This deck illustrates the architecture of the solution, WSO2 as the main actor, the benefits for the patient (house follow-up, prediction of clinical condition, etc.), and the benefits for the organizations and users

1. Solution Architect
Engineering Ingegneria Informatica
Marco Mastroianni
Medical Device Integration
The Future is Here

2. What is MDI?
Medical Device Integration is an
Engineering Ingegneria Informatica
solution created to make software
communication secure, efficient,
robust and intelligent with devices to
support patient monitoring both in
hospital and for all patients who have
to follow up in their homes.

3. Abstract to application
solutions the complexity of
communication with
individual medical devices
providing them with a
standard, secure and reliable
communication protocol based
on multiple channels.
MDI: Challenging Goals

4. WSO2 Stream Processor and
WSO2 ML are able to offer
intelligence to the application
solutions by processing in real
time thousands of events and
thus providing alarms or
trends compared to the state
of the patients and the
devices
MDI: Challenging Goals

5. WSO2 Dashboard Server permits you to have a synoptic
monitoring dashboard that allows you to have immediately the
operating status of the devices and the related technical
parameters that devices transmit (reagents, State of health,
Battery, etc.) under control
MDI: Challenging Goals

6. Architecture
Logical decomposition of system

7. Logical Architecture

8. WSO2 Components
inside
Architecture
WSO2 Iot
WSO2
Stream
Processor
WSO2
Identity
Server
WSO2
Enteprise
Integrator/MB
WSO2 API
Manager

9. WSO2
Identity Server
• Is the main responsible for
communicating to make the
communication of the whole
components of architecture secure
using the most secure protocols
like OAuth2 with JWT Token
• Is used also for generating tokens
to access the API Manager by
integrated applications (Service
providers)

10. WSO2 Enteprise
Integrator/ WSO2
Message Broker
• Is the main responsible for communicating
with external modules to the Platform
solution
• Provides an integrated multi-protocol
Message Broker that allows communication
between different devices and the centralized
MDI system.
• Also allows you to generate alert or
management notifications from MDI to
application solutions.
• Is able to communicate with the Terminology
Services to perform compensation and
transformation of incoming/outgoing
streams
• Is able to provide standard communication
API (proxy services) like HL7

11. WSO2 IoT
1 of 2
• Allows to develop device drivers
with installation packages, which
are easy to deploy and reuse.
• Each device driver has «health
module» that sends technical
information, like heart-beat, status
of hardware and component
• Is able to have Map of location of
device (when Geolocalization is
supported by hardware)

12. WSO2 IoT
2 of 2
• Each «driver» is able to transform
specific device protocol (RS232, HL7,
ecc..) into an encrypted generic platform
message in order to avoid MDI to know
each protocol.
• Stream processor can configure
algorithms which allow you to run
"Predictive maintenance ". It often
happens that the hardware shows a
series of symptoms that lead to a
technical problem without the system
itself sending a specific alarm.

13. WSO2 Stream
Processor
• Analyses in realtime several
health/clinical messages coming
from device driver: technical and
clinical information has splitted into
different CEP flow; so it is possibile
to manage technical warning or
feed CEP/ML flow about clinical
data
• ML acquires and refines its
classified algorithms in order to
predict any critical situations.

14. Machine Learning/CEP Architecture

15. The benefit for Patient

16. The benefits for the patient
Through this type of architecture, the patient obtains different
benefits, both in the hospital and in the follow-up phases carried
out in other places (e.g. own home).
• In the hospital context everything revolves around the communication
and management of the instruments, from the management of the
state of the devices, to the programming and reception of the clinical
data of the patients.
• This way the probability of errors is lowered considerably, because the
data flows in the folder automatically without any human transcription.

17. Machine Learning
Complex Event
Processing
inside
INTENSIVE CARE
UNIT
Case Study

18. • Patients to monitor H24
• Heterogeneous and isolated devices
• Vital information to be integrated with data from other
solutions
• Communication speed
• Timeliness of decision making
Intensive Care Unit - Requirements

19. • Critical care patients are often technologically dependent
on monitoring or life-sustaining devices
• The assessment of prognosis and patient stratification
making use of these combined data sources is extremely
important, which highlights the importance of medical device
connectivity in this patient-centric environment as an
essential element of a complete clinical decision support
framework
Patient Monitoring and Alarm Algorithms in
Critical Care

20. • The data made available by these devices often take the
form of signal, making it the perfect target for signal
processing techniques based on ML and related methods
• One of the key ICU problems addressed to this approach is
patient monitoring and to the related design of algorithms for
the implementation of patient alarms.
Intensive Care Unit - Requirements

21. A Real Sample
Severe hemodynamic
instability or shock
• MDI solution is able to notify
immediately Hemodynamic
Instability or Shock analyzing in real
time Observations (event) coming
from ICU monitoring device (ex.
Philips Intellivue device)
• Alert can be based on CEP rule and
Classified Machine Learning
Algorithms

22. A Real Sample
Severe hemodynamic
instability or shock
Scenery:
• Several patients under monitoring in
ICU room
• Device connected to a specific patient
(identified with PID.1.1234 code) send
observations
• The trends of these observations
become dramatically critically
• MDI can intercept the trends and
notify to specific application the
alarm

23. Hemodynamic Shock – CEP Analysys
PID SpO2 SisArt BPM Timestamp ……
PID.1.1234 96 130 80 12/10/2018 10:00:00
PID.1.1235 95 120 67 12/10/2018 10:00:02
PID.1.1234 90 110 88 12/10/2018 10:01:00
PID.1.1235 95 120 69 12/10/2018 10:01:01
……… ………….. …………. ……….. ……….
PID.1.1234 80 75 110 12/10/2018 10:05:00

24. A Real Sample
Severe hemodynamic
instability or shock

25. Hemodynamic Shock – CEP Analysys

26. THANK YOU
wso2.com