Swan(sea) Song – personal research during my six years at Swansea ... and bey...
Personalized robotic intervention strategy by using semantics for people with dementia in nursing homes
1. PUBLIC
PERSONALIZED ROBOTIC INTERVENTION STRATEGY BY USING
SEMANTICS FOR PEOPLE WITH DEMENTIA IN NURSING HOMES
FEMKE ONGENAE & FEMKE DE BACKERE
GHENT UNIVERSITY - IDLAB
2. 2
100,000 PWD in Flanders, exhibiting Behavioral Disturbances (BDs)
Up to
80%
Agression/Agitation
Sleep/wake disturbance
42%
Apathy
72%
9 - 63%
Delusions
3. 3
BDs can be prevented and reduced
by preserving dignity and promoting quality of life
Financial constraints and lack of technology
hinder the adoption of person-centric care principles
4. PERSONALIZED ROBOTIC INTERVENTIONS
WONDER project
2 nursing homes, ZoraBots, XETAL & Flemish Universities
Design of an Internet of Robotic Things platform
Enabling contextualized steering of robot actions & intervention strategy
Enabling personalized interactions
Design of personalized intervention algorithm to intercept BDs
9. Sensor data
Profile info
BD detection
algorithm
Assignment
algorithm
Task
assignment
algorithm
3 ALGORITHMS UNDER DESIGN...
Detected BD
Send Zora
Send message
to caregiver
10. USER-DRIVEN ONTOLOGY & ALGORITHM DESIGN APPROACH
Observation
s &
interviews
Mind
mapping
Decision tree
workshops
11. DECISION TREE WORKSHOP
Goal:
Capturing algorithms/decision processes that stakeholders/domain experts
perform intuitively (“experience thaught me how to do it, there is no process”,
“we don’t think about it”)
Formalize those processes in a decision tree
Capturing the context/parameters on which people base those decisions
Prioritize the different decisions that make up the process
Wonder:
How do you decide when & how to intervene when a PwD is exhibiting a BD?
4 workshops with each time 4 caregives of the involved residential care centers
13. DECISION TREE WORKSHOP
Knowledge engineer
Asked questions insight into which concepts needed
Example answers possible data/sub concepts
Order of questions insight into importance
14
16. DECISION TREES
Taking into account:
Type of incident & its priority (wandering, yelling,...)
Resident’s historical behavior (BDs exhibited, agression, ...)
Resident’s profile (liked/disliked activities, family (history), ...)
Staff information (who is working, locations, relationship to residents, ...)
Robot information (where are they, what are there capabilities, ...)
Previous interactions of robot and PwD
When robot is sent constant updates to staff & allowance to intervene
Staff preferred to send the robot as much as possible when there is low probability of
agression
Personalization of actions of the robot
Based on context & profile information create list of possible actions personalized to
situation & resident
Learning through reinforcement, what works and what not
20. AMBIENT-AWARE CONTINUOUS CARE ONTOLOGY (ACCIO)
HTTPS://GITHUB.COM/IBCNSERVICES/ACCIO-ONTOLOGY/
SSNIoTSAREFIoT
PhysicalAss
et
TaskAccio
RoleCompetenc
eACCIO
PersonProfil
eACCIO
Localization
ContextAcci
o
DUL
NurseCall FallRisk CareCoBots
General
Medical
SS
N
SARE
F
Time OWL-S
GEO-
SPARQ
L
V-card
Organ
ization
Unit Galen
Know
Rob
22. TASK ASSIGNMENT SERVICES
Yelling BD
Detection
Service
Wandering
BD
Detection
Service
Distance
Measure
to Robot
Service
Priority BD
Assessme
nt Service
Wandering
BD
Solution
Assignme
nt Service
Yelling BD
Solution
Assignme
nt Service
Robot
Task
Assignme
nt Service
Staff
Comm
Service
Semantic Communication Bus
Robot
Comm
Service
Adapters
LocationFilter ≣ hasContext some (BLETagObservation or XetalObservation)
23. RobotPlaysSoothing
MusicAction
RobotSolution
TASK ASSIGNMENT SERVICES
Yelling BD
Detection
Service
Wanderin
g BD
Detection
Service
Distance
Measure
to Robot
Service
Priority BD
Assessme
nt Service
Wanderin
g BD
Solution
Assignme
nt Service
Yelling BD
Solution
Assignme
nt Service
Robot
Task
Assignme
nt Service
Staff
Comm
Service
Semantic Communication Bus
Robot
Comm
Service
Adapters
NoiseObservation
SoundObservation
YellingPersonFault
YellingPersonFault
WanderingPersonFault
YellingPersonFault
High Priority
YellingPersonFault
Priority
RobotSooth
PersonSolution
RobotSolution
SendNurseNotification
RobotAction
RobotPlaysSoothing
MusicAction
Distance
RobotAction
Distance
26. ANNOTATION OF BDS BY STAFF
Wandering
4 types (purposefully, pacing, circles, random)
Escaping
Standing by the door
Escaped outside
Yelling
Short
Long
A B
A B
B
A
D C
29. CONCLUSION
User-driven approach towards design of personalized care cobot algorithms
IoRT platform
Contexualized task generation & assignment for robots
Plug-in of new situational & personalized services based on semantic reasoning
Realizing P4 Medicine through Semantic Reasoning enabling:
- Heterogeneous data integration
- Personalized event detection & intervention
Along with the ageing population, the number of people with dementia (PWD) is steadily increasing [1]. The Alzheimer Liga Vlaanderen estimates a prevalence of 100,000 PWD in Flanders. Nursing homes around the globe are pulling out all the stops to provide the best possible care for residents with dementia. Almost all PWD exhibit behavioral disturbances (BDs) [2] like agitation (e.g., wandering or aggression), mood disorders (e.g., depression or apathy), sleep disorders and psychotic symptoms (e.g., delusions or hallucinations).
These BDs can be prevented by non-pharmaceutical interventions [3], like personal interaction, revisiting positive personal memories and promoting comfort and quality of life of the PWD. However, because of increased strain on the available resources within healthcare, a dwindling number of caregivers needs to care for an increasing number of elderly [4]. This inhibits the staff from allocating a lot of their time to these interactions.
A robot solution could help to provide such person-centric care by interacting with the PwD in order to alleviate BDs [5]. By audio-visual stimuli or by engaging communication, humanoid robots can elicit memories with associated positive feelings that have a calming effect on PwD. This improves the well-being of the PwD, but it can also be used to temporarily distract the PwD until the staff arrives in acute situations. As manifestations of dementia and the stimuli they react to, vary widely amongst PwD, a personalized approach is required. The key idea is that personalized interaction by robots improves 1) the well-being of residents, thus reducing the prevalence of behavioral disturbances, and 2) can temporarily distract PWD until the alerted staff arrives in acute situations. Personalization is especially important for companion robots, as it has been shown that this leads to more enjoyable experiences for the person and prevents losing interest in in the long-term [6], [7].
In the WONDER project1, we are developing an Internet of Robotic Things (IoRT) platform to enable personalized robot interactions with PwD to reduce and intercept BDs. In an IoRT platform the robot is integrated in a smart en- vironment outfitted with a variability of sensors and wearables that capture the current context. A semantic cloud back-end then analyzes this captured information and combines it with other context information sources (e.g. profile of the PwD or day schedule of the institution) to extract valuable knowledge about the context and activities of the persons active in it. This derived knowledge is then used to steer the actions of the robot. The IoRT platform autonomously detects whether a BD occurred and determines which actions should be per- formed by the robot to respond to it in a personalized manner. Although several researchers have looked towards semantics to support the care for people with dementia, none of them focused on the combination of IoT and semantics in or- der to detect behavioral disturbances and derive the accompanying personalized intervention strategy [8], [9], [10].
7
Task Assignment Services: uses semantic reasoning to decide which tasks should be executed by the nurses and robots
Action Assignment Services: Path planning, charging, translating tasks into lower level steps
Before we dive into the different task assignment services, Femke will explain the methodology we used to design these services
Start: stakeholders each describe a complex situation pertaining to the goal of the workshop
Layout of the institution = central discussion point
Stakeholders are asked to play the “system”
This system has a task e.g. a BD is occurring assign it to a person or robot and give them something to do (action)
As the system, they have a bird-eye view, they can ask additional questions about the context (queries)
Not immediately answered
Discuss importance of question
Discuss example answers
Answer + visualize