Life-changing bionic pancreases simplify diabetes management

Welcome to the 18th issue of the AstroFeather AI newsletter!

[ Daily AI News | Newsletter Signup | View Online ]

The bionic pancreas (artificial pancreas) market is heating up as investors take note of the impact that recently FDA-approved devices will have on people with type 1 diabetes (T1D), and in this issue we'll take a deep dive into a leading artificial pancreas system and some of its competitors.

In this week's issue, you'll also find some helpful updates on medical AI technology (including product launches) and recent implementations of AI and technology in hospital systems, including Mass General Brigham and HCA Healthcare.

I hope you enjoy reading and if you have any helpful feedback, feel free to respond to this email, contact me directly on LinkedIn (@adideswilliams), or follow the AstroFeather LinkedIn page!

Thanks - Adides Williams, Founder @ AstroFeather

In this week’s recap (15 min read):

• Update #1: Life-changing bionic pancreases simplify diabetes management

• Update #2: Medical Tech: Product Launches, Trials, and FDA Approvals

• Update #3: Healthcare Industry News: AI and tech implementations in the hospital setting

Must-Read News Articles and Updates

Update #1. Life-changing bionic pancreases simplify diabetes management.

The latest: The iLet Bionic Pancreas is an exciting device (and automated insulin delivery software) innovation that simplifies daily diabetes management for people with type 1 diabetes and has recently raised $100 million in new equity funding (following FDA approval).

Type 1 diabetes (T1D) is a chronic (lifelong) inflammatory disease characterized by the body's inability to produce insulin due to autoimmune destruction of the insulin-producing cells in the pancreas, called pancreatic beta cells. Insulin is an important hormone that plays a role in regulating blood glucose (also known as blood sugar) levels by allowing the body's cells to take up glucose from the bloodstream and use it as a source of energy. Without insulin, the body's cells can't use glucose, resulting in dangerously high blood sugar levels that can (over time) cause serious damage to the heart, kidneys, eyes, nerves, and many other complications.

To help those diagnosed with T1D maintain safe blood glucose levels, Beta Bionics developed the iLet Bionic Pancreas, a pocket-sized insulin delivery device that doesn't require carbohydrate counting, is initialized only with the user's body weight, and is designed to make all insulin dosing decisions using a collection of adaptive, self-learning algorithms.

How it works (Part 1) - Artificial Pancreas Basics: Artificial pancreases (also known as hybrid closed-loop or automated insulin delivery (AID) systems) are remarkable medical device systems that mimic the way a healthy pancreas controls blood glucose levels by continuously tracking blood glucose and automatically adjusting insulin delivery as needed, with minimal user intervention. The artificial pancreas generally consists of three components:

1. Continuous glucose monitoring (CGM) device: This device uses a small sensor, inserted under the skin (often in the arm or stomach) to continuously track glucose levels in the body's interstitial fluid (the thin layer of fluid that surrounds the cells). The CGM then wirelessly transmits glucose level information to a program (also known as a control algorithm) stored on a computer, smartphone, or insulin pump (most likely embedded on a chip in the pump hardware).

2. Insulin delivery software (control algorithms - the "brains" of the artificial pancreas): Based on the glucose readings received from the CGM, the control algorithms work together to calculate the appropriate insulin dose and delivery rate to maintain blood glucose levels in an acceptable range. The control algorithms make these calculations and adjustments in real time (mimicking the function of a healthy pancreas) and send dosing instructions to an insulin infusion pump to deliver the insulin accordingly.

3. Insulin infusion pump (and infusion set): Once instructions are received from the dosing software, the insulin pump delivers insulin into the body through an infusion set (a system of tubes that connects the pump to the body) to help keep blood glucose levels in a healthy range.

How it works (Part 2) - The iLet Bionic Pancreas System: Armed with our basic understanding of artificial pancreases, we can examine the iLet Bionic Pancreas System, which is a collection of wearable medical devices that work together to orchestrate all insulin dosing with minimal user input. The iLet System consists of a bionic pancreas device (which combines an insulin pump and automated insulin dosing software), an infusion set, and a continuous glucose monitor (CGM):

1. User setup and device initialization: The initial setup process for the iLet system involves several tasks. Users must first wirelessly pair the iLet device with a compatible CGM (Dexcom G6). Next, they are instructed to fill an iLet cartridge with insulin (Humalog or Novolog) and then insert the filled cartridge into the device's insulin chamber. An insulin infusion set is then connected to the iLet device's insulin chamber to fill the infusion set's tubing and cannula with insulin. Finally, users place the insulin infusion set on the body and enter their body weight to initiate the system. [Hands-on demo]

2. Glucose readings from the CGM sensor/transmitter: Once the user has placed a compatible CGM sensor on their body and paired its transmitter with the iLet device, glucose readings can begin to appear on the iLet device's LCD screen. At the time of this writing, the iLet Bionic Pancreas is only advertised as compatible with the Dexcom G6 CGM, which measures interstitial glucose via a glucose oxidase reaction. The Dexcom G6 sensor has a glucose oxidase coated filament that reacts with glucose in the interstitial fluid to produce gluconic acid and hydrogen peroxide. Electrochemical oxidation of the hydrogen peroxide produces an electrical signal that is converted to a corresponding glucose value. The transmitter attached to the sensor records the glucose value and transmits it every 5 minutes to a receiver (in this case, the iLet bionic pancreas device) where it is used by adaptive control algorithms to determine insulin dosing.

3. Insulin dosing software (adaptive control algorithms): Once the iLet device receives glucose readings from the CGM, a collection of three insulin dosing algorithms processes the readings and converts them into dosing instructions for the insulin pump. These dosing algorithms continuously adapt to the user's food intake and insulin needs and are described as follows: 1) the basal algorithm determines requirements for basal insulin ("background" insulin delivered between meals and overnight), eliminating the need to set basal rates; 2) the bolus correction algorithm automatically adds insulin above basal requirements and reduces insulin as needed to prevent hypoglycemia (unhealthy low blood sugar); and 3) the meal announcement algorithm provides meal insulin doses tailored to the user, automatically adapts based on dosing history of similar meals, and eliminates the need to count carbohydrates or know insulin-to-carbohydrate ratios.

4. Insulin infusion pump: After receiving instructions from the dosing software (based on CGM blood glucose readings), the insulin pump delivers customized doses of insulin through a small cannula inserted under the skin every five minutes (or as needed based on the user's changing needs).

Results: User feedback and the results of extensive clinical trials studies have been positive. In general, users experienced better blood sugar control and spent more time with their glucose levels in the target range compared to standard of care diabetes management methods (including real-time continuous glucose monitoring (CGM) paired with any insulin delivery method):

Clinical trial studies: Study results from a landmark multi-center clinical trial, suggest that the bionic pancreas is better at maintaining blood sugar levels in a healthy range for T1D patients than standard of care diabetes management methods:

• Study setup: During the 13-week study, conducted at 16 clinical sites across the US, participants with T1D between the ages of 6 and 79 years old who had been using insulin for at least one year, were randomly assigned to a treatment group using the iLet Bionic Pancreas or a standard of care control group (using their personal insulin pre-trial delivery method and a CGM).

• Bionic pancreas improves blood sugar control: Participants using the bionic pancreas experienced a reduction in glycated hemoglobin level (A1c – a measure of a person’s average blood glucose (sugar) levels over the past three months) from 7.9% to 7.3%. However, the standard of care group did not experience a change in glycated hemoglobin levels over the same time period. Interestingly, in an extension study, participants in the standard of care control group switched to using the bionic pancreas for 13 weeks and experienced improvements in glucose control that were similar to the initial bionic pancreas group in the randomized trial!

• Bionic pancreas improves blood sugar levels for children: Results from a companion paper focused only on the child participants (6 – 17 years old) showed that the bionic pancreas users experienced a decrease in mean A1c (8.1% at baseline to 7.5%) during the 13-week study, while A1c remained unchanged in the standard of care group. Notably, children with poor diabetes control (baseline A1c greater than (or equal to) 9.0%) experienced the greatest reduction in mean A1c from 9.7% to 7.9%.

• Bionic pancreas extends time in target glucose range: Participants who used the bionic pancreas spent 11% more time, or 2.6 hours per day, in the target glucose range (70 to 180 mg per deciliter) than the standard of care control group. The bionic pancreas group also spent significantly less time in hyperglycemia (high blood glucose greater than 180 mg per deciliter) and severe hyperglycemia (blood glucose greater than 250 mg per deciliter) than the standard of care control group.

Yes, but: While the benefits of the iLet Bionic Pancreas are many and the system is seen as a well-received additional option for diabetes management, a system that fully automates insulin dosing may not be the best fit for some:

• User confidence in automated insulin delivery: Those that are new to an automated insulin delivery system will undoubtedly experience some initial discomfort or unease with the idea of giving control of insulin dosing to an algorithm. In an interview with Evaluate Advantage, Ed Damiano (inventor of the iLet Bionic Pancreas) emphasizes that “you can’t override the insulin dosing decision,” acknowledging that some users may be uncomfortable with the idea of letting the system control all insulin doses. In an interview with Diabetes Strong, Sean Saint, CEO of Beta Bionics, acknowledges that the iLet Bionic Pancreas may not be the best fit for the “hyper-accurate carb counter” and may be more appropriate for those who comfortable with approximating carb counts and allowing the automated system to manage insulin delivery.

Behind the news: The concept of an artificial pancreas is not new and has been around for several decades. The first experimental closed-loop insulin pump, worn like a backpack, was described in the early 1960s, while the first computerized closed-loop system, the refrigerator-sized Biostator, was developed in 1974. Fortunately, advances in technology have led to artificial pancreases that consist of discreet, wearable devices designed for everyday use!

Since the FDA cleared Medtronic's MiniMed 670G system in 2016, there has been a race to develop increasingly advanced artificial pancreas systems. Below, I briefly highlight a few recently FDA-cleared additions to the competitive artificial pancreas market:

MiniMed 780G System from Medtronic: The MiniMed 780G is an automated insulin pump system designed (and FDA cleared) for the management of T1D in individuals aged 7 years and older and is comprised of the Guardian 4 CGM, MiniMed 780G insulin pump (infusion set), and dosing software (“SmartGuard” and “Meal Detection”):

• CGM: The MiniMed 780G system has been FDA cleared for use with the Guardian 4 sensor (and transmitter), which provides real-time glucose readings (every 5 minutes) and doesn’t require finger-stick calibrations like the Guardian 3.

• Dosing software (control algorithms): Proprietary SmartGuard and Meal Detection algorithms work together to control glucose to a target as low as 100 mg/dL (close to the average glucose level for people without diabetes). The SmartGuard algorithm uses current and past blood glucose levels to anticipate, adapt, and correct insulin delivery, while the Meal Detection algorithm corrects for underestimated carbohydrate counts or when a user occasionally misses a meal dose.

• Insulin pump (and infusion set): Based on instructions from the dosing software, the MiniMed 780G insulin pump automatically adjusts and corrects insulin delivery through an infusion set that can be worn for up to 7 days.

Tandem Mobi with Control-IQ from Tandem Diabetes: The Tandem Mobi and Control-IQ (dosing software) combination is one of the newer entrants in the artificial pancreas market, is fully controllable via an iPhone mobile app, and is FDA approved for the management of T1D in individuals 6 years old and above:

• CGM: The Tandem Mobi is compatible with the Dexcom G6 CGM (like the iLet Bionic Pancreas) which provides glucose readings every 5 minutes.

• Dosing software (control algorithm): The Mobi pump features Tandem’s proprietary Control-IQ software, which works to control blood sugar levels to a target range of 112.5 – 160 mg/dL during normal use and can automatically adjust insulin levels based on CGM readings. Control-IQ also uses a predictive algorithm that minimizes fluctuations in glucose levels by predicting how they may change up to 30 minutes in advance.

• Insulin pump (infusion set): Marketed as the world's smallest pump, the Tandem Mobi pump features a 200-unit insulin cartridge and is compatible with a collection of compatible (detachable) infusion sets.

Why it matters:

• Type 1 diabetes (T1D) isn't just about counting carbohydrates and administering insulin. Across the lifespan, T1D can have a profound and lasting impact on emotional and psychological well-being. According to a writeup by the CDC, more than half of people with diabetes report experiencing stigma (including negative attitudes and discrimination) because they have diabetes. On top of this, the day-to-day self-management of T1D, while absolutely critical, can be stressful and burdensome for the millions of our friends, family members, and co-workers who are living with the condition.

• In 2021, it's estimated that there were approximately 8.4 million people living with T1D worldwide, and this number is projected to increase to 13.5 - 17.4 million people with T1D by 2040. In the US, there are approximately 1.4 - 1.9 million people living with T1D (including children and adolescents), with an estimated 64,000 people diagnosed with T1D each year, and an estimated $16 billion in annual T1D-related healthcare costs and lost income.

• Although they are not a cure for T1D, artificial pancreases have emerged as a promising option for helping the millions of people with T1D manage their diabetes. By continuously monitoring blood glucose levels and delivering precise doses of insulin when needed, artificial pancreases can help improve blood glucose control, reduce short-term complications such as hyperglycemia and hypoglycemia, and potentially minimize long-term risks associated with T1D, such as kidney disease. As the stories in this update illustrate, artificial pancreases can also reduce the burden of constant blood glucose monitoring, carbohydrate counting, and manual insulin delivery, allowing people with T1D to spend more time enjoying their lives without the constant worry of managing blood glucose levels.

Update #2. Medical Tech: Product Launches, Trials, and FDA Approvals.

Zeta Surgical has received FDA clearance for its digital navigation system: The Zeta Cranial Navigation System helps surgeons perform precise procedures in the brain and uses computer vision to create mixed reality overlays on a standard screen, providing "GPS-like" guidance in different environments. Zeta's AI continuously adjusts the displayed images for maximum accuracy, enabling minimally invasive neurosurgical procedures without general anesthesia or rigid cranial immobilization. Interestingly, this news follows OnPoint Surgical also receiving FDA clearance for its AR-based surgical guidance system for spinal implant placement procedures.

Insight Medbotics has received FDA clearance for the world's first surgical robot designed to operate inside the bore (opening) of an MRI machine: The image-guided automated robot (IGAR) currently assists in performing breast biopsies to evaluate suspicious lesions for cancer. The robot is constructed primarily of plastic and non-ferrous metals to avoid interference with the magnetic field, its moving parts are built into the patient’s bed, and it’s powered by a separate control cart that is placed outside the scanner's range.

Asensus Surgical has partnered with Nvidia to accelerate the development of its Intelligent Surgical Unit (ISU): The collaboration aims to enhance the ISU's real-time intra-operative surgical image analysis platform, which uses augmented intelligence to reduce surgical variability. Asensus will use Nvidia's tools and accelerated computing technology to enhance the ISU's augmented intelligence capabilities, including features such as digital tags, 3D measurement, and enhanced camera control.

Techsomed has received FDA clearance for its VisAble.IO ablation treatment planning and confirmation software: The software helps physicians plan and assess ablation coverage of liver tumors and offers benefits such as 3D visualization of the ablation target within a patient-specific anatomical view. It integrates common imaging modalities such as ultrasound, computed tomography, and magnetic resonance imaging to provide real-time visualization of the entire ablation zone and personalized treatment planning tools. The software also provides an interactive 3D view of ablation margins and missed volumes, helping to assess ablation target coverage after the procedure.

Noah Medical has released the first-in-human clinical data from its Galaxy system: The Galaxy robotic system is designed to navigate the airway and probe the lungs for potential cancers and other conditions. The system combines single-use hardware with real-time imaging software and allows surgeons to collect biopsy samples from suspicious nodules in the lungs. Preliminary results from the clinical trial showed a 100% success rate in navigating the bronchoscope to targeted lung lesions, with diagnostic yields ranging from 89.5% to 94.7%. No major complications were reported during or after the procedure.

German Bionic has introduced the Apogee+ powered exoskeleton to support caregivers: The exoskeleton provides personalized lifting assistance to caregivers, addressing safety and job satisfaction concerns. It provides up to 66 pounds of back support during lifting and ambulation, with integrated grips to facilitate patient repositioning. The Apogee+ is dustproof and waterproof, making it suitable for washing or showering patients, and features extensive reporting capabilities for workplace management.

Update #3. Healthcare Industry News: AI and tech implementations in the hospital setting.

Google Cloud: Google has been quite busy with healthcare partnerships this year! In addition to collaborating with Mayo Clinic, Cognizant, Pager, and Redox, they've also added Huma, HCA Healthcare, and Ginkgo Bioworks to their growing list:

• Huma (a digital health company) will use Google Cloud's generative AI (genAI) services to enhance its digital disease management product. The platform, which centralizes patient data for clinician assessment or patient self-management, will now have genAI to automate the generation of clinical summary reports. Huma also plans to use Google Cloud's Med-PaLM 2, a large language model (LLM) developed for healthcare, to improve communication between patients and providers and optimize patient triage.

• HCA Healthcare, in collaboration with Google Cloud and Augmedix, has launched a pilot program equipping emergency room physicians with Google's AI technology to document medical information more efficiently. The physicians use the Augmedix app to create medical notes from patient conversations, which are then converted to medical notes and transferred to the hospital's electronic health record. HCA Healthcare is also considering deploying Med-PaLM 2 to summarize insights from medical texts.

• Ginkgo Bioworks has entered a five-year partnership with Google Cloud to bring AI to synthetic biology. Ginkgo plans to train AI models on genomic and protein data to advance drug discovery, agriculture, industrial manufacturing, and biosecurity. As part of the agreement, Ginkgo will make Google Cloud its primary provider of computing services, and Google Cloud will provide funding to help Ginkgo achieve specific milestones over the next 3 years.

Microsoft and Paige AI: Paige, a digital pathology company, has partnered with Microsoft to develop the "world's largest" AI model using images of cancer biopsies. Paige previously developed an AI model for tumor detection using more than 1 billion images from 500,000 pathology slides. With Microsoft's Azure cloud computing support, Paige plans to build a larger model with billions of parameters and variables using up to 4 million digitized microscopy slides. The collaboration aims to provide greater accuracy and new capabilities in oncology.

GE Healthcare: GE HealthCare has partnered with Mass General Brigham to develop an AI algorithm for healthcare scheduling prediction. The algorithm is designed to predict missed appointments, missed follow-ups and late arrivals. The first application of the algorithm will be integrated into the Radiology Operations Module (ROM), a digital imaging tool designed to reduce costs and administrative burdens. The collaboration aims to provide healthcare providers with streamlined data and actionable insights powered by AI and machine learning, allowing them to improve productivity and efficiency and focus on patient care.

Meta: Surgeons are utilizing Meta's Quest 2 virtual reality (VR) headsets to simulate medical procedures, allowing doctors to practice from home. CNBC reports that an orthopedics resident at Kettering Health Dayton in Ohio, used the VR headset to simulate a shoulder-replacement surgery before performing the actual procedure. The VR simulation helped him fine-tune his skills and avoid potential risks, resulting in a successful surgery. Overtime, VR will prove to be a valuable tool in healthcare training and treatment.

Thanks for reading this issue of the AstroFeather newsletter and I’ll see you in the next issue!

If you enjoy AstroFeather content, be sure to share this newsletter!

Adides Williams, Founder @ AstroFeather (astrofeather.com)