - ITU KALEIDOSCOPE, ATLANTA 2019

ICT for Health: Networks, standards and innovation

forest. Then, we extract atomic care demands according User Layer Nature Language User Living
Geriatric Disease
to their slot values. Additionally, an expert knowledge Information Care Needs Requirements
forest is built by expert knowledge trees which integrate Knowledge
Expert
expert knowledge and care demands for every geriatric Demand Layer
disease. Thereafter, it is a requisite to obtain smart services Care ĊĊ Care
for providing geriatric care for the elderly by the smart Demand 1 Demand N
home, based on mapping similarity consisting of functional Atomic Atomic Atomic Atomic Atomic
Care
Care
Care
Care
Care
similarity and non-functional similarity between atomic Demand 1 Demand 2 Demand 3 Demand 4 Demand G
demands and smart services. In order to make the selection Service Layer Smart Smart Smart
of sensing devices unambiguous, we convert the problem Service Service Service
S3
S2
of selecting suitable sensing devices to realize selected smart S1
services into a multi-objective knapsack problem of searching Device Layer Sound Infrared Smart Switch
Pressure
for the optimized solution. In this paper, SAFSA is selected Sensor ĊĊ Sensor Sensor Bracelet ĊĊ Sensor
to search for the optimized solution, which could be replaced
by other optimization algorithms. Finally, we decode the Figure 2 – The system workflow in UDSD model
optimized artificial fish to recommend the user-oriented and so on. As shown in Definition 1, the care demand
SDSP for the target elderly, including types, commodity
selections, quantities and installation positions of sensing contains ID, Label, Description, Children and Quality
Requirement(QR), where ID the unique identifier of one
devices.
care demand, Label are keywords of functional description,
Description expresses user demand in natural language,
4. THE SMART-DESIRE MAPPING METHOD
and QR expresses quality constraints of required service.
Moreover, Children contains the ID of its sub-demands, if this
4.1 The UDSD architecture for data formalization
care demand is a composited demand. Generally, QR consists
As the informal descriptions of user demands and smart of some non-functional attributes, including price, response
services limit the efficiency of automatically generating time, availability, reliability and so on. In this paper, the
optimized SDSP, we proposed a UDSD architecture to non-functional attribute QR with accuracy, response_time,
formalize elderly information, care demands, smart services smart_level, availability and reliability.
and sensing devices, and which consisted of a user layer,
demand layer, service layer and device layer, as shown in Definition 1 Every care demand in this paper contains ID,
Figure 2. Label, Description, Children and Quality Requirement.

• User layer: The concept of the user layer is used De = {ID, Label, Description,Children,QR}
to sort and formalize key information of diseases
description, natural language care needs, and other
living requirements of elderly.
4.1.2 Service Layer
• Demand layer: After we extract expert knowledge
in expert diagnosis, medical literature and clinical In order to obtain the suitable smart services for geriatric
diagnosis of geriatric diseases, the demand layer fuses care automatically, we need to formalize smart services into
user demands and expert knowledge. Specifically, care digital description in the service layer, including ID, Label,
demands are divided into composite care demands and Description and Quality of Service(QoS). Therefore, smart
atomic care demands. services in this paper are formalized with three modules.
• Service layer: It is proposed to formalize technologies 1. Definition: The definition of the smart service indicates
and services in recent research into smart services with a the functional description for geriatric care, including
fixed format, including name, label, description, quality ID, Label, Description.
of service and input requirement.
2. QoS: This represents the non-functional parameters of
• Device layer: This layer formalizes digital descriptions the smart service, including accuracy, response-time,
of sensing devices, includingprice, precision, measuring smart-level, availability and reliability.
range and so on, which are important in analyzing the
ability and performance of sensing devices. 3. Input requirements: It is proposed to summarize the
hardware requirements of the smart service.

4.1.1 Demand layer 4.1.3 Expert knowledge forest
Generally, care demands are extracted in expert diagnosis Knowledge excavation is the mining of potential patterns
of geriatric diseases, which represents the real care needs and behaviors by highly automated analysis of legacy data
of elderly for 24-hour geriatric care in the smart home, to help people make the right decisions. The expression of
including diet care, sport care, daily care, danger warning expert knowledge directly affects the efficiency of knowledge

– 127 –