Diabetes And Technology Robert J. Rushakoff, MD Professor of - - PDF document

3/17/16 ¡ 1 ¡

Diabetes And Technology

Robert J. Rushakoff, MD Professor of Medicine University of California, San Francisco robert.Rushakoff@ucsf.edu

Disclosures

n None

3/17/16 ¡ 2 ¡

"Each ¡blind ¡man ¡perceived ¡the ¡elephant ¡as ¡something ¡ different: ¡a ¡rope, ¡a ¡wall, ¡tree ¡trunks, ¡a ¡fan, ¡a ¡snake, ¡a ¡ spear..." ¡ ¡

Telemedicine ¡ Central ¡ Pla/orms ¡ Wearable ¡ Devices ¡

¡ ¡

Apps ¡

Medica9on/ insulin ¡ Delivery ¡

Automa9on ¡

inpaCent/outpaCent ¡

Monitoring ¡

Personal/central ¡

3/17/16 ¡ 3 ¡

Diabetes And Technology

n Journals

n JOURNAL OF DIABETES SCIENCE AND TECHNOLOGY n DIABETES TECHNOLOGY & THERAPEUTICS

n National/International DM technology

meetings

n International Inpatient DM meetings

3/17/16 ¡ 4 ¡

Diva: Romeo and Juliet

Romeo

n a six-ounce, hand-held device that resembles a pocket calculator. n Glucose Monitor n Programmed to beep at set times as reminder when to test blood sugar, take insulin, eat meals and exercise n 3 month storage n Records blood sugar n With push of button, records insulin doses, amount of food eaten, intensity of exercise done and the times at which all those activities took place

Juliet

n device produces printouts n Can send data to provider using a telephone modem.

Robert Ratner, MD: It's not perfect for everybody. It's a lot of work, a lot of effort, and a lot of patients are unwilling to do that. And, frankly, for a lot of patients, it's not necessary. Patient’s MD: Those who benefit most those whose diabetes is out of control and those who are newly diagnosed and need to become aware of how different things affect them. Most people can use the system, for several months and then "graduate" to using just a diary and a simple blood sugar monitor. Those who want to, can buy their own system - hospitals lend or rent them to patients - but the system is expensive and not always reimbursable by insurance. Romeo costs about $495; Juliet, $275.

3/17/16 ¡ 5 ¡

• The treatment with DIANET vs conventional showed a better

metabolic control

• lower before breakfast: 87 +/- 6 vs 104 +/- 4 mg
• Lower before lunch: 85 +/- 5 vs 104 +/- 4 mg
• Lower after dinner: 102 +/- 5 vs 124 +/- 6 mg)
• These results were associated with higher insulin doses in

the DIANET vs conventional treatment, and a significant reduction of hypoglycemic reaction in both group

3/17/16 ¡ 6 ¡

Diva: Romeo and Juliet

n Chemstrip bG n When strip gone -

• device worthless

3/17/16 ¡ 7 ¡

Diva: Romeo and Juliet

n Chemstrip bG n When strip gone - - device worthless n Technology limited to single device

(expensive and was not covered by insurance)

n Time consuming n ? Who really needed it n Who will pay

Diva: Romeo and Juliet

Now 2016 Has anything changed since 1988?

3/17/16 ¡ 8 ¡

3/17/16 ¡ 9 ¡

Requirements for Successful “technology” Use

n Make stuff easier to do

n For the patient; For the MD/Nurse/Pharmacist

n Integration n Supports normal Workflow n Scalable n Sustainable n Cost effective

3/17/16 ¡ 10 ¡

General Concerns with Data

n Numbers, numbers and more

numbers:

n Potential to overwhelm patients,

clinicians or other care givers

n ? How to actually interpret all the

data and actually make real time use

• f the information

Key Issues

n While new technology is cool - -

n Have to show some improved

• utcomes

n Not short term studies

3/17/16 ¡ 11 ¡

Glucose Meters

n Generally - - still have to prick finger n You get glucose value n ? Remains on value for patients not on

insulin

n Patients on insulin - - more is better

Glucose Meters

n Numbers have to be in the context of what you’re

eating and doing

n The patient would love to have something they could

beam onto the food to figure out how many carbs, to figure out how much insulin to give.

n The key is trying to integrate numbers into action, so

the devices are trying to become smarter, to give some sort of narrative with the data.

So . . . . . . .

3/17/16 ¡ 12 ¡

Glucose Meters: Accu-Chek Aviva Expert

Works like pump for calculating doses

n calculates the amount of insulin

needed based on:

n the test result n expected carbohydrate intake n past bolus doses, often referred to

as “insulin on board.”

Glucose Meters: Abbot Freestyle Flash

• Measures glucose every minute in

interstitial fluid through a small (5mm long, 0.4mm wide) filament that is inserted just under the skin and held in place with a small adhesive pad.

• No finger prick calibration
• Disposable, water-resistant sensor

can be worn on the back of the upper arm for up to 14 days

• A reader is scanned over the

sensor to get a glucose result painlessly in less than one second.

• Scanning can take place while the

sensor is under clothing

3/17/16 ¡ 13 ¡

Glucose Meters: Abbot Freestyle Flash

Insulin Pumps

n Newer pumps more user friendly n Some integration with CGM n Touch screen, small n BUT - - still just pumps and requires a user who really

knows how to interpret data, make changes, input correct information.

3/17/16 ¡ 14 ¡

Stupid stuff

n Wrong time on meters n Wrong time on pumps

Continuous Glucose Monitoring

n CGM devices continue

to improve, with interfaces that wirelessly transmit data to smartphones

• r a cloud-based
• system. As an

example, the Dexcom G5 mobile CGM helps caregivers to monitor their family members with diabetes and also allows physicians to monitor several of their patients at once

3/17/16 ¡ 15 ¡

Smart Continuous Glucose Monitoring

Medtronic Pumps/CGM

SmartGuard™

• low glucose suspend feature
• two different
• suspension of insulin delivery

when the glucose levels are predicted to hit the low limit in the next 30 minutes

• Suspension set to when the

glucose levels hit the low limit.

• can automatically suspend insulin

infusion for a maximum of 2 hours when sensor glucose (SG) levels are predicted to approach a pre-determined threshold and, without intervention, will resume basal insulin delivery to its pre- set rate.

Reviewing Data Type 1 DM

n EVER download data from one

• r more devices

n Adults: 31% n Caregivers: 56%

n ROUTINE reviewer of Data

n Adults: 12% n Caregivers: 27%

n ROUTINE reviewer HbA1c vs

no review

n Adults: 7.2% vs. 8.1%; P = .03 n Children: 7.8% vs. 8.6%; P = .001

Wong JC, et al. Diabetes Technol Ther. 2015

Ever Download: black Routine Download: Striped Routine Review: white

3/17/16 ¡ 16 ¡

2014 Survey of Diabetes Apps

Arnhold M, et al. J Med Internet Res. 2014.

Review of top 6 apps in 2011

EndoGoddess: gone Bant: .99 too much 3 others gone 1 still there - -out of date info

3/17/16 ¡ 17 ¡

mySugr Diabetes Logbook

3/17/16 ¡ 18 ¡

Some Current Diabetes Apps

n Diabetek n Diabetic Connect n Diabetes Pilot Pro. Food database n Diabetes Tracker n BG Monitor Diabetes n OnTrack Diabetes n Diabetes in Check n Carb Counting with Lenny

Diabetes tracker

3/17/16 ¡ 19 ¡

Diabetes tracker BG Monitor Diabetes

3/17/16 ¡ 20 ¡

3/17/16 ¡ 21 ¡

Smartphone apps for calculating insulin dose: a systematic assessment

46 calculators that performed simple mathematical

• perations using planned carbohydrate intake and

measured blood glucose.

n

59% (n = 27/46) of apps included a clinical disclaimer

n

30% (n = 14/46) documented the calculation formula.

n

91% (n = 42/46) lacked numeric input validation,

n

59% (n = 27/46) allowed calculation when one or more values were missing

n

48% (n = 22/46) used ambiguous terminology

n

9% (n = 4/46) did not use adequate numeric precision

n

4% (n = 2/46) did not store parameters faithfully. BMC Medicine 2015 13:106

Smartphone apps for calculating insulin dose: a systematic assessment

n

67% (n = 31/46) of apps carried a risk of inappropriate output dose recommendation that either violated basic clinical assumptions (48%, n = 22/46) or did not match a stated formula (14%, n = 3/21) or correctly update in response to changing user inputs (37%, n = 17/46).

n

Only one app, for iOS, was issue-free

n

No significant differences were observed in issue prevalence by payment model or platform.

n

majority of insulin dose calculator apps provide no protection against, and may actively contribute to, incorrect or inappropriate dose recommendations that put current users at risk of both catastrophic overdose and more subtle harms resulting from suboptimal glucose control.

BMC Medicine 2015 13:106

3/17/16 ¡ 22 ¡

Telediab1 software system

n

Diabeo software is a bolus calculator with validated algorithms, taking into account SMPG level before meals, carbohydrate counts, and planned physical activity. Parameters personally tailored for adjustment

• f prandial and basal insulin dose are entered into the system for each
• patient. If fasting or postprandial SMPG do not meet target levels, the

system can suggests adjustments for carbohydrate ratio, long-acting insulin analog dose, or pump basal rates.

n

usual quarterly follow-up (G1), home use of a smartphone recommending insulin doses with quarterly visits (G2), or use of the smartphone with short teleconsultations every 2 weeks but no visit until point end (G3).

Efficacy of electronic logbook ± teleconsultation.

Guillaume Charpentier et al. Dia Care 2011;34:533-539

3/17/16 ¡ 23 ¡

Telediab1 software system

n Telemedicine and type 1 diabetes: is technology per

se sufficient to improve glycaemic control?

n

Among the high users, the proportion of informed meals remained stable from baseline to the end of the study 6months later (from 78.1±21.5% to 73.8±25.1%; P=0.107), but decreased in the low users (from 36.6±29.4% to 26.7±28.4%; P=0.005). As expected, HbA1c improved in high users from 8.7% [range: 8.3-9.2%] to 8.2% [range: 7.8-8.7%] in patients with (n=26) vs without (n=30) the benefit of telemonitoring/teleconsultation (-0.49±0.60% vs -0.52±0.73%, respectively; P=0.879). However, although HbA1c also improved in low users from 9.0% [8.5-10.1] to 8.5% [7.9-9.6], those receiving support via teleconsultation tended to show greater improvement than the others (-0.93±0.97 vs

• 0.46±1.05, respectively; P=0.084).

n

CONCLUSION:

n

The Diabeo system improved glycaemic control in both high and low users who avidly used the IDA function, while the greatest improvement was seen in the low users who had the motivational support of teleconsultations.

Diabetes Metab. 2014 Feb;40(1):61-6

Platform Overload

n

Every company has different platform

n

Patient shows up, you can quickly pull up one but spend 10 minutes figuring out how to do downloads

n

Multiple reports on platform and can take 10 more minutes to find best report

n

Would be nice to say -- everyone use this meter/pump/cgm but insurance companies (and patients) have other ideas

n

Devices and software with more sophisticated algorithms that can perform pattern recognition. The question is how well that type of advancement can work.

3/17/16 ¡ 24 ¡

Platforms: Glooko

n MeterSync technology can download

diabetes data from more than 40 meters, pumps and CGMs directly to a smartphone, integrate food and activity data, and share results with caregivers

• r providers

3/17/16 ¡ 25 ¡

Glooko’s Product Line Glooko Suite of Solutions

Glooko MeterSync Device and Mobile App For personal use Improves self-management Glooko Kiosk for Offices For clinical use Improves office workflow MyGlooko Web App + Glooko Population Tracker For remote monitoring Enables on-demand care Glooko APIs EHR and other system integrations

Platforms: Tidepool

nonprofit company based in San Francisco, is currently building three applications

n Uploader:

uploading data from insulin pumps, CGMs and blood glucose meters to the platform.

n Blip:

• nline platform with numbers viewable in one shareable interface.

n Nutshell:

mobile app, helps patients with diabetes to better manage the meals they eat and to properly dose insulin for them.

3/17/16 ¡ 26 ¡

tidepool

3/17/16 ¡ 27 ¡

nutshell

3/17/16 ¡ 28 ¡

? Smarter Apps

n Im2Calories

n

Google “automatic food diary,”

n

Google is tapping the artificial intelligence researchers it acquired when it bought DeepMind for $400 million to develop an system that can measure the calories in food from pictures.

n

system determines the depth of each pixel in an image, matches the results to a vast database of nutritional information, and then takes into account portions by gauging the size of the food relative to the plate it’s on.

n

In one test was able to calculate the accurate caloric total of two eggs, two pancakes, three strips of bacon, and the accompanying condiments. n Another by SRI international

Healbe’s GoBe Body Manager band

n Shows how many calories your eat and

how many you burn each day

3/17/16 ¡ 29 ¡

Beyond CHO counting

n Mealtime insulin dosing calculation

should focus on meal composition— including fat, protein, and glycemic index—fat, protein, and glycemic index can impact on blood glucose levels,”

n carbohydrate counting alone is too

simplistic and monolithic

Group at Univ of Melbourne

n

many patients struggle to accurately (within 10-15 g) calculate carbohydrate content (as their research group has shown in other research) in the case of younger patients, to remember to bolus for meals at all

n

group also has previously shown that carbohydrate counting does not need to be all that accurate (+/- 10 g) to obtain good prandial insulin cover

n

translation of the algorithm for insulin dosing by Bell et al into a practical tool that can easily be used by patients is challenging.

n

Developed a food insulin index - vs cho

n

Compared with carbohydrate counting, the FII algorithm significantly decreased glucose incremental area under the curve over 3 h (–52%, P = 0.013) and peak glucose excursion (–41%, P = 0.01) and improved the percentage of time within the normal blood glucose range (4–10 mmol/L) (31%, P = 0.001). There was no significant difference in the occurrence of hypoglycemia.

Diabetes Care October 2011 vol. 34 2146-2151

3/17/16 ¡ 30 ¡

Impact of Fat, Protein, and Glycemic Index on Postprandial Glucose Control in Type 1 Diabetes: Implications for Intensive Diabetes Management in the Continuous Glucose Monitoring Era Diabetes Care June 2015

• vol. 38 no. 6 1008-1015

Impact of Fat, Protein, and Glycemic Index on Postprandial Glucose Control in Type 1 Diabetes: Implications for Intensive Diabetes Management in the Continuous Glucose Monitoring Era Diabetes Care June 2015

• vol. 38 no. 6 1008-1015

n

digital health tools and cloud computing will open

• pportunities to develop sophisticated analytic

systems to automatically evaluate postprandial glucose data and provide dosing recommendations.

n

Carbohydrate counting is a challenging aspect to diabetes self-management, and requiring that fat and protein intake also be quantitated and incorporated in insulin dosing decisions will create an additional burden that few patients will be able to accomplish. The need for both practical simplicity and widespread use of advanced dosing algorithms will ultimately be resolved with the development of data analysis and decision-support tools that evaluate meal patterns to identify whether macronutrients are contributing to glycemic fluctuations and to provide individualized dosing recommendations to patients for their common meals, thereby eliminating the need for patients to routinely count carbohydrates and other macronutrients.

3/17/16 ¡ 31 ¡

All together

n Take picture of meal

n Figures out carbs/fat/protein

n Glucose from CGM n Auto calculation of insulin dose n What’s missing?

Ohio predictive model

n case-based reasoning to determine causes of

glucose excursions on retrospective cgm

n using physiologic monitors (Basis & Microsoft

bands) plus smart phone activity tracking apps (Map my Run/Ride), to take live cgm & with support vector analysis predict future hypoglycemia before it occurs (especially at night or during exercise)

3/17/16 ¡ 32 ¡

Privacy Policies of Android Diabetes Apps and Sharing of Health Information

n In 2012

n 20% of smart phone users have health

care app

n 7% primary care MDs recommended some

app

n Chose DM apps - 271

n In 6 months- - 60 became unavailable n 211 apps analyzed

• JAMA. 2016;315(10):1051-1052

Privacy and DM apps

n 81% had no privacy policies

n Of the 19% with policies:

n 80% collected user data n 48% shared data

n (only 10% of these said they would ask

about sharing)

3/17/16 ¡ 33 ¡

Privacy and DM apps

n Initial Download Permissions

n 82% full network access n 64% modify USB storage n 30% read phone status and identity n 14% find accounts on phone n 11% view wifi connections n 4% modify users contacts n 4% view call logs

Privacy and DM apps

n Transmission Analysis

n 82% collected and sharted data (insulin/

glucose) with 3rd party

n 82% placed cookies

n Sharing of sensitive Health info by apps

not prohibited by HIPAA

3/17/16 ¡ 34 ¡

Privacy and DM apps

n Patients might mistakenly believe that

health information entered into an app is private (particularly if the app has a privacy policy), but that generally is not the case. Medical professionals should consider privacy implications prior to encouraging patients to use health apps.

3/17/16 ¡ 35 ¡

Smartsox

For Preventing ulcers Sox with embedded thermal and pressure sensors smartphone to alert wearer if problem (ie pressure, change in blood supply) Smartsox - Univ Arizona fiber optics monitor pressure, temperature, joint angles

Implanted technology

n ITCA 650: exenatide - implanted

yearly

3/17/16 ¡ 36 ¡

Implanted technology

n

VC-01 bio-artifitial pancreas

n contains stem cells capable of producing insulin.

Implanted technology

n

InSmart

n implanted into the abdomen of

the patient. It contains a gel barrier that reacts in response to increasing blood glucose levels in the blood, and releases insulin accordingly.

n Unfortunately the gel reservoir

• nly lasts two weeks before it

needs to be filled again. This has to be done via an external port on the outside of the body. Insulin injected into the port fills the reservoir again.

3/17/16 ¡ 37 ¡

Bionic Pancreas Bionic Pancreas

3/17/16 ¡ 38 ¡

Open Artificial Pancreas System (OpenAPS)

Simplified Artificial Pancreas System (APS)

• designed to use existing approved

medical devices, commodity hardware, and open source software

• designed primarily for safety,

simplicity, and interoperability with existing treatment approaches as well as existing devices.

• we believe that OpenAPS can be demonstrated to

be both safer and more effective than current state-of-the-art standalone insulin pump therapy, and that this can be demonstrated far more easily than for the completely novel therapy approach employed by the full APS systems that have been in clinical trials for years and are still years away from FDA approval.

3/17/16 ¡ 39 ¡

Goals for “Inpatient Technology” Use

n Make stuff easier to do n Improve glucose

control

n Increase in glucoses in

range

n Reduce hypoglycemia

n Improve outcomes

n Infections n Mortality

n Reduce errors

n Orders n Administration n Documentation

n Reduce Costs

n Reduced length of

stay

n Reduced rate of

readmission

Hyperglycemia in Hospitalized Patients ucsf.logicnets.com Transition from Home to Hospital

3/17/16 ¡ 40 ¡ Simple Subcutaneous Insulin Algorithm for Management of the NPO Hyperglycemic Patient in the ICU

3/17/16 ¡ 41 ¡

3/17/16 ¡ 42 ¡

3/17/16 ¡ 43 ¡

SQ Algorithm Team

n Mekhala Patwardhan n Heidemarie W Macmaster n Andrew Maruoka n Amy Kuwata n Craig Johnson

Hypoglycemia Prediction Model

n Based on Washington St. Louis Model n Real time inpatient data (glucose/renal

function/hepatic function/weight/ medications/insulin doses)

n Reduction of severe hypoglycemia by

50%

n New Team. Programming into EPIC like

sepsis alert

3/17/16 ¡ 44 ¡