Diabetes, a chronic metabolic disorder, is a global healthcare burden. There are 463 million cases of diabetes in persons aged 20-79 and another 374 million cases of impaired glucose tolerance, as reported by the International Diabetes Federation (IDF). 

Primary domains

The three primary domains of diabetes care that AI may impact and enhance are people with diabetes, healthcare providers, and healthcare systems. AI has improved resource efficiency in healthcare systems, brought quick and dependable decision-making, flexible follow-ups for medical professionals, and new dimensions of self-care for diabetic patients.

For patients

AI educates and empowers patients. Digital solutions affect patient comorbidities, behaviours, time spent in healthcare facilities, and the necessity for frequent travel and provider interaction, affecting healthcare systems. AI has also improved hospital admissions and transfers.

For medical personnel

Neural networks, deep learning, machine learning, and case-based reasoning allow for improved decision-making, self-management, automated retinal screening, and predictive population risk stratification. AI facilitates remote monitoring and decision-making, which benefits medical personnel.

Applications:

AI is widely used in automated retinal screening, clinical decision support, predictive population risk stratification, and patient self-management systems for diabetes. 

• Automated retinal screening - Deep learning algorithms automate diabetic retinopathy diagnosis. AI-based retina screening is feasible, accurate, and well-accepted for diabetic retinopathy identification and monitoring.
• Clinical decision support - Machine learning-based clinical decision support systems predict type 2 diabetes patients' short- and long-term HbA1c response after insulin introduction. These methods also reveal clinical factors that affect HbA1c response.
• Healthcare recommendation system (HRS) - It uses machine learning and analyzes lifestyle, physical health, mental health, and social network activities to predict illness risk, including diabetes.
• Genomics - The detection and treatment of diseases have advanced with molecular phenotyping, genomics, epigenetic changes, and digital biomarkers. Due to its heterogeneity and chronicity, diabetes generates massive data sets.
• Patient self-management tools - Self-management is crucial to diabetes treatment. Patients can now manage their diabetes, collect parameter data, and be their health experts thanks to AI.

Recent AI models

Numerous AI-based methods have been used in the treatment of diabetes. Diabetes diagnosis has advanced since the development of AI, beyond simple tests of haemoglobin A1c and blood glucose. AI has transformed how diabetes is identified, treated, and avoided, potentially reducing the disease's 8.8% worldwide prevalence. 

• AI model detects diabetes early in chest X-rays. The new AI algorithm indicates that ordinary medical X-ray pictures can see diabetes even in patients who don't fulfil high-risk criteria. This model could help doctors spot the condition earlier and avoid problems.
• A risk factor framework compared dozens of AI models for type 2 diabetes prediction.
• According to the latest study, scientists used six to 10 seconds of people's voices and primary health data like age, sex, height, and weight to construct an AI model that can detect Type 2 diabetes. The model is 89% accurate for women and 86% for men.

Conclusion

AI has been utilized in several recent research to anticipate diabetes. However, these machine-learning approaches have yet to show greater effectiveness in forecasting illness onset compared to standard statistical techniques that integrate risk factors. Despite this, we believe that ongoing machine learning research and efforts toward its practical application will maximize AI's predictive performance by using large amounts of organized data and abundant computational resources, thereby significantly enhancing the predictive accuracy of disease diagnosis, prevention, and treatment in diabetes.

Image source: Unsplash