Cardiovascular/Stroke Risk Stratification in Diabetic Foot Infection Patients Using Deep Learning-Based Artificial...

Area of Science:

• Cardiology
• Artificial Intelligence
• Diabetology

Background:

• Diabetic foot infections (DFI) are severe, costly conditions complicating cardiovascular disease (CVD) and stroke risk stratification.
• Existing machine learning (ML) models face limitations due to nonlinearity, comorbidities, and validation challenges.

Purpose of the Study:

• To investigate deep learning (DL) solutions for predicting CVD/stroke risk in patients with DFI.
• To explore the efficacy of DL models, specifically Long Short-Term Memory (LSTM) and Recurrent Neural Networks (RNN), in this high-risk population.

Main Methods:

• A systematic literature search using Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) identified 207 relevant studies.
• DL models were developed to integrate diverse data, including biomarkers, carotid ultrasound image phenotype (CUSIP) lesions, and DFI severity.
• Analysis included assessment of AI bias and strategies for early CVD/stroke diagnosis in DFI patients.

Main Results:

• Deep learning models demonstrated viability for CVD/stroke risk stratification in DFI patients.
• Research confirmed strong designs in DL architectures for this specific risk prediction task.
• The study confirmed the hypothesis that DFI exacerbates atherosclerotic disease, increasing CVD/stroke risk.

Conclusions:

• Deep learning paradigms are highly effective for predicting CVD/stroke risk in DFI patients.
• DFI patients exhibit aggressive atherosclerotic disease, necessitating advanced risk prediction tools.
• The proposed DL approach supports early diagnosis and improved management strategies for CVD/stroke in DFI.