Gestational diabetes mellitus (GDM) is a prevalent complication during pregnancy, affecting a significant percentage of pregnant women globally. Defined as “diabetes diagnosed in the second or third trimester of pregnancy that was not clearly overt diabetes prior to gestation,” GDM poses serious health risks for both mother and child if not managed effectively. In Mexico, the prevalence of GDM is estimated to be between 10 and 12%, highlighting the urgent need for early detection and intervention strategies .
The study, led by Héctor Gallardo-Rincón and a team of researchers, introduces the MIDO GDM model—an innovative artificial intelligence (AI)-based prediction model designed specifically for the early detection of GDM in Mexican women. This pioneering research, published in Scientific Reports, underscores the potential of AI in transforming healthcare by providing accurate, early predictions that can lead to timely and effective interventions .
Methodology and Development of the MIDO GDM Model
The MIDO GDM model was developed using data from 1709 pregnant women who participated in the multicenter prospective cohort study ‘Cuido mi embarazo’ conducted between April 2019 and May 2021. Out of these, 860 women were used for building the model, half of whom had GDM. The model utilized an artificial neural network (ANN) approach, which offers several advantages over traditional statistical methods. These include the ability to handle non-structured and missing input data, identify complex non-linear relationships, and perform parallel data processing .
Key predictive variables identified for the model included maternal age, family history of type 2 diabetes, previous diagnosis of hypertension, pregestational body mass index (BMI), gestational week at first prenatal visit, parity, birth weight of the last child, and random capillary glucose level. These variables are easily collected and do not require extensive laboratory testing, making the model highly applicable even in low-resource settings .
The ANN model demonstrated high accuracy and sensitivity, with a precision of 70.3% and a sensitivity of 83.3% for identifying women at high risk of developing GDM. The model’s robustness was further validated through external datasets, achieving an area under the ROC curve of 0.9308, indicating excellent predictive performance .
Implementation and Impact
The practical application of the MIDO GDM model within the Mexican healthcare system is a critical step forward. Integrated into the MIDO healthcare strategy, which includes a series of digital platforms aimed at preventing various chronic diseases, the MIDO GDM model will facilitate early detection and personalized healthcare interventions for pregnant women. The system, already utilized in 388 primary healthcare centers across Mexico, offers healthcare professionals and patients access to risk assessments and personalized health recommendations through the Mi Salud Integral app .
This model is not only a tool for early detection but also a strategic innovation that promises to enhance the overall quality of prenatal care. By enabling early identification of GDM risk, healthcare providers can implement preventative measures sooner, potentially reducing the incidence of complications associated with GDM.
Conclusion
The development of the MIDO GDM model represents a significant advancement in the use of AI for healthcare. Its ability to accurately predict GDM using easily obtainable data makes it a valuable tool, particularly in low-resource settings. As the model continues to be refined and integrated into broader healthcare strategies, it holds promise for improving maternal and child health outcomes significantly.
The research, supported by a collaboration of experts from various institutions and the Carlos Slim Foundation, highlights the transformative potential of AI in healthcare. With continued innovation and application, AI-driven models like MIDO GDM could redefine the standards of prenatal care and chronic disease prevention worldwide .
By focusing on early detection and intervention, the MIDO GDM model not only addresses a critical healthcare need but also sets a precedent for future AI applications in medicine. This approach exemplifies how technology can bridge gaps in healthcare accessibility and quality, ultimately contributing to better health outcomes for populations at risk.
References:
Gallardo-Rincón, H., Ríos-Blancas, M. J., Ortega-Montiel, J., Montoya, A., Martinez-Juarez, L. A., Lomelín-Gascón, J., Saucedo-Martínez, R., Mújica-Rosales, R., Galicia-Hernández, V., Morales-Juárez, L., Illescas-Correa, L. M., Ruiz-Cabrera, I. L., Díaz-Martínez, D. A., Magos-Vázquez, F. J., Vargas Ávila, E. O., Benitez-Herrera, A. E., Reyes-Gómez, D., Carmona-Ramos, M. C., Hernández-González, L., … Tapia-Conyer, R. (2023). MIDO GDM: An innovative artificial intelligence-based prediction model for the development of gestational diabetes in Mexican women. Scientific Reports, 13(1), 6992. https://doi.org/10.1038/s41598-023-34126-7
