Introduction to Biosignals

darkfryingpanMobile - Wireless

Dec 10, 2013 (3 years and 7 months ago)

79 views



Introduction to
Biosignals

Dr. Gari D. Clifford



Head: Intelligent Patient Monitoring Group


University Lecturer & Associate Director


Centre for Doctoral Training in Healthcare Innovation


Institute of Biomedical Engineering


Department of Engineering Science


University of Oxford

gari@robots.ox.ac.uk



Today’s course









A bit of end
-
of
-
week fun ...



Four short lectures this morning:


Me


Why Affordable Healthcare



Technology?


Aoife



Physiology & Heart Sounds


Marco


Blood Pressure


Mauro


Pulse
Oximetry



And a lab in the afternoon on these three signals

But first ... Why these signals?









They are fundamental physiology



There’s so much you can do with
them!


Asthma, COPD, CVD, ...



They illustrate an important point:


There is no correct answer in medicine


(This is the point of today’s lectures)



They are cheap to collect!



... and healthcare spending is making us bankrupt

G. D. Clifford

Health Care spending Per Person 2007


Per capita health care spending in
poorer countries is ~100 times lower
than higher GDP countries


Healthcare quality is not
f
(spending)


WHO rankings


US 37
th

(Costa Rica is 36
th

!)


Germany 25
th



Canada 30
th



UK 18
th



Mexico 61
st


China 144
th



India 112
th






http://www.photius.com/rankings/healthranks.html







The poorest have the worst deal!

Why?








...

G. D. Clifford

1. Poor Technology & Infrastructure in
Resource
-
Poor Settings for Health Care


Old legacy hardware & instruments


Generally paper notes, static information


no QA!


Poor communication channels


Poor drug storage and supply


Poor waste disposal


Poor transport


Intermittent electricity


Information shortage


Skills shortages


Medical errors
!






Image © Robert Malkin / Annu. Rev. Biomed. Eng.

Malkin, RA, Design of Health Care

Technologies for the Developing World

Annu. Rev. Biomed. Eng. 2007. 9:567

87

G. D. Clifford

Ad
-
hoc

medical facilities

Images © Gari Clifford, Creative Commons License
-

http://creativecommons.org/licenses/by
-
nc
-
nd/3.0/us/



Free
-
market without regulation / standards



Out of calibration, incorrect supplies, little training

2. Too few trained medical people
....








G. D. Clifford

Global Maps: Peters Projection


Each country is drawn in proportion to its relative surface area

(‘normal’ world map erroneously enlarges countries nearer the poles.)








G. D. Clifford

Global Maps: Population 2002


Each country is drawn in proportion to its relative global population









G. D. Clifford

Global Maps: Working Physicians 2004


Relative proportions of physicians working in each country









G. D. Clifford

Global Maps: Population 2002


Each country is drawn in proportion to its relative global population









G. D. Clifford

3. Spiraling costs


Ageing populations & chronic disease epidemics



Increased awareness of diseases
-

obligation to treat



Increased demands for ‘heroic’ interventions



Lack of preventative measures / compliance



Increased complexity of treatments & diagnostics => more
infrastructure & training



NOT SCALABLE!


G. D. Clifford

Our solution?



Smart phones,

d
umb (cheap)
sensors

G. D. Clifford

Design principles



Sensors sourced in
-
country!




Use USB for
comms

and power




Clever signal processing on the
phone (it’s built for it!)




Upload directly to EMR (reduces
errors & allows data mining)




Use in
-
built sensors


Interested?:
ewh
-
oxford.org



G. D. Clifford

What now?


A quick 10m break



Then
Aoife

will introduce the cardiovascular system


& explain heart sounds.



Then Marco will explain how we can obtain blood pressure
measurements without actually putting a sensor into the blood
stream!



And then Mauro will explain how to find out how much oxygen
is in the blood stream, just by shining a light at you!




G. D. Clifford

Break Time


G. D. Clifford

Recent trends (2010)


739 million
mobile users in China, >99.5%
coverage



277 million
Chinese accessed the Internet
through mobile devices in June 2010 (
up 18.4%
in 6 months
!)



China Mobile projects
957 million
mobile
Internet users in 2014



Android handset sales grew
886%
worldwide
during 2010. (The closest competitor


the
iPhone

-

only grew 86% in the same period.)




G. D. Clifford

mHealth Projects in my group


Open Source Telemedicine Infrastructure for
COPD monitoring (SanaMobile.org)



Sleep Monitoring



ECG Monitoring



Heart Sounds for TB



Pulse Oximetry & Respiration
Analysis; Apnea of Prematurity



BP monitoring for hypertension
& CVD risk predictor



HIV drug
-
drug interactions





G. D. Clifford

£5
Blood Pressure Monitor



Phone replaces all expensive parts



(Joao, Marco, Mauro, Carlos, David & Arvind)




Use cheap BP cuff


remove costly analogue manometer



Insert cheap electronics

Freescale MPxx5050 family

Regular Blood Pressure
cuff

USB 2.0 full
-
speed, 10
-
bit ADC

MCU. Microchip PIC18
series

Android Phone with
USBOTG

G. D. Clifford

Standard Oscillometric Method


Humans are error
-
prone at reading BP


Machines are consistent

Cuff pressure signal and oscillation waveform (Lin.C.T et al,
2003)

MAP @ peak, SBP @ 70% of peak, DBP @

50% of max

G. D. Clifford

User Interface

G. D. Clifford

Assisted interface


G. D. Clifford

Integration with Health Records

G. D. Clifford

Summary & Future


Opportunity to transform healthcare


Low cost, high frequency health monitoring
(at Nyquist?) using mobile phones


Portable, integrated personalised health
record at the back
-
end


Local calibration of equipment


Multiple independent expert labelling of
medical events


Train algorithms to do most of the
diagnostics & provide feedback to improve
data



G. D. Clifford

Appendices

G. D. Clifford

What is wrong with data


analysis in healthcare?


We’ve essentially digitized a paper process



Technology is interfering with data recording, slowing the
process



Transcribing data with time delays leads to clinically significant
errors



G. D. Clifford

Under
-
sampling & under
-
automating


Compare nurse
-
verified
IBP with re
-
derived IBP
from original waveform


Nurse ‘samples’ every 1
-
120 minutes (median 60
min, IQR 30 min)


Nurse over
-
estimates BP
by av of 20 mmHg


AND hypotensive events
missed for av 4 hours



Repeated transient hypotension (linked
with


mortality) lasts minutes

Hug
et al .
Critical Care Medicine ,2010

G. D. Clifford

How can we satisfy Nyquist and make the
data more ‘honest’?


Obviously we could sample more frequently


Impractical, so sample irregularly!




(.. to beat average Nyquist)



Avoid human data transcription & deletion by
using personal monitoring with low cost
sensors



Even so


how are we going to synthesise all
this new data? We do not cope with existing
data



This is a GLOBAL opportunity to reform
healthcare data analysis

The global burden of disease & injury


WHO:
In 2030
Disability
-
Adjusted Life Years (DALYs) lost will be:



20% of total will be from
c
ommunicable

diseases, maternal and
nutritional conditions (c.f. 40% now)


66% will be from
non
-
communicable disease (all income groups
globally)

The Global Burden of Disease: 2004 Update. World Health Organization. 2004.

Healthcare in the future



9 out of the 10 top burdens on our healthcare systems in
2030 are likely to be
noncommunicable

diseases



Basic issues are diagnostics, treatment
delivery (supply chains) and compliance



The lack of trained healthcare workers in
resource poor regions means we need to
deliver healthcare through telemedicine



Mobile phones are almost ubiquitous and represent the
cheapest method for transmitting and recording
information to and from remote locations









They solve the supply chain issue!





G. D. Clifford

Where is the need/opportunity largest?



The more resource
-
constrained the region, the greater the need


The Guatemalan highlands


The Scottish Islands


Rural China, etc



These places also offer the greatest opportunity


They have the biggest incentive to drive change


Smaller financial incentive for large companies to push ‘traditional’ economics


Lack of regulation : need to self
-
regulate before politics & industry coalesce


The ‘consumer’ or ‘patient’ will demand (and pay for low
-
cost healthcare)


… and take responsibility for their own care (and medical data)



How will they do this?


Look at what is changing the fastest …

World Mobile (GSM) Coverage,
Jan
2005

(from http://www.coveragemaps.com/gsmposter.htm)

World Mobile (GSM) Coverage,
Jan
2007

(from http://www.coveragemaps.com/gsmposter.htm)

G. D. Clifford

TiGo in Guatemala, Honduras & El Salvador


http://alum.mit.edu/www/gari/

TiGo GSM Coverage: Image © TiGo Guatemala
http://tigo.com.gt/

G. D. Clifford

Global Maps: Public Health Spending 2002


Each country is drawn in proportion to its relative global spending on
public health










http://alum.mit.edu/www/gari/

G. D. Clifford

Global Maps: Private Health Spending 2002


Each country is drawn in proportion to its relative global spending on
private

health









G. D. Clifford

The
global

problems with healthcare


Spiraling costs


Ageing populations & chronic disease epidemics


Increased awareness of diseases
-

obligation to treat


Increased demands for ‘heroic’ interventions


Lack of preventative measures / compliance


Increased complexity of treatments & diagnostics => more infrastructure & training



Lack of longitudinal medical records (& portability)


Data!


Lack of standards & monitoring of compliance


Interchangeability / interoperability


Lack of data collection


Artifacts & noise in data (Lack of data fusion)



Medical errors


Human fallibility (tiredness, confusion, overwork, transcription)


Contradictory or missing information (broken equipment, different manufacturers)


Drug issues (interactions, incorrect dosage/drug, counterfeit)


Errors of omission (forgetting to perform tests, treatment compliance, losing results)


Training (Lack of suitable training courses or available skilled professionals)



Too few trained medical professionals


With poor communication & physical infrastructure / equipment









G. D. Clifford

Changing landscape of communication devices

Smart
phones cross
over with
cheap
phones

G. D. Clifford

Growth of mobile subscriptions


left
-

Mobile cellular subscriptions by level of development in 2000, 2005
and 2010*; right
-

Mobile cellular subscriptions per 100 inhabitants
between 2000
-
2010*. (International Telecommunication Union Statistics
)

Moving to patient
-
centric care


Mobile phone infrastructure to capture and transmit


Written/typed notes


Voice notes


Images (e.g. photos, radiologic, etc)


Videos (e.g. ultrasound, ECHOs)



Removing technology from being a barrier between patient and doctor


it’s now a (passive?) intermediary



Phone=security (something you have/know)



Access to a longitudinal, portable medical record to which you can add
information



Automatically notes your activity and provides feedback


moving to
proactive care



Your patterns can be compared with your past and 6billion+ others


rare events now have precedents.


Automated diagnostics


the mobile
phone as a medical instrument


Suite of sensors + USB connections



Ubiquitous connections to database
and network of experts



Product of expert labeling



Sufficient computational power for
automated or semi
-
automated
diagnosis and treatment
recommendations



http://sanamobile.org/demo.html