New AI Tool Aims to Catch Misdiagnosed Type 1 Diabetes in Adults

Adult-onset Type 1 diabetes is frequently misclassified as Type 2 diabetes. This mix-up is more than a paperwork error—it has real consequences. When patients receive the wrong diagnosis, they may miss out on specialist referrals, get treatment that doesn't address their actual condition, and face a higher risk of serious complications, including diabetic ketoacidosis (DKA), a dangerous acute condition that requires emergency care.

Researchers have developed an artificial intelligence algorithm that analyzes electronic medical records to identify adults likely to be misclassified. The algorithm looks at patterns in patient data collected over the previous 24 months—factors that are typical of Type 1 diabetes—and assigns a risk score. When a patient's score meets a certain threshold, the system flags them for a clinician to review more closely.

In earlier testing on historical medical data, the retrained algorithm showed promising results. But the researchers knew that testing on old data isn't the same as seeing how the tool performs in real clinical practice.

To answer that question, researchers launched a prospective study across two healthcare organizations in southeastern Pennsylvania. Unlike retrospective studies that look backward at data already collected, this study follows patients going forward in real time.

Here's how it works: The algorithm scores all adult patients currently diagnosed with Type 2 diabetes, looking for signs they might actually have Type 1. Patients who meet the high-risk threshold are presented to endocrinologists or primary care doctors for careful chart review. If misclassification is suspected, doctors can order appropriate follow-up testing, including autoantibody tests—the standard way to confirm Type 1 diabetes.

The primary goal is to measure the positive predictive value: how often the algorithm correctly identifies misclassified Type 1 diabetes patients among those it flags. The study will also track whether the referral process actually reaches patients and leads to appropriate testing and diagnosis.

Getting the right diagnosis matters. Patients with Type 1 diabetes need insulin; patients with Type 2 may be managed with other medications first. When Type 1 is missed, patients may spend months or years on the wrong treatment plan, only to eventually experience a crisis that forces diagnosis.

If this algorithm proves reliable in practice, it could become a routine part of healthcare systems, quietly working in the background to catch misclassified patients early. The goal is simple: help more people get the correct diagnosis sooner, so they can receive the treatment they actually need.

Source: BMJ open. Evidence type: PubMed indexed literature. Type1Cure is an information and intelligence hub, not a medical advice service. This article summarizes published research and does not provide diagnosis, treatment, or personal medical guidance. Always talk to your own care team before changing anything about your Type 1 diabetes management.