Rebecca Araujo

Urban Health Today

Results of a randomized, controlled trial published in JAMA Network Open showed that an <a href="https://urbanhealthtoday.com/post/openai-and-fda-in-talks-to-accelerate-drug-evaluations-with-ai">artificial intelligence</a> (AI) digital tool provided effective inpatient clinical support for patients with type 2 diabetes (T2D).1
The study randomized 149 adult patients with T2D to receive insulin dosage titration based on recommendations of either an AI-based insulin clinical decision symport system (iNCDSS, n=75) or a senior physician. The AI-based system analyzes patient information and provides insulin titration recommendations, including dosage recommendations. For this analysis, insulin dosage titration was conducted for 5 consecutive days. Data from <a href="https://urbanhealthtoday.com/post/continuous-glucose-monitoring-is-under-prescribed-in-federally-qualified-health-centers">continuous glucose monitoring</a> were also collected. The primary outcome was amount of time patients spent in the target glucose range (70-180 mg/dL) during the study period.
The mean time spent in the target range was 76.4% in the iNCDSS group and 73.6% in the physician group, which achieved the prespecified noninferiority criterion of a 2.7% mean difference. Adverse events were similar between groups.
In addition, most physicians reported satisfaction with the iNCDSS system. On a scale of 1 to 5, physicians rated the AI digital tool on the following parameters:
<ul>
<li>Clarity (4.6)</li>
<li>Time-saving (4.2)</li>
<li>Efficacy (4.0)</li>
<li>Safety (4.4)</li>
</ul>
The overall satisfaction score was 4.1 of 5.0.
<h2>Limitations of AI Use in Diabetes Care</h2>
In an accompanying editorial also published in JAMA Network Open, Eileen R. Faulds, PhD, of The Ohio State University College of Nursing, acknowledged a few limitations of this study and potential roadblocks for implementation in clinical care.2
Dr. Faulds pointed out that “patients included in the trial were specifically selected for their glycemic control needs, with exclusions for acute illness or other inpatient medications that could interfere with glucose regulation.” She emphasized that acute or <a href="https://urbanhealthtoday.com/post/correlation-between-diabetes-thyroid-disorders-and-the-elevated-risk-for-hypertension">chronic illnesses</a> and the use of other medications, such as steroids, vasopressors, or dialysis, can affect glycemic control in the hospital setting. Thus, the decision support tool cannot be generalized to patients with these unique needs. “These results, although encouraging, may not be fully replicable in typical hospital environments.”
In addition, the glucose target range of 70 to 180 mg/dL is less conservative than recommended targets for patients with critical illness from the American Diabetes Association and the Endocrine Society. “Given these considerations, it is essential that AI-driven systems for inpatient care factor in these more conservative, higher target ranges to better align with clinical guidelines and ensure patient safety,” Dr. Faulds noted.
Despite these limitations, the findings of the study underscore the promise of AI tools in diabetes management. “As AI continues to evolve, there is tremendous potential for improving diabetes care in hospitals, provided that future models incorporate the full range of variables that influence glycemic control in these settings,” Dr. Faulds wrote.
<h4>References</h4>
<ol>
<li>Ying Z, et al. JAMA Netw Open.2025;8(5):e258910. doi:<a href="https://jamanetwork.com/journals/jamanetworkopen/fullarticle/2833619" target="_blank" rel="noopener">1001/jamanetworkopen.2025.8910</a>.</li>
<li>Faulds E. JAMA Netw Open. 2025;8(5):e258924. doi:<a href="https://jamanetwork.com/journals/jamanetworkopen/fullarticle/2833625" target="_blank" rel="noopener">1001/jamanetworkopen.2025.8924</a>.</li>
</ol>

Results of a randomized, controlled trial published in JAMA Network Open showed that an <a href="https://urbanhealthtoday.com/post/openai-and-fda-in-talks-to-accelerate-drug-evaluations-with-ai">artificial intelligence</a> (AI) digital tool provided effective inpatient clinical support for patients with type 2 diabetes (T2D).1The study randomized 149 adult patients with T2D to receive insulin dosage titration based on recommendations of either an AI-based insulin clinical decision symport system (iNCDSS, n=75) or a senior physician. The AI-based system analyzes patient information and provides insulin titration recommendations, including dosage recommendations. For this analysis, insulin dosage titration was conducted for 5 consecutive days. Data from <a href="https://urbanhealthtoday.com/post/continuous-glucose-monitoring-is-under-prescribed-in-federally-qualified-health-centers">continuous glucose monitoring</a> were also collected. The primary outcome was amount of time patients spent in the target glucose range (70-180 mg/dL) during the study period.The mean time spent in the target range was 76.4% in the iNCDSS group and 73.6% in the physician group, which achieved the prespecified noninferiority criterion of a 2.7% mean difference. Adverse events were similar between groups.In addition, most physicians reported satisfaction with the iNCDSS system. On a scale of 1 to 5, physicians rated the AI digital tool on the following parameters:<ul>
<li>Clarity (4.6)</li>
<li>Time-saving (4.2)</li>
<li>Efficacy (4.0)</li>
<li>Safety (4.4)</li>
</ul>The overall satisfaction score was 4.1 of 5.0.<h2>Limitations of AI Use in Diabetes Care</h2>In an accompanying editorial also published in JAMA Network Open, Eileen R. Faulds, PhD, of The Ohio State University College of Nursing, acknowledged a few limitations of this study and potential roadblocks for implementation in clinical care.2Dr. Faulds pointed out that “patients included in the trial were specifically selected for their glycemic control needs, with exclusions for acute illness or other inpatient medications that could interfere with glucose regulation.” She emphasized that acute or <a href="https://urbanhealthtoday.com/post/correlation-between-diabetes-thyroid-disorders-and-the-elevated-risk-for-hypertension">chronic illnesses</a> and the use of other medications, such as steroids, vasopressors, or dialysis, can affect glycemic control in the hospital setting. Thus, the decision support tool cannot be generalized to patients with these unique needs. “These results, although encouraging, may not be fully replicable in typical hospital environments.”In addition, the glucose target range of 70 to 180 mg/dL is less conservative than recommended targets for patients with critical illness from the American Diabetes Association and the Endocrine Society. “Given these considerations, it is essential that AI-driven systems for inpatient care factor in these more conservative, higher target ranges to better align with clinical guidelines and ensure patient safety,” Dr. Faulds noted.Despite these limitations, the findings of the study underscore the promise of AI tools in diabetes management. “As AI continues to evolve, there is tremendous potential for improving diabetes care in hospitals, provided that future models incorporate the full range of variables that influence glycemic control in these settings,” Dr. Faulds wrote.<h4>References</h4><ol>
<li>Ying Z, et al. JAMA Netw Open.2025;8(5):e258910. doi:<a href="https://jamanetwork.com/journals/jamanetworkopen/fullarticle/2833619" target="_blank" rel="noopener">1001/jamanetworkopen.2025.8910</a>.</li>
<li>Faulds E. JAMA Netw Open. 2025;8(5):e258924. doi:<a href="https://jamanetwork.com/journals/jamanetworkopen/fullarticle/2833625" target="_blank" rel="noopener">1001/jamanetworkopen.2025.8924</a>.</li>
</ol>

ADA 2025: 2 Studies Spotlight Diabetes Care for LGBTQ+ Patients

Diabetes

Lou Portero

Cannabis Use Disorder Linked to Higher Mortality in Colorectal Cancer Patients

Colorectal Cancer

Barry Greenberg, PhD

Monique Nugent, MD, MPH

Van Ta Park, PhD, MPH

Julie Zissimopoulos, PhD

A Personalized Approach Can Transform Alzheimer's Disease Care

Alzheimer's & Dementia

Latest News

AI-Driven Insulin Support: Are These Tools Ready for Diabetes Care?

An AI-based digital tool provided effective insulin dosage titration recommendations for inpatients with type 2 diabetes.