Feasibility Study of PID Versus MPC and HMS

Sansum Diabetes Research Institute

Status and phase: Completed; Early Phase 1

Conditions: Type 1 Diabetes Mellitus

Treatments: Device: PID control algorithm; Device: MPC control algorithm

Study type: Interventional

Funder types: Other

Identifiers

NCT01987206

2013-11 C2T

Details and patient eligibility

About

The goal of this proposed study is to explore the feasibility of using a PID (Proportional-Integral-Derivative) controller versus an MPC (Model Predictive Control) controller algorithm in an artificial pancreas system, all other components and study design being equal.

The study consists of an evaluation of either type of control algorithm as a part of the Artificial Pancreas (AP) device during two periods of 27.5-hour closed-loop control in a clinic environment (Sansum Diabetes Research Institute, Santa Barbara, CA) separated by a minimum of 5 days and a maximum of 2 weeks. The 27.5-hour period includes: 2 announced meals (dinner and breakfast of 65g and 50g CHO respectively) preceded with a dose of rapid-acting insulin equivalent to 100% bolus based on each subject's Insulin to Carbohydrate (I:C) ratio and 1 unannounced meal (lunch of 65g carbohydrates, same meal content as dinner); complete night from 12:00 am to 7:00 am. The goal is to demonstrate that the AP device is able to maintain the subject blood glucose within a safe range at all times.

Enrollment: 10 patients

Sex: All

Ages: 21 to 65 years old

Volunteers: Accepts Healthy Volunteers

Inclusion criteria

• Clinical diagnosis of type 1 diabetes for at least one year and using an insulin pump for at least 6 months with commercially available rapid acting insulin
• The diagnosis of type 1 diabetes is based on the investigator's judgment; C peptide level and antibody determinations are not needed.
• Age 21 to 65 years
• For females, not currently known to be pregnant or nursing
• HbA1c between 5 to 10%, as measured with DCA2000 or equivalent device
• Willing to perform the calibration of the study CGMs using a finger stick only and willing to follow instructions for insulin pump and CGM wear.
• Willing to use the study CGM and study insulin pump during closed-loop.
• Able to and agrees to avoid the following medication starting 24 hours before sensor wear through completion of the close loop study visit: acetaminophen, prednisone, and pseudoephedrine.
• An understanding of and willingness to follow the protocol and sign the informed consent.

Exclusion criteria

• Exhibit hypoglycemia unawareness.

• Indications of cardiac arrhythmia.

• Pregnancy (as determined by a positive blood pregnancy test performed in females of childbearing capacity during screening visit and urine test at time of admission for in-patient visit) or nursing mother.

• Females who are sexually active and able to conceive that do not use contraception.

• Diabetic ketoacidosis in the past 6 months prior to enrollment requiring emergency room visit or hospitalization

• Severe hypoglycemia resulting in seizure or loss of consciousness in the 12 months prior to enrollment

• Current treatment for a seizure disorder; Subjects with a history of seizures may be included in the study if they receive written clearance from their neurologist

• Active infection

• A known medical condition that in the judgment of the investigator might interfere with the completion of the protocol such as cognitive deficit.

• Mental incapacity, unwillingness or language barriers precluding adequate understanding or co-operation, including subjects not able to read or write.

• Coronary artery disease or heart failure.

• Subjects with a history of coronary artery disease may be included in the study if they receive written clearance from their cardiologist

• Presence of a known adrenal disorder

• Active gastroparesis

• If on antihypertensive, thyroid, anti-depressant or lipid lowering medication, lack of stability on the medication for the past 2 months prior to enrollment in the study

• Uncontrolled thyroid disease; Adequately treated thyroid disease and celiac disease do not exclude subjects from enrollment

• Abuse of alcohol

• Current use of a beta blocker medication

• Laboratory results:

  1. Hematocrit < 30% or >55%
  2. A1C > 10%
  3. Abnormal liver or renal function (Transaminase >2 times the upper limit of normal, Creatinine> 1.5 mg/dL)
  4. Labs drawn at screening visit or within one month prior to screening (for other purposes) will suffice for enrollment purposes related to hematocrit

• Subject has skin conditions that, in the determination of the investigator, would preclude wearing the study devices (infusion set and sensor), in the abdomen. Examples include but are not limited to: psoriasis, burns, scaring, eczema, tattoos, and significant hypertrophy at sites of device wear; any known allergy to medical adhesives.

• Currently on long-term treatment using prednisone. If subject had been on short term treatment of prednisone, defer enrollment until underlying condition and prednisone treatment have resolved.

• Allergy to study drug, food or other study material.

• Clinically significant screening ECG, physical examination, laboratory test, or vital sign abnormality.

• Exposure to any investigational drug within 30 days.

• History of malignancy within the 5 years before screening (other than basal cell carcinoma).

• Currently smoking or discontinued smoking (including cigarettes, cigars, pipes) over the past 6 months.

• Current participation in another investigational trial.

Trial design

10 participants in 2 patient groups

PID algorithm with HMS

Active Comparator group

Description:

The control algorithm, at its core, is a Proportional-Integral-Derivative (PID)controller that incorporates an Internal Model Control (IMC) based tuning rule using an explicit model of human T1DM glucose-insulin dynamics. Parameters of the model are personalized based on a priori easily available subject parameters. This controller divides the control action into three components - the proportional distance between the current measurement and the target setpoint, the accumulated integral error as expressed by the area between the current state curve and the target set point over time, and the derivative rate of change of the current measurement. The Health Monitoring System algorithm uses the same glucose monitoring (CGM) data as the PID control algorithm but utilizes a separate algorithm for trending and predictions of future glucose values. Using a redundant and independent algorithm is an important safety feature of the overall AP device.

Treatment:

Device: PID control algorithm

MPC algorithm with HMS

Experimental group

Description:

The first control strategy is a flavor of Model Predictive Control (MPC) algorithm. MPC employs an explicit model of the process to be controlled when optimizing the input. Specifically, MPC controllers for glycemia control use a model of a human's T1DM insulin-glucose dynamics to predict the evolution of the blood glucose values over a so-called prediction horizon of controller steps, and optimize a predicted insulin input trajectory in order to optimize a specified cost objective that penalizes unsafe glycemic values, and also insulin usage. The Health Monitoring System algorithm uses the same CGM data as the MPC control algorithm but utilizes a separate algorithm for trending and predictions of future glucose values. Using a redundant and independent algorithm is an important safety feature of the overall AP device.

Treatment:

Device: MPC control algorithm