Key result
Artificial intelligence models, particularly neural networks, achieved a median accuracy of 90% for peripheral artery disease diagnosis, though clinical use is limited by biases and poor reporting.
Why the study?
PAD is prevalent globally and early, accurate diagnosis is critical, while recent advances in AI offer new opportunities to improve PAD diagnosis.
Do artificial intelligence models provide accurate and reliable diagnosis of peripheral artery disease?
Systematic Review
Do artificial intelligence models provide accurate and reliable diagnosis of peripheral artery disease?
While AI models for PAD diagnosis demonstrate high reported accuracy, their clinical utility is currently limited by significant methodological flaws, high risk of bias, and poor reporting standards.
Abstract Objective Peripheral artery disease (PAD) is a prevalent global condition where early and accurate diagnosis is critical for improving outcomes. Recent advances in artificial intelligence (AI) offer new opportunities to improve PAD diagnosis. This systematic review evaluates current AI approaches, their limitations, and their potential for future clinical application. Methods A systematic search of Embase, MEDLINE, PubMed, Cochrane, and Scopus identified studies on AI in PAD diagnosis published between May 2019 and May 2025. The review followed the PRISMA guidelines and was registered with PROSPERO. Model transparency and reporting quality were assessed using the Transparent Reporting of a multivariable prediction model for Individual Prognosis or Diagnosis method, and the risk of bias was evaluated using the Prediction model Risk of Bias Assessment Tool. Results Of 429 studies screened, 32 met the inclusion criteria. Sample sizes ranged from 5 to 28,868 participants. Data input varied widely across 13 measurement modalities, with Doppler ultrasound examination being the most common (n = 5). Neural networks were the most frequently applied model (n = 18), achieving a median accuracy of 90 ± 7.4%. Demographic data were often unreported; only 10 studies accounted for age and sex balance, and just 7 included a full spectrum of disease severity. No study fully adhered to Transparent the Reporting of a multivariable prediction model for Individual Prognosis or Diagnosis method, with 65% scoring low for reporting transparency. Under the Prediction model Risk of Bias Assessment Tool, 40% of the studies were rated at a high risk of bias in analysis. Conclusions AI methods for PAD diagnosis show promising accuracy, but their clinical applicability is hindered by dataset biases, poor reporting, and a lack of generalizability. Addressing these barriers through collaboration between engineers and clinicians, adherence to reporting standards, and the use of representative clinical datasets is essential. Future studies should prioritize clinical relevance to enable safe and effective translation of AI into patient care.
No takes yet. Share an insight, caveat, or question.
Wilson et al. (2025) conducted a systematic review in Peripheral artery disease (PAD). Artificial intelligence (AI) models was evaluated on Model accuracy. Artificial intelligence models, particularly neural networks, achieved a median accuracy of 90% for peripheral artery disease diagnosis, though clinical use is limited by biases and poor reporting.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: