Isc2 vitali

Interconnecting Processes through IoT in a Health-Care Scenario

Monica Vitali, Barbara Pernici

Dip. Elettronica Informazione e Bioingegneria (DEIB)Politecnico di Milano, Italy

CONTEXT

Internet of Things has a great potential for improving many aspects of life.

The value of data coming from IoT is null if they are not analyzed properly

MOTIVATION

Cooperating organizations act separately, but data used by the processes are not independent.

ISSUEHidden relations between business processes of different organizations affect the global result.

GOALTo Identify and discover interconnections between business processes in an IoT environment enriched with events.

PiE – Processes in Events

PiE - ScenarioOrganization 1

Event Pool

Organization 2

PiE - ScenarioOrganization 1

Event Pool

Organization 2

PiE - ScenarioOrganization 1

Event Pool

Organization 2

How processes of different organizations interact?

How the environment affects the processes?

Connecting events

APPROACH: merge the unstructured information coming from IoT with the structured information coming from the monitoring of business processes.

Hybrid correlations – to find relations between processes through device events.

Several possible relationsTemporal-based relationsContent-based relations

Connecting events: STEP 0

Connecting events: STEP 1

Connecting events: STEP 2

Connecting events

We considered temporal/causal relations.

Two event types are related if A always follows B

Confidence how many time A is observed if B is observedweighted by the time distance between A and Bcontinuously updated accounting for history

We used synthetic logs generated integrating real observations with realistic ones.

PiE Application to AALSocial Care

Health Care

Operator Recipient

Event Pool

WATCH

TOTEM

WEB INTERFACE

PiE Application to AAL

PiE Application to AAL

PiE Application to AAL

eD2<T1>

: operator in

eD2<T2>

: operator out

eD1<W1>

: watch std

eD3<I2>

: urgent drug ordereD1<W2>

: watch unfasten

eD3<I1>

: regular drug order

PiE Application to AAL

eP1<A1>

eP1<A2> eP1

<A3> eP1<A4>

eP1<A5>

eP1<A6>

eP2<A1>

eP2<A2>

eP2<A3>

eD2<T1>

eD1<W2> eD2

<T2>

eD1<W1>

eD3<I1>

eD3<I2>

A1 A2 A3 A4 A6

A5

B1 B2 B3

PiE Application to AAL

eP1<A1>

eP1<A2> eP1

<A3> eP1<A4>

eP1<A5>

eP1<A6>

eP2<A1>

eP2<A2>

eP2<A3>

eD2<T1>

eD1<W2> eD2

<T2>

eD1<W1>

eD3<I1>

eD3<I2>

A1 A2 A3 A4 A6

A5

B1 B2 B3

0.3 0.990.450.94

0.81

0.99 0.96 0.990.99

0.990.29

PiE Application to AAL

eP1<A1>

eP1<A2> eP1

<A3> eP1<A4>

eP1<A5>

eP1<A6>

eP2<A1>

eP2<A2>

eP2<A3>

eD2<T1>

eD1<W2> eD2

<T2>

eD1<W1>

eD3<I1>

eD3<I2>

A1 A2 A3 A4 A6

A5

B1 B2 B3

0.3 0.990.450.94

0.81

0.99 0.96 0.990.99

0.990.29

PiE Application to AAL

eP1<A1>

eP1<A2> eP1

<A3> eP1<A4>

eP1<A5>

eP1<A6>

eP2<A1>

eP2<A2>

eP2<A3>

eD2<T1>

eD1<W2> eD2

<T2>

eD1<W1>

eD3<I1>

eD3<I2>

A1 A2 A3 A4 A6

A5

B1 B2 B3

0.3 0.990.450.94

0.81

0.99 0.96 0.990.99

0.990.29

PiE Application to AAL

eP1<A1>

eP1<A2> eP1

<A3> eP1<A4>

eP1<A5>

eP1<A6>

eP2<A1>

eP2<A2>

eP2<A3>

eD2<T1>

eD1<W2> eD2

<T2>

eD1<W1>

eD3<I1>

eD3<I2>

A1 A2 A3 A4 A6

A5

B1 B2 B3

0.3 0.990.450.94

0.81

0.99 0.96 0.990.99

0.990.29

Future Steps

Discovered relations can be exploited for mining information about processes execution.

We plan to:validate the PiE methodology in a complex

real case studyapply PiE to other fields involving interacting

processes and IoT devicesenrich the relation discovering using content

based analysis

Interconnecting Processes through IoT in a Health-Care Scenario

Monica Vitali, Barbara Pernici

Dip. Elettronica Informazione e Bioingegneria (DEIB)Politecnico di Milano, Italy