Machine Learning-AI Approach for Early Detection of Latent Autoimmune Diabetes in Adults (LADA) in a Primary Care Setting

Latent Autoimmune Diabetes in Adults (LADA) is a subtype of Type 1 diabetes often mistaken for Type 2 due to its gradual onset and overlapping features, leading to misdiagnoses and adverse patient outcomes. Recognizing the critical need for precise and early identification, this study employs a machine learning-AI approach to develop a model aimed at detecting LADA in primary care settings. Utilizing data from 144 patients at FLCH, our model seeks to differentiate LADA patients more effectively. Key biomarkers such as C-peptide, Glutamic Decarboxylase, IA2 Antibody, BMI, Glucose, and A1C levels were incorporated as predictive features. Implementing Stochastic Gradient Descent (SGD) and Balanced Random Forest (B-RF) algorithms, the model achieved detection accuracies of 93% and 91%, respectively, surpassing other tested methodologies. The most influential features contributing to this accuracy were identified as C-Peptide and the latest A1C levels. Validation on hypothetical patient data demonstrated the model's robustness, with both SGD and B-RF algorithms successfully classifying all samples. These results highlight the significant potential of machine learning-AI in supporting early LADA diagnosis and facilitating personalized treatment, thereby mitigating the risks associated with misdiagnosis. Nevertheless, challenges such as ensuring data privacy, integrating into clinical workflows, and addressing algorithmic bias persist. Future research and collaborative efforts will focus on refining the model, enhancing interpretability, and promoting its adoption in clinical practice, ultimately improving care quality for individuals with or at risk of LADA.

Laith Abu-Lekham is a Data Scientist with a strong background in machine learning, text mining, and healthcare analytics. He currently works at Finger Lakes Community Health in New York, where he focuses on improving clinical productivity, optimizing healthcare processes, and advancing patient outcomes, particularly in underserved rural communities. He holds an MS in Industrial and Systems Engineering from Binghamton University and a BS in Industrial Engineering from Jordan University of Science and Technology. Laith has published extensively in peer-reviewed journals and conferences, and he is passionate about using data-driven approaches to solve complex healthcare and operational challenges.

Dr. Jacob M. Sprouse is a board-certified clinical pharmacist with over 12 years of experience in diabetes care, education, and chronic disease management. He holds national credentials as a Certified Diabetes Care and Education Specialist and in Advanced Diabetes Management. Dr. Sprouse currently leads CGM and insulin pump integration at Finger Lakes Community Health, where he supports primary care teams in delivering advanced diabetes and chronic disease state services. He is also a 2025 North Star Health Leadership Fellow and was recently honored as Preceptor of the Year by Wegmans School of Pharmacy. His work focuses on bridging innovative diabetes technology with real-world patient care across diverse clinical settings.

Dr. Jacob M. Sprouse is a board-certified clinical pharmacist with over 12 years of experience in diabetes care, education, and chronic disease management. He holds national credentials as a Certified Diabetes Care and Education Specialist and in Advanced Diabetes Management. Dr. Sprouse currently leads CGM and insulin pump integration at Finger Lakes Community Health, where he supports primary care teams in delivering advanced diabetes and chronic disease state services. He is also a 2025 North Star Health Leadership Fellow and was recently honored as Preceptor of the Year by Wegmans School of Pharmacy. His work focuses on bridging innovative diabetes technology with real-world patient care across diverse clinical settings.