The document describes a smart healthcare monitoring system for independent living. Some key points: - The system collects data from sensors monitoring daily living activities, physiological signals, and the environment to determine a person's wellness and ability to live independently. - Sensors are deployed throughout the home to monitor activities like using appliances, mobility, and vital signs. The data is analyzed to recognize patterns and forecast wellness. - Wellness is determined based on indices measuring inactive time and excess usage of appliances. The indices are improved over time using dynamic thresholds. - Patterns in sensor activity are analyzed to detect irregular behaviors that could indicate issues. Forecasting is used to predict daily activities and identify deviations.

1. A Smart Healthcare Monitoring
System for Independent Living
N. K. Suryadevara
Ph.D student
Under Supervision
Prof. Subhas Chandra Mukhopadhyay
SEAT-Massey University

2. Courtesy: National Geographic Magazine
It’s not just
hype
New Science
could lead to
very long lives
Good News
But…

3. Need Assistance…….

4. How does someone die alone in their home without anyone realising?

5. • Ambient Assisted Living
• Wellness Determination
• Sensor Deployment
• Wellness Indices and Forecasting
• Sensor Activity Pattern Analysis
• Conclusion
• Q & A

6. The concept of Ambient Assisted Living
• To extend the time, people can live in
their preferred environment by increasing
their autonomy, self-confidence and
mobility;
• Support maintaining health and functional
capability of the elderly individuals;
Ref: http://www.aal-europe.eu/

7. www.standards.co.nz
www.stats.co.nz

8. Functional Blocks of Health Informatics System for
Wellbeing and Independent Living
Health
Informatics
System
Activities of
Daily Living
Monitoring
Physiological
Monitoring
Environmental
Monitoring
Wellness
Determination
Non -Invasive
Wearable and
Non-Wearable-
Unobtrusive

9. Health Informatics System collects more data
 Plenty of Data Collection methods/tools
 More resources are required for Analyzing the data
Proper Information can be gained from the analysis
(translation) of data
Important indications for proper decision-making
Notify policy development
Our Solution:
Web-based reporting tool to analyze and infer right
decisions

10. Key Components – Health Informatics System
Instrumentation
Sensing Objects
Wireless
Communication
Information
Processing
 Information and Communication Technology
• Compatibility of Sub-Systems
• Flexibility
• Robust
• Real-Time Processing of Data

11. Smart Home Monitoring System
AAL Services
(Energy Consumption)
(Human Physiological)
Recognition of
Human ADL’s
Human Wellness
Determination
β1, β2
Domestic Objects usage
Trend through
Time Series Data
Mining
Sensors Data
Acquisition
Remote
Interoperability
Internet
www.iots2is.org
Health Care
Provider/
Relatives/
Tele-Care Services
Data
Base
WSN

12. • How “Well” a person living alone in their home is
able to perform their essential daily activities in
terms of using household appliances?
• Performance of Daily Activities
Performance behavior
• Livelihood activities are Cyclic
• Monitor usage of household objects for
recognizing the habitual nature of the person.
Wellness of an Independent Living Person

13. Monitoring Basic ADL
Microwave, Water
Kettle, Toaster or (any
other item used
regularly in Kitchen)
Room Heater
Television/Radio
Bed, Couch, Chair, Toilet
Any other appliance
used as habitual
Preparation of Food
(Breakfast,Lunch,Dinner)
Sleeping
Toileting, Self Grooming
Dinning
Relax
Watching TV(while sitting
on Couch)
ADL α Household Appliance Usage
N.K.Suryadevara, S. C. Mukhopadhyay, R. Wang, R.K. Rayudu, “Forecasting the behavior of an elderly using
wireless sensors data in a smart home”, Elsevier: Engineering Applications of Artificial Intelligence, Available online
12 September 2013, ISSN 0952-1976, http://dx.doi.org/10.1016/j.engappai.2013.08.004.
Passive Infra Red(PIR) Sensors: Mobility(Movements monitoring)

14. Sensing Units-HMS

15. Human Emotional State Recognition Unit
An a ad Ne t a

16. Robust Supervisory Control Unit of
Home Monitoring System

17. Integrated system for simultaneous human emotion
and ADL recognitions

18. Medicine Dispenser Unit
ZigBee Based
Wireless
Receiver Unit
Micro Controller
based Opening +
Locking
Arrangement
Automatic
Electronic
Medicine
Dispenser
Real-Time Clock and
Control Unit, ZigBee
based Wireless
Communications
(Central Coordinator)

19. Sensor activity status
Subject #1 Subject #2
Subject #3 Subject #4
Suryadevara N.K, Mukhopadhyay S.C, “Wireless Sensor Network based Home Monitoring System for Wellness
Determination of Elderly”, IEEE Sensors Journal-2012, Vol: 12 Issue: 6, Page(s): 1965 – 1972

20. • Domestic objects are used at regular time
intervals in the day to day life
• Usage durations and the frequency of use
are varied
• “Human behaviours in constant contexts recur,
because the processing that initiates and
controls their performance becomes automatic”
• “Frequency of past behaviour reflects the habit
strength and has a direct effect on future
performance”

21. Selection of Sensors and Using Minimum Number of
enso s fo Monito in Basic ADL’s
Life Style of
the Elderly
Sensors for
basic ADL
monitoring
Determination
of minimum
sensors
| |
( ) 1/ ( )
l
c
l
c
s
c
loc TT
s S
f s

 
Frequency of Sensor usage
Room
Type
Sensor
Type
Connected to
Device
η
Trail
Test
Living Force,
Electrical
Couch, Chair,
TV, Heater
0.03,
0.05,
0.05,0.1
0.03,
0.04,
0.03,.1
Kitchen
Electrical Microwave,
Toaster,
Kettle
0.05,
0.05,
0.02
0.04,
0.06,
0.00
Bed Force Bed 0.29 0.37
Bath Force Toilet 0.35 0.33
Storage Contact Cupboard 0.01 0.00
N.K.Suryadevara, S. C. Mukhopadhyay, R. Wang, R.K. Rayudu, Forecasting the behavior of an elderly using wireless sensors data in
a smart home, Elsevier: Engineering Applications of Artificial Intelligence, Available online 12 September 2013, ISSN 0952-1976,
http://dx.doi.org/10.1016/j.engappai.2013.08.004.

22. N.K.Suryadevara, S. C. Mukhopadhyay, R. Wang, R.K. Rayudu, Forecasting the behavior of an elderly using
wireless sensors data in a smart home, Elsevier: Engineering Applications of Artificial Intelligence, Available online
12 September 2013, ISSN 0952-1976, http://dx.doi.org/10.1016/j.engappai.2013.08.004.
Block diagram of the Wellness Determination

23. Behaviour Detection
Regular or Irregular
Sensor Activity
Pattern
ForecastingWellness
Indices
To Minimize “Fa se A a ms”
N.K.Suryadevara, S. C. Mukhopadhyay, R. Wang, R.K. Rayudu, Forecasting the behavior of an elderly using
wireless sensors data in a smart home, Elsevier: Engineering Applications of Artificial Intelligence, Available online
12 September 2013, ISSN 0952-1976, http://dx.doi.org/10.1016/j.engappai.2013.08.004.

24. Sensor Deployment Layout

25. WSN-Communication
Using XBee Module
• IEEE-ZigBee Protocol, ISM 2.4 GHz frequency
Configuration:
Mesh Topology-Reliable Data Transmission.
Sampling Rate:
Depending on the Type of Sensing Unit.

26. Wellness Indices
Inactive usage measurement of appliances
β1 = 1 – t/T
t = Time of Inactive duration of all appliances
(i.e.) duration of time during which no appliances are used.
T= Maximum inactive duration during which no appliances are
used under normal condition.
Excess usage measurement of appliance
β2 = 1 + (1 – Ta / Tn)
Ta= Actual(Current) usage duration of a appliance.
Tn = Maximum usage time of appliance under normal
condition.
Suryadevara N.K, Mukhopadhyay S.C, “Wireless Sensor Network based Home Monitoring System for Wellness
Determination of Elderly”, IEEE Sensors Journal-2012, Vol: 12 Issue: 6, Page(s): 1965 – 1972

27. Limitations
• Seasonal variations such as day of the week,
weekly, monthly are not taken into
consideration therefore it is likely that more
false warning messages will be generated.
• The threshold value of wellness indices was
derived to 0.5 and has been considered as safe
limit beyond which a warning message is sent to
the elderly/healthcare provider regarding the
daily activity behaviour.
(Need for Dynamic Wellness Functions)
Suryadevara N.K, Mukhopadhyay S.C, “Wireless Sensor Network based Home Monitoring System for Wellness
Determination of Elderly”, IEEE Sensors Journal-2012, Vol: 12 Issue: 6, Page(s): 1965 – 1972

28. Improved Wellness Function (β1, new)
β1,new = e
−t
T

29. Improved Wellness Function (β2, new):
β2,new = e
Tn−Ta
Tn

30. Advantages of Improved Wellness Functions:
• For linear wellness indices (β1 and β2) the
threshold value was kept at 0.5 for generating
irregular behaviour warning messages.
• The improved wellness index β1,new, β2,new allow
more time to generate warning messages for the
same threshold.
• It was also observed that at 50% of the time
period the new wellness functions indicates a
wellness of 62%, better than 50% of the
previous wellness indices.

31. Dynamic – Maximum Inactive Usage and
Excess Active Usage Durations (T, Tn):
• T = δ (C1t – C1t-1) + (1 −δ) Tt-1
C1t = α (xt) + (1 − α) (C1t-1+ Tt-1) + St
Tn = δ (C2t – C2t-1) + (1 −δ) Tnt-1
C2t = α (xt) + (1 − α) (C2t-1+ Tnt-1) + St
• T: Trend of the Maximum Inactive usage Durations,
• Tn: Trend of the Maximum excess active usage durations,
• C1t, C2t: Seasonal trends;
• xt is the object usage observation at the current time,
• s is the number of periods in one cycle (week) (i.e. s=7),
Tt=(1/s)((xs+1-x1)/s+ (xs+2-x2)/s+…. (x2s-xs)/s)
• α, δ are the smoothing parameters (0 to 1), selected by minimizing mean
square errors.
• St is the seasonal term (for spring =1, summer=2, monsoon=3, autumn=4,
winter=5, prevernal=6)
N.K.Suryadevara, S. C. Mukhopadhyay, R. Wang, R.K. Rayudu, Forecasting the behavior of an elderly using
wireless sensors data in a smart home, Elsevier: Engineering Applications of Artificial Intelligence, Available online
12 September 2013, ISSN 0952-1976, http://dx.doi.org/10.1016/j.engappai.2013.08.004.

32. Forecasting
τ =Tt = δ (Lt − Lt−1) + (1 −δ) Tt−1
Lt = α (xt − St−s) + (1 − α) (Lt−1 + Tt−1)
St = γ (xt − Lt) + (1 − γ) St−s
(Lt=(1/s) (x1+x2+x3+…..xs)
Tt=(1/s)((xs+1-x1)/s+ (xs+2-x2)/s+…. (x2s-xs)/s)
St= xk -Ls, where k=1, 2….s.)
Ft+m=Lt+Ttm+St-s+m
• Allowable duration of regular activity = Forecasted
duration ± 2 * standard deviation
Trend using activity duration time series
N.K.Suryadevara, S. C. Mukhopadhyay, R. Wang, R.K. Rayudu, Forecasting the behavior of an elderly using
wireless sensors data in a smart home, Elsevier: Engineering Applications of Artificial Intelligence, Available online
12 September 2013, ISSN 0952-1976, http://dx.doi.org/10.1016/j.engappai.2013.08.004.

33. Wellness Functions and forecast of the basic ADL’S
N.K.Suryadevara, S. C. Mukhopadhyay, R. Wang, R.K. Rayudu, Forecasting the behavior of an elderly using
wireless sensors data in a smart home, Elsevier: Engineering Applications of Artificial Intelligence, Available online
12 September 2013, ISSN 0952-1976, http://dx.doi.org/10.1016/j.engappai.2013.08.004.

34. 600.
700.
800.
900.
1000.
1100.
1200.
0 10 20 30 40 50 60 70
Fig 9(a) Toilet usage Trend for70 days Fig 9(b) Toilet usage (Ninth week forecast pattern)
(b)
Fig 9 (c) Toilet usage (Tenth weekforecast pattern) Fig 9(d) Chair Usage Trend for 70 days
600.
700.
800.
900.
1000.
1100.
1200.
0 10 20 30 40 50 60 70
600.
800.
1000.
1200.
1400.
1600.
1800.
2000.
2200.
2400.
0 10 20 30 40 50 60 70
Forecast
Observed
Forecast
Trend
Observed
Observed
Trend
Observed
Days Days
Days Days
N.K.Suryadevara, S. C. Mukhopadhyay, R. Wang, R.K. Rayudu, Forecasting the behavior of an elderly using
wireless sensors data in a smart home, Elsevier: Engineering Applications of Artificial Intelligence, Available online
12 September 2013, ISSN 0952-1976, http://dx.doi.org/10.1016/j.engappai.2013.08.004.

35. Sensor Activity Status
Subject #1
Monitor the Movements of the Person

36. Sensor Activity Pattern
Discovering Interesting Patterns in Data
Perform Classification
of New Data based on
Training Data
Sequential Patterns/Rules
(Finding inherent regularities in data)
With Time Constraints
Cluster of Similar
Instances
Associations
Top-K Sequential Rule Mining
(Redundant/Non-Redundant)

37. Sequence Pattern Analysis* of Data Mining
Sequence Pattern: A pattern (a set of items, subsequences,
substructures, etc.) that occurs frequently in a data set.
*First proposed by Agrawal, Imielinski, and Swami [AIS93] in the context of frequent
item sets and association rule mining.
Ex: Let Sensor_ID={FR,CR,R1,KT}
Let t1=(FR,CR),t2=(FR,CR,R1) and t3=(KT) be three sets of sizes 2,3 and 1.
A eq ence ‘ ’=<t1,t2,t3> = <(FR,CR),(FR,CR,R1),(KT)> e esents a
length |S|=6.
A Sequence S1=<(FR,CR),(FR,CR,R1),(FR,CR,KT),(CR),(CR,CR)>
supports S
A “Sequential Database” Expected Data Base(EDB) is derived from
Data Base(DB) by eliminating certain Sequences

38. • Algorithm:
Sensor Activity Pattern–Pruning (node n= 𝑠1, 𝑠2, 𝑠3, … . . 𝑠𝑙 , Sn)
1.EDB = ∅,Support=0
2. For each (i ∈ Sn)
If ( (𝑠1, 𝑠2, 𝑠3, … . . 𝑠𝑙, 𝑖 ) is frequent )
EDB = EDB ∪ 𝑖
3. For each (i ∈ EDB)
4. Sensor Activity Pattern-Pruning((𝑠1, 𝑠2, 𝑠3, … . . 𝑠𝑙, 𝑖 , EDB, all
e ements in EDB eate than ‘i’ and satisfies Support(EDB) //
Generating EDB at node i that satisfies Support for the SAP
Support (EDB): If SAP is infrequent, and in order for the SAP to
propagate to frequent then it should have a length at-least σ−1(SAP).

39. BK #}BK #}</00>
FR #}BK R3 #}FR #}</02>
R2 BK #}R2 BK #}CR #}</04>
FR KT #}BK KT #}R2 KT #}R2 #}R2 #}R2 #}R2 #}R2 #}R2 #}R2 #}R2 #}R2 #}R2 #}R2 #}R2 #}R2 #}R2
#}R2 #}R2 #}R2 #}R2 #}R2 #}R2 #}R2 #}R2 #}R2 #}R2 #}R2 #}R2 KT #}CR R2 KT #}R2 #}R2 #}R2
#}</05>
R2 #}R2 #}R2 #}R2 #}R2 #}R2 #}FR CR KT #}BK #}R2 BK CR KT #}BK KT #}CR KT #}CR R2 #}#}</06>
BK #}R2 BK #}R2 #}R2 #}R2 #}BK #}R2 #}R2 #}R2 #}R2 #}BK KT #}KT #}CR R2 #}R2 #}#}</09>
BK #}BK #}BK #}CR FR #}FR CR #}CR FR KT #}KT FR #}BK FR KT #}KT #}KT CR FR #}KT #}CR KT #}CR
FR #}R3 #}</10>
R2 #}R2 CR FR BK #}BK FR CR KT #}R2 #}R2 KT #}R2 KT #}KT #}KT #}KT #}KT #}KT #}FR R3 KT #}R2
#}R2 #}R2 #}R2 #}R2 #}R2 #}R2 #}R2 #}R2 #}R2 #}CR KT #}BK #}R3 #}R3 #}</11>
06-Oct-2013(Sunday)
Parser to generate set of Sequences(Sensor ID’s)
based on the (day of the week) and (hour) and (Minute)
enso ID’s eq ence within the Min te “#}” “</” o “>”
Pattern Sequence Detection

40. Tracking System for Inhabitant
Movements inside the house

41. Tuesdays(21May-06Aug)

42. Grouped matrix-based visualization (Hahsler and Chelloboina, 2011)
Tuesdays(21May-06Aug)

43. 43
WSN Assisted Intelligent Integrated Healthcare
Platform for Wellbeing and Independent Living
The healthcare platform consists of
1. Appliances Monitoring Unit
2. Physiological parameters monitoring unit
3. Human Posture and Position Detection Unit
4. Human Emotion Recognition Unit
5. Automatic Medicine Dispenser Unit
6. Power Management Unit
7. Robust Supervisory Control Unit
8. Safety Box Unit
Conclusion

44. • A WSN Assisted and Embedded Processing
based smart home to care elderly people.
• The integrated system is able to support
people who wish to live independently.
• The developed system is robust and is
possible to develop at a low cost due to
ingenious development.
•The technology assisted home will alert the
caregiver in advance about the trend of the
health status, so that necessary precaution can
be taken. 44

45. s.c.mukhopadhyay@massey.ac.nz
n.k.suryadevara@massey.ac.nz