Transcript Slides - Georgia State University

Health Informatics:
Applications, Requirements,
and Emerging Research
Upkar Varshney
Department of CIS
Georgia State University
E-mail: [email protected]
1
Format of the Tutorial




Health Informatics (HI)
Information and Communications
Technologies in Healthcare
Specific Examples of Health Informatics
Conclusions and Future
2
Health Informatics (HI)




The intersection of several fields including computer
science, healthcare, and business
To provide the needed information anywhere
anytime to anyone authorized in prompt, correct
and secure ways
Resources, devices and methods to provide
healthcare services
Consumer health informatics, nursing health
informatics, organizational health informatics, public
health informatics, and medical health informatics
3
E-health and Health Informatics
E-health
Health
Info Syst
EMR
Health
and
EHR Informatics
M-health
Telemedicine
Devices
4
Major Goals of HI









Reducing Various Errors in Healthcare
Extending the Coverage and Delivery of Healthcare
Improving Medication Adherence/Adverse Drug
Events
Support Independent Living for the Elderly
Improved Decision Making
Wellness and Managing Chronic Conditions
Improving Efficiencies and Reducing Overall Cost
Supporting Patient Empowerment
5
Addressing Limited Human Resources
Chronic Diseases





A disease for which there is no cure
Managing it well to reduce other complications
CDC: Chronic diseases – such as heart disease,
stroke, cancer, diabetes, and arthritis – are among
the most common, costly, and preventable of all
health problems in the U.S.
Heart disease, cancer and stroke: 50% of all deaths
The four most common reasons: lack of physical
activity, poor nutrition, tobacco use, and excessive
alcohol consumption
6
Wellness and Proactive Health

People with good health need to maintain it


People with chronic conditions need to manage it


Medications, sleep, weight
Elderly want to live independently


Exercise, stress, food, sleep, weight
Activities of daily living, medications, sleep, weight,
behaviour
Use of ICT to enable monitoring and management
of health
7
IT-based Wellness Management

Various IT/communications technologies for







sensors in shoes
Internet-aware exercise machines
cell-phone based applications for managing
wellness
Wellness diary
Social networking/group communications/twitter
Suitability/evaluation of technologies
Theoretical Support (or lack of suitable
theories)
8
Information and Communications
Technologies in Healthcare







Smart computing
Wearable computing
Sensors
RFID
Wireless LANs
3G/4G networks
Personal area networks
9
The Big Picture
Quality of
integrated data
Quality of
Processed data
Health
Databases
Quality of
stored data
Quality of data
from sensors
Quality of
retrieved data
Quality of service
Quality of
transmitted data
Devices
Compressed and
Processed data
Quality of life
for the patient
Data from
multiple sources
(patients, doctors,
labs, pharmacies)
Networking
Infrastructure
Quality of network
control data
Quality of
received data
Healthcare
Decision
Systems
Quality of
healthcare
decisions
Quality of healthcare services
10
Improvements with Wireless
Technologies



How wireless technologies can help improve
healthcare systems worldwide
Wireless technologies can lead to the desired
evolution of healthcare system
In general, these technologies




can allow information to be available anywhere any time
to anyone who is authorized to access it
make the delivery of healthcare services more efficient
reduce the number of tasks that need to be done by
healthcare professionals
encourage patients to take better control of their
healthcare needs and life style
11
Healthcare Quality of Service (H-QoS)
and Wireless Requirements

Real-time delivery


Reliability


Wireless LANs and 4G networks
Location Management


Cellular/3G/4G, wireless LANs, satellites, ad hoc networks
Bandwidth


Challenges for most wireless networks
Wide Coverage


Cellular/3G/4G
Cellular/3G/4G and wireless LANs
Pricing

Wireless LANs
12
Suitable Technologies for Healthcare





Suitable technologies for healthcare: implanted (inside
body), wearable, portable, and environmental
technologies
Implanted technologies: RFID to store information and
sensors to measure medical parameters
The wearable technologies: Smart Shirts with sensors
designed to wear for extended monitoring of health
conditions. Could be washed, ironed and charged for
use, and in future, networked with devices and people
The portable devices, such as handheld devices and
phones, used in monitoring and recording health
conditions
Environmental technologies: computing and
communications close to the patients (“Smart” Home) 13
Wireless Technologies in Healthcare







Cellular Networks
Wireless LANs
Sensors
Radio Frequency Identification
Bluetooth and ZigBee
Satellites
Characteristics: Indoor vs outdoor, real time vs no
real-time services, coverage (PANs to WANs),
reliability, varying bit rates and levels of locationawareness
14
Sensors & Applications



Small devices with sensing, computation and
wireless communications capabilities (not mobile!!)
Sensors measure ambient conditions in their
surrounding environment and then transform these
into signals (which can be processed to determine
the conditions of the “sensed” environment)
Applications



Use in wearable, portable and environmental
implementations (Smart Shirt, Smart House, Appliances)
Monitoring of vital signs
Many more applications with sensor networks (fall
detection with embedded “smart carpet”)
15
Mobile monitoring devices




Implanted, portable, wearable or in the surrounding
environment
Devices with sensors to measure a range of vital signs and
other parameters for its patient
The devices with intelligence would detect certain
conditions by the touch of a user
Many of the smaller medical devices can be integrated in
the hand-held/wearable wireless device


Pulse-rate, blood pressure, level of alcohol
Specific requirements of vital signs


how to measure and process vital signs such as blood
pressure (BP), ElectroCardioGram (ECG), temperature,
oxygen saturation
Each of these requires different type of sensor(s) at a certain
16
part of human body
Sensors in Healthcare
+
ECG
Sensors
Temperature
sensors
Sensors on Neck, Bed, Kitchen,
Appliances, and Bathroom
Sensor locations and user movement
Blood pressure Reliability of sensors
sensors
Unintentional removal of sensors
Wear and tear of sensors
SpO2 sensor Sensor-body contact/noise
Connectivity for body area networks
False positive/false negative
17
Vital Signs
Sampling rate
Quantization
Breathing (12-18/min) 1 sample/sec
4 bits/sample
Minimum bit rate
4 bps
ECG Signal (60-80/min)
240 samples/sec
12-36 bits/sample
Blood pressure (Sys<120, Dia<80)
D
1 beat
2.9-8.7 Kbps
1 sample/minute
1 bps
64 bits/sample
Oxygen Saturation (95-99%) 1 sample/sec
16 bps
16 bits/sample
Multiple
messages
per minute
Body core temperature (97.1-99.1F)
1 sample/min
0.3 bps
16 bits/sample
18
Radio Frequency Identification (RFID)

Location tracking of






Patients and healthcare professionals
Supplies, equipments, and blood
Authentication of expensive medications
Storage of (compact) information (such as EMR)
(Ingestible) RFID on “smart” capsules to monitor
the condition of internal organs (GI tract diseases)
Further Research:


Requirements of new applications/bio compatibility/side
effects/long-term use
Cost-benefit of RFID deployment
19
Smart House


Assistive environments (for older and/or disabled
people) for sensing themselves and their residents
Gator Tech Smart House at University of Florida








smart blinds to control ambient light (and privacy)
smart bed to monitor sleep patterns
smart closet to make clothing suggestions
smart mirrors for messages & reminders for medications
smart bathroom with sensors for measurement of
weight, height and temperature, and ECG
SmartWave to refuse to heating up the items that you
are not suppose to eat
social-distant dining using immersive video
smart floor for fall detection
20
Smart House
Smart Medication System
Smart
Laundry
(Future)
Smart Blinds
Smart Floor
Smart
Projector
Home
Security
Monitor
Driving
Simulator
Smart mail-box
Smart Bed
Smart
Plugs
Smart
Closet
(Future)
Smart Mirror
& Bathroom
Smart Front
Door
Social Interaction/Entertainment System
E
N
E
R
G
Y
A
W
A
R
E
Fall Prevention and Detection Systems
S
Y
S
T
E
M
Daily Activity Support System (context-aware)
A
L
E
R
T
&
M
O
N
I
T
O
R
I
N
G
Memory Support System
21
R
E
L
I
A
B
L
E
S
Y
S
T
E
M
Cellular/3G/4G Networks






Offers from about 100 Kbps to 2 Mbps
Designed to support multimedia, data, and video
Short Messaging Service
 stored & delivered in few seconds (not real-time)
 reminders, or compressed information on patient
General Packet Radio Service (GPRS): 160 Kbps
Enhanced Data rate for GSM Evolution (EDGE):384
Kbps
4G (not available in places where patients may live) but
useful for video-oriented healthcare applications


Telemedicine, tele-radiology, tele-surgery
Video-clips of patients, healthcare professionals
22
Wireless LANs: IEEE
802.11a, 802.11g,
802.11n

Cell
Supports 54 Mbps in 5 GHz band (smallest range)
Supports 54 Mbps in 2.4 GHz band
802.11n



Cell
802.11g


User
802.11a


BS/Terminals
As high as 600 Mbps and up to 70 meters (indoors)
and 250 meters (outdoors)
5 GHz and 2.4 GHz both possible
Useful for independent homes, assisted living
and nursing homes
23
Bluetooth and ZigBee In Healthcare






Bluetooth: Unlikely to be a standalone technology due
to short range (10m), limited bit rate (few hundred
Kbps) and not more than eight devices in a piconet
The range could be increased by adding Bluetooth
adapters in hand-held devices and phones
ZigBee: ad hoc and mesh networking format
One of the intended environments is Hospital Care
More likely to be a front-end technology and will
require another network to carry monitoring messages
to one or more healthcare professionals
Sensors: Bluetooth or ZigBee for communications to
other devices or among themselves (sensor networks)
24
Comparison of Wireless Technologies
Wireless
Technologies
Personal Area
Networks
(Bluetooth)
Coverage
Bit rate
Cost
Small (about
10meters)
Few
hundred
Kbps
Low with
inexpensive
adapters
Number of
patients
A few (8 or less
active per
piconet)
Wireless
LANs (IEEE
802.11)
Small (about
100 meters)
Several
Mbps
Low
10-100
Cellular/3G
Wide area
(nationwide)
High
subscription
/usage
charges
1000s
Satellites
Wide area
Few
Kbps to
several
hundred
Kbps
Few
Kbps to
several
Mbps
High
subscription
/usage
charges
100s (limited
uplink
bandwidth)
Fixed
wireless
Metropolitan
Up to
several
Mbps
High
subscription
/usage
charges
100s (limited to
the same general
area)
Suitability
Could work well
in conjunction
with other
wireless networks
Suitable for
homes, assisted
living, and
nursing homes
With commercial
traffic, sometimes
bandwidth not
available
Primarily outdoor
or line-of-sight
and may work
well as the
secondary
network
Patient mobility is
difficult to
support
25
Wearable Computing: Variations
User programmable
Privacy support
Monitoring of vital signs
Smart
Shirt
Networkable
Chargeable
Washable
26
Context-awareness
Thresholds
Vital Signs Rate of Change
Previous Values
Prescribed
Medicines
Medicine1, ..N
Recent doses
Missed doses
Patient
Handicaps
Physical
Cognitive
Sensory
Sweat
Sensory
Palpitations
Information Breathing
Current
Activities
Sitting
Walking
Running
Sleeping
Environmental Temperature
Humidity
Variables
Air quality
Unusual Conditions
Recent Lab-results
Type of
Health
Monitoring
Filtering
&
Information
Integration
Patient’s
Medical
History
Context
Generation
and
Processing
Patient Info
And Context
27
Obtain Vital_signs and thresholds
No
If Vital_sign<Thres2(low)
If Vital_sign<Thres1(low)
Yes
EM-Points = EM-Points + Thres1(low) - Vital_sign)
Yes
Yes
Emergency Level = High
Transmit Emergency Signal
No
If Vital_sign<Thres1(high)
No
If Vital_sign<Thres2(high)
No
Yes
EM-Points = EM-Points + Thres2(high) - Vital_sign
No
If RT_CHNGE>RT_THRSLD
Yes
EM-Points = EM-Points + POINTS_RT_CHNGE
If CURR_ACTIVITY = RESTING
No
Yes
EM-Points = EM-Points + POINTS_RESTING
If Curr_TM>= RPRT_TM
Yes
Transmit Normal Signal
If EM-Points >= EM_Thresh-H
No
No
Yes
If EM-Points >= EM_Thresh-M
Yes
Emergency Level = Medium
Transmit Abnormal Signal
28
The Monitoring of Monitoring System
Monitoring System
Monitoring of
Alert Generation
Monitoring of
Alert Transmission
Monitoring of
Alert Processing
Patient
Information
Wireless Networks
Monitoring
device
Monitoring of
Decisions/updates
Healthcare
Professional
29
Specific Components





EMR/EHR
Telemedicine and Health monitoring
Medication monitoring and management
Independent Living and Activity of Daily Life
Medical and clinical decision making
30
EMR/EHR
Source A
Source B
EMR
Source C
Patient Information
Hospital
2
1
Orders for lab tests
6
Orders for medications
9
Pharmacy
Updated refills
Physician’s notes
8
Test results
3
5
Physician
7
4
Laboratory
31
More Research in EMR/EHR

Patient information from multiple sources



Use of handheld devices to access EMR



Displaying important information first/Cognitive load
Viewing part of EMR on small screen vs all on big screen
Reliability of Mobile Infrastructure


Verification by patients/healthcare professionals
Tagging (source, time)
Coverage, access and delay
Access to EMR in Emergency

Use of stored information on patient’s body (RFID/shirt)
32
Health Monitoring: Vital Signs and
Processing
Vital Signs
(Nominal Values)
Sampling rate
Quantization
Breathing (12-18/min)
Fb samples/sec
Bb bits/sample
Fb x Bb
ECG Signal (60-80/min)
1 beat
FECG x BECG
FECG samples/sec
BECG bits/sample
Blood pressure (Sys<120, Dia<80)
FBP samples/sec
BBP bits/sample
Oxygen Saturation (95-99%) F
SPO2 samples/sec
FBP x BBP
Messages
FSPO2 x BSPO2
BSPO2 bits/sample
Body core temperature (97-99 F) FT samples/sec
FT
x BT
BT bits/sample
33
5.4 Processing of Vital Signs and Parameters.
Vital Sign & Parameters Sampling rate
Quantization
Total bit rate
Breathing rate
1 sample/sec
4 bits/sample
4 bps
ECG
Blood Pressure
240 samples/sec
1 sample/minute
12-36 bits/sample
64 bits/sample
2.9 to 8.7 Kbps
1 bps
16 bits/sample
16 bits/sample
16 bps
0.3 bps
Oxygen Saturation
1 sample/sec
Core body temperature 1 sample/minute
Transmitting video can add considerable traffic depending on the duration,
frequency, resolution, frame rate, and compression
34
Types of Monitoring
Monitoring
Behavior
Medications
Adherence/abuse
Dementia
Delirium
Disability
Autism
Sleep
Hypertension
Diabetes
Rheumatoid Arthritis
Digestive
Respiratory
Cardiovascular
Neurological
Renal
Eating
Pattern
Sleep Apnea
REM
Sleep Behavior
Sleep Walking
Sleep Talking
Binging
Purging
Anorexia
Bulimia
Obesity
Diabetic
Renal
Cardiac
General Health
35
Threshold-based (multiple vital signs)
Vital-sign(W)
Vital-sign(X)
Threshold2(High)
Vital-sign(Z)
Emergency Level=High
Emergency Level=Medium
Threshold1(High)
Emergency Level=Low
Threshold1(Low)
Alerts/Alarm
Abnormally high (check
for related problems)
Normal (but still check
other vital signs)
Emergency Level=Medium
Abnormally low (check
for related problems)
Emergency Level=High
Alerts/Alarm
Threshold2(Low)
Vital-sign(Y)
Inter-relationship among
Vital-signs
36
Cellular/3G/4G for Monitoring

Advantages








Real-time Delivery
Wide Coverage
Bandwidth for WHM
Ability to Work with other Wireless Technologies
Widely Used Technology
Secure
Location Management
Limitations




Availability and presence of dead-spots
Reliability Challenges
Lack of Broadcast/Multicast
Pricing and the impact of commercial traffic
37
Experience in using Cellular/3G/4G






The cost was a major factor for many users and
even some hospitals
The quality was variable (packet loss, delays,
disconnections)
Sometimes healthcare professionals were not
reachable (coverage, network overload problems)
Patients were not always able to access the
network (access and coverage problems)
Sometimes the device battery was a limitation
Video quality was variable (bandwidth issues)
38
Wireless LANs for Monitoring

Advantages:





Bit Rates
Transmission from Patients to AP (access point)
Support for Mobile Patients
Location Management
Limitations






Limited Coverage
Security
Monitoring Delays
Co-located Networks
Reliability
Multicast
39
Experience in Wireless LANs





The coverage was unpredictable
The data speed was variable (monitoring delays
were highly variable)
 Shared bandwidth
 Interference in shared unlicensed ISM band
The device could not access the network
Reaching to HP was difficult
Sometimes video quality was not good (variable
delays)
40
Lack of Medication Adherence





With prescriptions (2010) at 3.5 billion/year,
prescription medications a major component of
healthcare expenses
From no-use (about one third), infrequent use (about
one third), to overuse to abuse (about one third)
The non-adherence leads to 125,000 deaths and $90
billion in additional hospitalization and procedures
People who miss their doses are three times more
likely to see doctors again, resulting in further increase
in healthcare expenses
About 20% people in US have used prescription drugs
for non-medical reasons (prescription drug abuse)
41
Smart Medication Management System




Medication is only dispensed certain doses at
certain number of times a day to certain people
SMMS to keep track of the time and the number of
times/day a certain medication was taken (also
how many times the medication system was
attempted to be opened unsuccessfully)
Physicians can check/communicate with SMMS on
medication adherence and/or abuse before
renewing the prescriptions
SMMS can prepare and transmit short video clips of
various actions of the patient
42
Satellites
Device-Networks
Interaction
Notification/Alerts
To physicians
Wireless LANs
Wireless LANs
To pharmacists
To designated
family members
Cellular Networks
Reminders/alarms
To insurance
companies
Healthcare ProfessionalsPatient Interaction
Program/controls
Adherence monitoring Medication
Management
System
Abuse monitoring
Inter-device
Interaction
Wireless
Personal
Area
Network
MMS-Patient
Interaction
43
Context-aware Reminders
Medication Time
Medication Time
Medication Time
Med ABC (1T)
Med ABC (1T)
Med ABC (1T)
Med XYZ (2T)
Med XYZ (2T)
Med XYZ (2T)
Reminder 1
Reminder 2
Reminder 3
44
Using SMMS in Multiple Interventions
Dispensing of Dose
Sensing of Patient’s actions
Report
(adherence and
patterns of use)
Advice
(Scheduling)
Context-aware reminder(s)
Smart
Medication Adherence/side-effects
Management
Dosing-changes
System
Social Support/Motivation
Support from
Healthcare
Professional
Support from
family and friends
45
The (Seniors or) Elderly




Eligibility for Medicare = 65 = senior citizen/geriatric
“old” (65-85) vs “very old” (85+)
700 million seniors Worldwide (1.3 billion in 2040)
US life expectancy=78
 People at 65, expected to live another 18.7 years
 Women outnumber the man in the elderly population
Hospital
Independent
House
Assisted Living
Facility
Nursing Home
Hospice/Terminal
Care Facility
46
Age, Abilities and Deficits in Percentages
Age
Sensorimotor and Deficit
cognitive abilities (average)
(average)
Technology support
needed for daily
activities
60-70
100-90
0-10
Minimal
70-80
70
30
Moderate
80-90
50
50
High
0-10
90-100
Very high
90+
47
Activities of Daily Living and Falls




ADL includes hygiene, food, social needs,
medications, sleep, managing chronic conditions,
safety and financial needs (video clips of ADL)
Elderly with increased susceptibility to falls (hours
or days before someone finds out)
Health complications due to falls and the delayed
response increases the severity of such conditions
Detection of falls an important requirement


Automatic
detection
of
falls
based
on
detection/estimation of posture and pressure on sensorequipped floors
Visual fall detection along with context information
48
Grand Challenge









A 70 year old widow at home all by herself
Mild cognitive impairment, but does most of the household
work on her own
Wants to remain as much independent as possible (grown up
children in another state, but call to check if she is fine)
Wants to help her friends with similar problems
Multiple chronic illnesses requiring multiple medications
If not taking medications, her condition may become acute
Partial compliance: one or more side effects
A visit from home health nurse once a week
Goal to manage chronic conditions and delay her transition
to assisted living/nursing home for 10 years (independence
and $500,000 savings in her nursing home expenses)
49
Suitable ICT for the Elderly



Appropriate for sensori-motor and cognitive problems
wearable, portable, implanted, and environmental
Another classification can be





simple: computers, Internet, websites, cell phones and alarm
Intermediate: RFID, emergency alarms, medication and task
reminder systems, fall detection systems
Elaborate: Smart Home and all smart devices
Reliable, smart and context-aware, personalized,
robust, self-configuring, and no harm to the patients
Cognition, including executive function, decision
making and dual-task performance, decline with age
(dual-task could lead to increased task complexity, may
lead to falls, especially if the home is not well kept)
50
Some Examples of Current Research




Monitoring and stray prevention system (RFID, GPS,
GSM and Geographical Information Systems)
Markov decision processes (MDPs) to provide prompts
to a user for guidance through the activity of handwashing
Wrist-worn integrated health monitoring device
(WIHMD) for fall detection, ECG, blood pressure,
pulse oximetry, respiration rate, and body
temperature
Selective video-clips for automated/human analysis
(even remote help based on the context)
51
Medical Decision Making




Complex in terms of number of parameters and
variables, outcome possibilities, and information
that must be processed
Healthcare professionals need to make these
complex decisions with no margins for errors
15% of medical decisions lead to misdiagnosis
Cognitive load (Sweller, 1988): Excessive cognitive
load could affect the quality of decision making
(medical errors and lower quality of healthcare
services) and mobile devices could make it worse
52
Cognitive Load and Decision Making
Cognitive Capacity
(Working Memory)
Intrinsic
Cognitive Load
Extraneous
Cognitive Load
Germane
Cognitive Load
Total
Cognitive Load
+
Visual
Component
Audio
Component
Quality of
Decision Making
Prior
Experiences
Residual Load
53
Mobile Alert to Healthcare Professional
BP = 140/90
PATIENT-ID = U93
SPO2 = 99
Temp=98
Pulse = 120
Medications
Screen 1
Screen 2
Screen N
54
Simple Decision Making
(context-aware system)
Do Nothing
Get More
Information
Hospitalize
PATIENT-ID = U93
Likely condition 1
Get More
Information
Likely condition 2
Call a
Physician
Hospitalize
55
Conclusions and Future



Health Informatics is one of the most exciting
advances in healthcare and computing
With the increasing cost of healthcare and limited
healthcare professionals, health informatics can
play a very important role in the future of
healthcare
How information can be used most effectively in
various healthcare processes (EMR/EHR, health
monitoring, medication adherence, and medical
decision making)
56
Conclusions and Future

HI can further lead to many important advances in
healthcare and technologies





Proactive health and wellness management
Design and use of medications that are most suited to
individual patients
Healthcare systems that are context aware to provide
necessary interventions for health and medications
Smart technologies that can sense and support the needs
of elderly in independent living
Personalized and intelligent monitoring of patients can
lead to better health outcomes at lower healthcare cost
57
Questions?
For more information:
Pervasive Healthcare
Computing
EMR/EHR, Wireless and
Health Monitoring
Varshney, Upkar
2009 (Hardcover)
ISBN: 978-1-4419-0214-6
Springer.com
58