July 2009 Darrell M. Wilson, MD (Stanford)Slide 1Submission Doc: 15-09-0537-00-0006 July 2009 Slide 1 Project: IEEE P802.15 Working Group for Wireless.

July 2009

Darrell M. Wilson, MD (Stanford)Slide 1Submission

Doc: 15-09-0537-00-0006July 2009

Slide 1

Project: IEEE P802.15 Working Group for Wireless Personal Area Networks (WPANs)Project: IEEE P802.15 Working Group for Wireless Personal Area Networks (WPANs)

Submission Title: BAN and Diabetes a template for medical device communicationDate Submitted: May 14, 2009Source: Darrell M. Wilson, MDContact: Stanford Voice: +1 650 723-5791, E-Mail: [email protected]: Diabetes Abstract:.Purpose: Same

Notice: This document has been prepared to assist the IEEE P802.15. It is offered as a basis for discussion and is not binding on the contributing individual(s) or organization(s). The material in this document is subject to change in form and content after further study. The contributor(s) reserve(s) the right to add, amend or withdraw material contained herein.

Release: The contributor acknowledges and accepts that this contribution becomes the property of IEEE and maybe made publicly available by P802.15.

July 2009

Darrell M. Wilson, MD (Stanford)Slide 2

IEEE Body Area NetworkDiabetes - July 09

• Darrell M. Wilson, MD

• [email protected]

• dped.stanford.edu

July 2009

Darrell M. Wilson, MD (Stanford)Slide 3Submission

Doc: 15-09-0537-00-0006

Goals

• Review diabetes for a few minutes

• Discuss current conventional treatment approaches

• Discuss cut-edge approaches include closed loop systems and there problems

July 2009

Darrell M. Wilson, MD (Stanford)Slide 4Submission

Doc: 15-09-0537-00-0006

Goals

• What “we” envision as  Body Area Network  upsides for diabetes

• What “we” envision as important features/functional aspects  to such a network

• Q and A

Diabetes MellitusMajor Forms

• Insulin dependent

• IDDM

• Juvenile onset

• Brittle

• Type 1

• Non-insulin dependent

• NIDDM

• Adult onset

• Type 2

Atypical DiabetesMinor forms

GeneticsEnvironmental

triggers

Insulitis

Type 1 Diabetes

Diabetes Exposure

RenalComplications

EyeComplications

LargeVessels

Time Course of Diabetes

Time .....0

20

40

60

80

100

Pe

rce

nt

DemandMassFunction

Trigger?

Insulinresistantperiods

ClinicalPresentation

Honeymoon

Travis, DM in Children, MPCP#29, 1987Diab Care 29:1150, 2006

July 2009

Darrell M. Wilson, MD (Stanford)Slide 9Submission

Doc: 15-09-0537-00-0006

Diabetes Impact

• Type 1 ~ 800,000 to 1,000,000– ~120,000 < 20 years of age

• Type 2 ~ 7 million– another ~ 7 million undiagnosed– Prevalence

• 1.3% 18-44 years of age• 6.2% 45-65 years of age• 10.4% 65-74 years of age

$92

$109

$138

$40$47

$54

$132

$156

$192

$0

$40

$80

$120

$160

$200

$240

Direct Indirect Total

2002

2010

2020

Diabetes Care 26:917-932, 2003

Costs Continue to Increase (U.S.)(in Billions of Dollars)

Mazze DTT 2008

Single Subject without DM

Mazze DTT 2008

Single Subject With DM

Hemoglobin A1c

http://www.cem.msu.edu/~cem252/sp97/ch18/ch18s20.GIF

Hemoglobin A1c

http://home.comcast.net/~creationsunltd/images/comparebsandhga1c.gif

DCCT

DCCT NEJM, 329:977,1993

Glucose ControlGlycosylated Hemoglobin

DCCT NEJM, 329:977,1993

RetinopathyPrimary Prevention

DCCT NEJM, 329:977,1993

DCCT Data

Glycosylated Hemoglobin (%)5.0 5.5 6.0 6.5 7.0 7.5 8.0 8.5 9.0 9.5 10.0 10.5

Pro

gre

ssio

n -

Ret

ino

pat

hy

(per

100

pt-

yr)

0

2

4

6

8

10

Sev

ere

Hyp

og

lyce

mia

(per

100

pt/

yr)

20

40

60

80

100

120

Insulin Action Curves

Hours

0 5 10 15 20 25 30

Act

ion

0

20

40

60

80

100 LisproRegularNPH & LenteUltra

Four Shots

Time

0 4 8 12 16 20 24

Act

ion

July 2009

Darrell M. Wilson, MD (Stanford)Slide 22Submission

Doc: 15-09-0537-00-0006

Pumps

• What do they do?• Basal(s) rates• Meal boluses• Correction bolus• What don't they do?• Still open loop• Require a great deal of attention to detail

Pumps

Pump Example

Time

0 4 8 12 16 20 24

Act

ion

July 2009

Darrell M. Wilson, MD (Stanford)Slide 25Submission

Doc: 15-09-0537-00-0006

How to Select the Correct Amount of Insulin

• Good carbohydrate counting– Frequently in error

• Using pumps– Use the calculators/wizards

• Using injections– Use a discrete plan

• Adjusting for exercise

• Bedtime snacks

July 2009

Darrell M. Wilson, MD (Stanford)Slide 26Submission

Doc: 15-09-0537-00-0006

Pumps and Injections

• I like dose calculators– Earlier age of dosing “competency”– The paradox of both greater dose flexibility and

consistency• Time of day• Fine tuning

– Better download and data analysis • Meal “buckets”

– Future “automatic” adjustment of parameters– Lead into the feedback controlled pump

July 2009

Darrell M. Wilson, MD (Stanford)Slide 27Submission

Doc: 15-09-0537-00-0006

Measurement of Glucose

• Direct– Methods

• meters• future sensors

– Data analysis• average• variability• extremes

Lucile Packard Children's Hospital at StanfordPackard Pediatric Diabetes Center Diabetes Educator Line: (650) 498-7353

Insulin Scale for: Test subject

Standard Insulin DoseBreakfast Lunch DinnerNPH: NPH:Lantus: Lantus:Humalog Humalog Humalog

Insulin to Carbohydrate Ratio Carb Ratio= Amount of grams of carbs covered by 1 unit of Humalog

Breakfast Carb Ratio: Lunch Carb Ratio: Dinner Carb Ratio:12 15 18

Grams of Carbs Insulin Grams of Carbs Insulin Grams of Carbs Insulin6 0.5 H 8 0.5 H 9 0.5 H12 1.0 H 15 1.0 H 18 1.0 H18 1.5 H 23 1.5 H 27 1.5 H24 2.0 H 30 2.0 H 36 2.0 H30 2.5 H 38 2.5 H 45 2.5 H36 3.0 H 45 3.0 H 54 3.0 H42 3.5 H 53 3.5 H 63 3.5 H48 4.0 H 60 4.0 H 72 4.0 H54 4.5 H 68 4.5 H 81 4.5 H60 5.0 H 75 5.0 H 90 5.0 H

Correction insulin Correction Factor= How many points 1 unit of Humalog lowers blood glucose Target BG= Correction factor tries to bring BG to this desired number Do not use correction scale if your last shot was less than 2 hours agoBreakfast Lunch Dinner

Correction Target BG: Correction Target BG: Correction Target BG:

50 120 60 120 80 120 My BG is between My BG is between My BG is between

70 100 = subtract 1H 70 100 = subtract 1H 70 100 = subtract 1H

101 120 no extra 101 120 no extra 101 120 no extra121 145 +0.5 H 121 150 +0.5 H 121 160 +0.5 H146 170 +1.0 H 151 180 +1.0 H 161 200 +1.0 H171 195 +1.5 H 181 210 +1.5 H 201 240 +1.5 H196 220 +2.0 H 211 240 +2.0 H 241 280 +2.0 H221 245 +2.5 H 241 270 +2.5 H 281 320 +2.5 H246 270 +3.0 H 271 300 +3.0 H 321 360 +3.0 H271 295 +3.5 H 301 330 +3.5 H 361 400 +3.5 H296 320 +4.0 H 331 360 +4.0 H 401 440 +4.0 H321 345 +4.5 H 361 390 +4.5 H 441 480 +4.5 H346 370 +5.0 H 391 420 +5.0 H 481 520 +5.0 H371 395 +5.5 H 421 450 +5.5 H 396 420 +6.0 H 451 480 +6.0 H 421 445 +6.5 H 481 510 +6.5 H 446 470 +7.0 H

Total Humalog insulin dose= correction insulin + insulin for carbs

Insulin Variability

Heinemann DTT 4:673, 2002

July 2009

Darrell M. Wilson, MD (Stanford)Slide 30Submission

Doc: 15-09-0537-00-0006

Maximizing Bolus DeliveryGetting the Bolus

• The price of a missed bolus is high

Burdick Peds 113:211e, 2004

July 2009

Darrell M. Wilson, MD (Stanford)Slide 31Submission

Doc: 15-09-0537-00-0006

Kinetics vs Dynamics

Approximate Time (min)

0 100 200 300 400

Per

cen

t

0

20

40

60

80

100pharmacokineticspharmacodynamics

Snacks

LOW FAT

30 gm CHO2.5 gm protein

1.3 gm fat138 kCal

HIGH FAT

30 gm CHO2 gm protein20 gm fat320 kCal

July 2009

Darrell M. Wilson, MD (Stanford)Slide 33Submission

Doc: 15-09-0537-00-0006

Sensor LagSensor Lag

Time (minutes) (0 = start if meal)

-40 -20 0 20 40 60 80 100 120 140

Blo

od G

luco

se (

mg/

dl)

0

100

200

300

400

500

Freestyle Sensor

Feature Summary

Paradigm 722

DexCom7-plus

Navigator

Rate of change arrows

Yes Yes Yes

Projected low alarm

No No Yes

Days of wear 5 3-7 5

Ability to download Yes Yes Yes

Ability to integrate with pump

Yes (MiniMed)

No Pending (Cozmo)

>2 (mg/dL)/min

< -2 (mg/dL)/min

-1 to -2 (mg/dL)/min

1 to 2 (mg/dL)/min

-1 to 1 (mg/dL)/min

Trend ArrowsNavigator MiniMed

Updated every minute Updated every 5 minutes

July 2009

Darrell M. Wilson, MD (Stanford)Slide 36Submission

Doc: 15-09-0537-00-0006FreeStyle Navigator™ Continuous Glucose Monitor

Receiver

Transmitter

Sensor/Sensor Mount

FreeStyle Navigator™ System

Intended Features– Home continuous monitoring system. – 3-day sensor continuously measures

glucose– Transmitter sends updated glucose

reading every minute– Alarms for hi/lo glucose– Alarms for projected hi/lo glucose– On-board trend and statistical reporting– Event entry (food, insulin, meds, exercise,

etc)– 60-day memory & upload to computer– Traditional glucose meter built in

• System calibration

• Backup glucose meter

Pilot Study to Evaluate the Navigator in Children with T1D

• 30 children with T1D• HA1c 7.1 ± 0.6%• Smart pumps• Ask to wear sensor

daily• Algorithm based

adjustments of insulin infusion rates

MiniMed Paradigm REAL-Time Insulin Pump and

Continuous Glucose Monitoring System

DexCom 7 Plus

• 91 insulin requiring adults– 75 Type 1– 16 Type 2

• Three 72 hour wears• Randomized

– Blinded– Shows 2/3 wears

Garg Diabetes Care 29:44–50, 2006

July 2009

Darrell M. Wilson, MD (Stanford)Slide 41Submission

Doc: 15-09-0537-00-0006

Modes of Glucose Sensor Use

• Meter replacement• Hypoglycemia alarm

– Down alert

• Hyperglycemia alarm– Up alert

• Pattern recognition• Dynamic adjustment• Infusion controller

– Suggestive vs closed loop• Nocturnal pump shutoff for unaddressed low alarms

– Non-diabetic inpatients– Research studies

Enrollment

Randomization

RT-CGM “Usual Care”

6 mo. Outcome

Continue RT-CGM

12 mo. Outcome

run-in phase

Start RT-CGM

0 – 6 mos.

6 – 12 mos.

0-6 Months of the Study

Changes in A1c in > 25 yr olds

Difference: -0.53%

+

P-value <0.001

Differences in Distribution of A1c Levels in > 25 yr olds at 26 Weeks

Cum

ulat

ive

%

26 week glycated hemoglobin (%)

Changes in A1c in 8-14 yr olds

P-value=0.29

P=0.04

P=0.009

P= 0.01

Secondary A1c Outcomes in 8-14 yr olds

Changes in A1c in 15-24 yr olds

P-value=0.52

Artificial Pancreas (-cell) • Artificial Pancreas Software (APS)

Features:– Communication with sensors & pumps

– Modularity, Plug-and-Play (PnP)

– Human Machine Interfaces (HMIs)

– Physician control

– Data storage

– Audio & Visual alarms

– Standalone application

– Data recording

– Safety and redundancy

Controller

Insulin PumpHMI

APS HMI

Sensor HMI

APS

Database

Startup Interface

July 2009

Darrell M. Wilson, MD (Stanford)Slide 50Submission

Doc: 15-09-0537-00-0006

B. Wayne Bequette

Proportional-Integral-Derivative (PID) Control

dt

tdedttetekutu

tytrte

D

t

Ic

00

1

Manipulated

Input (insulin)

Error = setpoint – measured output

= desired glucose – measured glucose

Proportional

gainIntegral time Derivative time

• Integral “windup” can lead to postprandial hypoglycemia

• Many possible tuning procedures

B. Wayne Bequette

Internal Model Control (IMC)

Controller

(approximate

“model inverse”)

sensor pump subject

glucose setpoint

Insulin-

Glucose

Model

Sensor

Model

~ymodel-

predicted

output

measured

output (glucose)

y

+_

manipulated input

(insulin infusion)

u

r

Estimates of other variables (possible)

model-reality mismatch

July 2009

Darrell M. Wilson, MD (Stanford)Slide 52Submission

Doc: 15-09-0537-00-0006

B. Wayne Bequette

Controller Information

• Feedback-only• Meal Announcement (feedforward)• Meal Detection

– Meal size estimation vs. priming bolus

• Insulin-on-Board– Impose infusion constraint vs. modified tuning

• Individualization– Subject supplied history vs. results of protocol

July 2009

Darrell M. Wilson, MD (Stanford)Slide 53Submission

Doc: 15-09-0537-00-0006

Closed-loop vs. hybrid control

6 12 18 24 30 36 420

100

200

300

400Closed Loop (N=5)

meals

setpoint

Hybrid Closed Loop (N=5)

Glu

cose

(m

g/d

l)

Mean Nocturnal Peak PP

Full CL 156 (149-163) 109 (87-131) 232 (208-256)

Hybrid 135 (129-141) 114 (98-131) 191 (168-215)

July 2009

Darrell M. Wilson, MD (Stanford)Slide 54Submission

Doc: 15-09-0537-00-0006What we would like to see in a Body Area Network

• A stable, company neutral system to reliably exchange data among diabetes related devices– Glucose meters– Glucose sensors– Insulin infusion devices – Control algorithm devices (if not

embedded)

July 2009

Darrell M. Wilson, MD (Stanford)Slide 55Submission

Doc: 15-09-0537-00-0006What we would like to see in a Body Area Network

• A stable, company neutral system to reliably exchange data among diabetes related devices– External alarms– Activity monitors– GPS– Phone– External communication devices– Ear buds?

July 2009

Darrell M. Wilson, MD (Stanford)Slide 56Submission

Doc: 15-09-0537-00-0006What we would like to see in a Body Area Network

• Note the difference – NOT just a sensor network– Insulin infusion devices– Insulin infusion algorithm

• NOW we are infusing insulin, a potential lethal medicine

July 2009

Darrell M. Wilson, MD (Stanford)Slide 57Submission

Doc: 15-09-0537-00-0006What we would like to see in a Body Area Network

• High security

• High specificity

July 2009

Darrell M. Wilson, MD (Stanford)Slide 58Submission

Doc: 15-09-0537-00-0006What we would like to see in a Body Area Network

• Bidirectional communications between devices with confirmation and error checking

• Reasonable transmission range– thru the body at least

• Monitoring of BAN status

• Fails safely with clear warnings

July 2009

Darrell M. Wilson, MD (Stanford)Slide 59Submission

Doc: 15-09-0537-00-0006What we would like to see in a Body Area Network

• Easily interrogated (downloadable)– Cell phone, internet

July 2009

Darrell M. Wilson, MD (Stanford)Slide 60Submission

Doc: 15-09-0537-00-0006What we would like to see in a Body Area Network

• Privacy and safety

• Privacy vs safety

July 2009

Darrell M. Wilson, MD (Stanford)Slide 61Submission

Doc: 15-09-0537-00-0006What we would like to see in a Body Area Network

• And of course– Cheap– Low energy requiring– Long lasting– Green– Easy to use

Thanks!