Can Type 2 Diabetes Be Reversed? Remission, Care, and AI Support
Episode 11 explores type 2 diabetes remission, supervised treatment, and how AI could support education and follow-up between appointments.
By Noah Vandal and Dr. Joseph Yoon, 2026-09-09
Some people with type 2 diabetes can achieve remission, with continued care to maintain and monitor their health. Episode 11 explores insulin resistance, supervised treatment, family support, and how AI could help patients stay connected to their care team between appointments.
Type 2 diabetes remission changes the conversation
Can someone with type 2 diabetes reach a point where their blood sugar stays below the diabetes threshold without glucose-lowering medication?
That question opens Episode 11 of the AI and Healthcare Podcast. In the 25-minute conversation, Noah Vandal and Dr. Joseph Yoon explore insulin resistance, approaches to treatment, and the practical support people need outside the clinic.
The episode builds on our earlier discussion of insulin resistance. Its central idea is hopeful: remission is possible for some people. Turning that possibility into a care plan requires a clear definition, appropriate treatment, and ongoing support.
This companion article adds current clinical sources to the discussion. It is educational; decisions about diet, medicines, or surgery need an individual assessment with a qualified healthcare professional.
Remission has a clinical definition
The international consensus supported by the American Diabetes Association uses HbA1c below 6.5% for at least three months without usual glucose-lowering medication as the standard definition. Glucose testing should continue at least yearly, along with routine checks for diabetes complications.
A good result while taking medicine is valuable evidence that treatment is working. It does not, by itself, establish remission. Stopping treatment simply to meet a definition would put the label ahead of the person's health. Medication decisions belong with the care team.
Insulin resistance helps explain what can improve
Dr. Yoon describes the relationship between the body's response to insulin and the pancreas's ability to supply it. When cells respond less effectively, the pancreas may compensate. Blood glucose can rise when that compensation is no longer sufficient.
As NIDDK explains, insulin resistance, prediabetes, and type 2 diabetes are related but distinct. Fasting glucose and A1C help assess glucose regulation; a fasting glucose reading alone does not directly measure insulin resistance. Tests specifically measuring insulin resistance are mainly used in research.
For patients, this makes the next conversation more concrete: what do the results show, what is the current treatment goal, and what would count as meaningful improvement?
Treatment needs a plan that can be maintained
The conversation covers metabolic surgery, low-calorie programmes, and carbohydrate reduction. Those approaches differ in their evidence, suitability, risks, and support requirements.
The ADA's 2026 guidance recommends individualized nutrition, activity, and behavioral support for people with type 2 diabetes and overweight or obesity. Very low-calorie interventions require carefully selected patients, trained practitioners, and close medical monitoring. A low-carbohydrate plan is one option to discuss; no single diet label guarantees remission.
The NHS Type 2 Diabetes Path to Remission Programme illustrates how much surrounds the food itself: specially formulated meal replacements, gradual food reintroduction, coaching, monitoring, and coordination with the patient's practice. That is a structured treatment pathway, not simply an instruction to eat less.
NIH-supported research on metabolic surgery has also found sustained glucose benefits and higher remission rates than medical and lifestyle treatment in the studied participants. Surgery still requires individualized assessment and long-term follow-up.
The episode's discussion of fasting should not be treated as a plan to stop eating. NIDDK warns that missed meals and fasting can increase the risk of low glucose, particularly with insulin or certain other diabetes medicines. Major food changes need medication review as part of the same plan.
Most of the work happens between appointments
One of the most useful moments comes late in the episode. Dr. Yoon recalls patients bringing glucose records to a visit and trying to reconstruct them in the waiting room. The clinician could see the patient that day, but had limited context about the weeks before it.
That gap is easy to recognize. A patient leaves an appointment with instructions, then meets the realities of shopping, work, appetite, cost, family meals, and questions that arise after the office closes.
For a practice, a useful follow-up process would answer:
- What should the patient record, and how can they do it with minimal effort?
- Who reviews the information, and when?
- How can the patient ask a question before the next visit?
- Who handles concerns that need a prompt clinical response?
- How will the team learn whether the plan is manageable at home?
Collecting more data only helps when someone can use it. A log that nobody reviews can create extra work without improving the conversation.
AI could help patients stay connected to care
At about 20 minutes, the discussion turns to AI. Noah and Dr. Yoon consider whether technology could extend education and encouragement beyond the time available in a clinic visit.
Building on that discussion, possible uses include explaining clinician-approved information, helping someone prepare questions, reminding them about an agreed task, or organizing a record for the care team. These possibilities do not establish that an AI assistant produces remission or that SpeechSage provides diabetes treatment.
Any clinical deployment would need a defined purpose and a reliable handoff to people. A general chatbot should not decide that someone is in remission, prescribe a restrictive diet, or change insulin or other medication doses. Patients also need to understand whether information is being reviewed and how quickly they can expect a response.
Our Episode 10 guide to evaluating healthcare AI offers a useful starting point: test the specific job, the errors that matter, and the human workflow surrounding the tool.
Family support can make the plan easier to live with
The episode closes with the home environment. Dr. Yoon points out how difficult a dietary change can be when a person is trying to make it alone while the rest of the household follows a different routine.
Support can be practical: preparing a shared meal, making time for an agreed activity, helping with transportation, or writing down questions together. The person living with diabetes should have a say in what help they want.
Remission is a possible outcome, not a test of effort or character. The work remains worthwhile when it leads to safer treatment, better communication, or a more manageable daily routine, whether or not remission occurs.
Common questions
Can type 2 diabetes be reversed?
Some people achieve remission. That means meeting clinical criteria for sustained improvement without usual glucose-lowering medication. Remission can end, so it is not a permanent cure. A clinician should assess and document it.
Does a good A1C while taking medication mean diabetes is in remission?
It can show that treatment is working, but it does not establish remission under the usual medication-free definition. Do not stop prescribed treatment to pursue that label; the care team should decide whether any medication change is appropriate.
Is a very low-calorie diet safe to start on your own?
A structured very low-calorie programme requires careful selection, clinical monitoring, nutritional support, and medication review. It is not a general recommendation to fast or stop eating. Discuss major changes in food intake with your clinician first.
Can AI put type 2 diabetes into remission?
The episode discusses possible support roles for AI, including education, encouragement, and organizing information for a clinician. It does not establish that an AI assistant causes remission. General chatbots should not diagnose remission or independently adjust diabetes treatment.