Discovery of Personal Processes from Labeled Sensor Data An - - PowerPoint PPT Presentation

▶
discovery of personal processes from labeled sensor data
SMART_READER_LITE
LIVE PREVIEW

Discovery of Personal Processes from Labeled Sensor Data An - - PowerPoint PPT Presentation

Nov 03, 2023 18 likes •299 views

Discovery of Personal Processes from Labeled Sensor Data An Application of Process Mining to Personalized Health Care Timo Sztyler, Johanna Vlker, Josep Carmona ATAED 2015 22.06.2015 Oliver Meier, Heiner Stuckenschmidt Motivation Health

slide-1
SLIDE 1

An Application of Process Mining to Personalized Health Care

Discovery of Personal Processes from Labeled Sensor Data

Timo Sztyler, Johanna Völker, Josep Carmona Oliver Meier, Heiner Stuckenschmidt 22.06.2015 ATAED 2015

slide-2
SLIDE 2

2

Motivation – Health Care

Noncommunicable diseases (NCDs) kill 38 million people each year. Such diseases include, for example, cardiovascular diseases, diabetes, osteoporosis, and certain types of cancer. (WHO, 2014)

  • Activities of daily living are important for assessing

changes in physical and behavioral profiles

  • In context of medicine, a correct compliance is

important.

  • We want use modern techniques to support people

and improve their healthiness.

slide-3
SLIDE 3

3

Motivation – Self-Tracking

Customary smart-phone platforms are equipped with a rich set of sensors which enable self-tracking.

  • Positioning technologies, sensor networks,

and spatiotemporal data are available

  • Personal behavior and processes can

be derived to learn the daily routine and allows to detect specific patterns.

  • Resulting predictions and recommendations

could help to achieve a healthier life

slide-4
SLIDE 4

4

Motivation - Scenario

Normally, (elder) people get a brief instruction from the doctor how they have to take their pills, e.g., three pills every eight hours without eating one hour after the intake. In general, current self-tracking approaches are helpful in many scenarios However, fine grain monitoring is not possible but necessary. Optimal Daily Routine 12:00 take pills 13:00 eating/drinking 20:00 take pills 21:00 eating/drinking 04:00 take pills 12:00 take pills Actual Daily Routine 12:00 take pills 12:10 eating/drinking 20:12 take pills 21:00 eating/drinking 06:25 take pills 14:55 take pills Time-based monitoring makes sense.

slide-5
SLIDE 5

5

Related Work

Smartphone and Healthcare Processes

  • Activity recognition from accelerometer data
  • n a mobile phone, 2009
  • Review of Healthcare Applications for

Smartphones, 2012

  • Smartphone Based Healthcare Platform

and Challenges, 2015

  • Trajectory pattern mining, 2007
  • Trace clustering in process mining, 2009
  • Process mining: discovery, conformance and

enhancement of business processes, 2011

slide-6
SLIDE 6

6

Overview

  • 1. Motivation
  • 2. Related Work
  • 3. Personalized Health Care
  • Self-Tracking
  • Use Cases and Experiments
  • 4. Challenges
  • 5. Summary
slide-7
SLIDE 7

7

Personalized Health Care – Self-Montoring

Location Inertial Sensors Self-Tracking

Activity Recognition (e.g. running) Mapping Database Compass Behavior / Daily Routine

slide-8
SLIDE 8

8

Personalized Health Care – Example

Activity Recognition Activity Recognition is a learning problem but there are still many

  • pen issues …

Accelormeter, X-axis readings for different activities (Ravi et. al., 2005)

?

slide-9
SLIDE 9

9

Personalized Health Care – Data Gathering

  • ~12 hours/day, 2 weeks, 8 subjects
  • recording inertial sensors and location
  • subjects have to label their activities (e.g., “playing football”)
  • it was possible to combine activities (e.g., “desk work” and “drinking coffee”)

Labels Records (avg±sd) Activities 20±7 Postures 80±62 Location 16±4

  • Dev. Position

8±6

  • Recorded: 74 cases, 1386 events
  • Average duration of one day: 12.1 hours
slide-10
SLIDE 10

10

Personalized Health Care as Process Mining

software system (process) model event logs

models analyzes

discovery

records events, e.g., messages, transactions, etc. specifies configures implements analyzes supports/ controls

enhancement conformance

“world”

people machines

  • rganizations

components business processes

Activity Recognition Software Daily Activities

slide-11
SLIDE 11

11

Overview

  • 1. Motivation
  • 2. Related Work
  • 3. Personalized Health Care
  • Self-Tracking
  • Use Cases and Experiments
  • 4. Challenges
  • 5. Summary
slide-12
SLIDE 12

12

Personalized Health Care – Use Cases Overview

I. Monitoring Record and analyze the personal behavior

  • II. Deviations

Compare personal processes with reference processes to detect deviations.

  • III. Operational Support

Combining spatio-temporal data and activity data to make predictions. Optimize the daily routine by adding missing activities or reorder them. Visualize their personal processes to highlight unconscious behavior. Make recommendations in order to accomplish certain goals.

slide-13
SLIDE 13

13

Personalized Health Care – Use Case I

  • Variability: for each individual,

#process variants = #traces !!

  • Fuzzy Models (using Disco)

allowed to focus on the main activity

  • Confirm tendencies:
  • Working vs. weekend days
  • Student vs. not Student

Personal Grooming Movement Meal Preparation Eating/ Drinking Housework Shopping DeskWork Relaxing Transporta tion Sleeping Socializing Sport

Personal activity during working week days

Personal Grooming Movement Meal Preparation Housework Shopping Relaxing Transporta tion Sleeping Sport

Personal activity during weekend days

Socializing DeskWork Eating/ Drinking

slide-14
SLIDE 14

14

Personalized Health Care – Use Case I

Model Enhancement Using Personal Data

  • personal activity-position map
  • space, time, and activity

(trajectory pattern)

  • New possibilities:
  • Geographical Label Splitting
  • Geographical Abstraction and

Clustering

Workplace Home Free Time

slide-15
SLIDE 15

15

Personalized Health Care – Use Case II

Reference Models

  • They can be obtained by
  • An expert (e.g., a doctor)
  • Using elite data
  • Elicitating them from textual

information using NLP+Process Extraction (Friedrich et al.)

  • Starting point to
  • Check deviations
  • Forensics
  • ...

DeskWork Relaxing Housework Sport Personal Grooming Eating/ Drinking Movement Meal Preparation Socializing Shopping Sleeping

Main Personal Activity

slide-16
SLIDE 16

16

Personalized Health Care – Use Case II

Reference Models

  • specific order, explicit choices,

concurrency actions

  • (flexible) conformance checking
  • deviations, costs, and quantities

Personal Grooming

Silent Event

could be expensive! Simplification: rules or patterns which should be satisfied by an individual.

slide-17
SLIDE 17

17

Personalized Health Care – Use Case III

State-based prediction 0.6 0.5 0.8

Sport Sleep Eating/Drinking

  • probability to reach a particular goal
  • process models help to determine the

influence of the next step

  • aggregate historical data/activities with,

e.g., amount of calories.

  • amount of calories vs. labels

Probability of a balanced day (calories consumption

  • vs. burning)
slide-18
SLIDE 18

18

Personalized Health Care – Use Case III

State-based prediction

  • Important question: Does concurrency plays an important role ?
  • Yes: then event-based models may be used for operational

support

  • No: state-based models like the one before are sufficient
  • Potential concurrency pairs in our context:
  • Movement/Transportation
  • Transportation/Socializing
  • Deskwork/Socializing
  • ... but in practice they were not so common !
slide-19
SLIDE 19

19

Overview

  • 1. Motivation
  • 2. Related Work
  • 3. Personalized Health Care
  • 4. Challenges
  • 5. Summary
slide-20
SLIDE 20

20

Challenges

  • 1. Trace Alignment
  • The behavior of a person is very individual any may depend on

the day (working day vs weekend) and other factors.

  • 2. Uncertainty
  • The daily routine of a person is flexible and does not follow a fix
  • rder of activities.
  • 3. Analytics
  • Several different dimensions such as space, time, and activity

has to be considered in context of the daily routine.

slide-21
SLIDE 21

21

Challenges (1) - Trace Alignment & Clustering [5]

  • Aims to extract common and frequent behavior but also highlight

exceptional behavior.

  • Cluster Log: (multi)set of cluster traces that may be the starting point for

analysis (discovery, conformance, ...)

  • How many clusters ?

Trace Alignment (ProM) Cluster Trace

slide-22
SLIDE 22

22

Challenges (2) - Uncertainty

Washing Working Stand Up Breakfast Trace:

, , , < >

Washing Social Jogging Breakfast Probabilistic Trace:

, , , < >

Stand Up Shaving Phone Working Hiking

0.6 0.2 0.2 0.2 0.8 0.7 0.1 0.2

A new theory for probabilistic process mining is needed

Models ? Algorithms ? Metrics ?

slide-23
SLIDE 23

23

Challenges (3) - Analytics

  • Process Cubes as a solution to handle, i.e., Spatial, Time,

Activity, and Transportation Modes.

  • Find tailored behaviors (e.g., reference models)

according to particular goals

  • May open the door to gamification (e.g., try to match

a very particular behavior)

Activities Time

Process Cubes

slide-24
SLIDE 24

24

Summary

However, we just started … … and these are the things we are working on. We hope for ideas for future work. Personal Healthcare is important and we want to support people automatically and we believe this is a very promising field for process mining. We outlined our ideas and challenges to support the following use cases:

  • Monitoring
  • Deviations
  • Operational Support
slide-25
SLIDE 25

25

Thank you

Thank you for your attention

slide-26
SLIDE 26

26

References

1. WORLD HEALTH ORGANIZATION, et al. Global status report on alcohol and health-2014. World Health Organization, 2014. 2. RAVI, Nishkam, et al. Activity recognition from accelerometer data. In: AAAI. 2005. S. 1541-1546. 3. GIANNOTTI, Fosca, et al. Trajectory pattern mining. In: Proceedings of the 13th ACM SIGKDD international conference on Knowledge discovery and data mining. ACM, 2007. S. 330-339. 4. ZHENG, Yu. Trajectory data mining: an overview. ACM Transactions on Intelligent Systems and Technology (TIST), 2015, 6. Jg., Nr. 3, S. 29. 5. BOSE, RP Jagadeesh Chandra; VAN DER AALST, Wil MP. Process diagnostics using trace alignment:

  • pportunities, issues, and challenges. Information Systems, 2012, 37. Jg., Nr. 2, S. 117-141.

6. DE LEONI, Massimiliano; VAN DER AALST, Wil MP. Aligning event logs and process models for multi- perspective conformance checking: An approach based on integer linear programming. In: Business Process Management. Springer Berlin Heidelberg, 2013. S. 113-129. 7. GÜNTHER, Christian W.; VAN DER AALST, Wil MP. Fuzzy mining–adaptive process simplification based

  • n multi-perspective metrics. In: Business Process Management. Springer Berlin Heidelberg, 2007. S. 328-

343. 8. VAN DER AALST, Wil MP. Process cubes: Slicing, dicing, rolling up and drilling down event data for process mining. In: Asia Pacific Business Process Management. Springer International Publishing, 2013. S. 1-22. 9. VAN DER AALST, Wil MP; DE BEER, H. T.; VAN DONGEN, Boudewijn F. Process mining and verification of properties: An approach based on temporal logic. Springer Berlin Heidelberg, 2005.

slide-27
SLIDE 27

27

Backup - Challenges (1) - Variability

Process Variant 1 Process Variant 2 Process Variant 3

slide-28
SLIDE 28

28

Backup - Challenges (3) – Analytics - Example

Daily activity in the period 2013-2014 for females between 20-30 years old, non-smokers Daily activity in the period 1980-1981 for athletes between 20-30 years old, non-smokers

Relaxing Sport Personal Grooming Eating/ Drinking Movement Meal Preparation Socializing Sleeping Shopping Housework Relaxing Sport Personal Grooming Eating/ Drinking Meal Preparation Socializing Sleeping Shopping Housework