Prediction and diagnosis of post-stroke epilepsy using artificial intelligence approaches: a systematic review and meta-analysis
Clinical Snapshot
PICO Framework
| P — Population | Patients with stroke at risk of developing post-stroke epilepsy |
| I — Intervention | Artificial intelligence approaches for prediction and diagnosis |
| C — Comparator | Standard clinical assessment or reference standard diagnostic methods |
| O — Outcomes | Sensitivity, specificity, and area under the receiver operating characteristic curve for PSE prediction/diagnosis |
Bottom Line
This meta-analysis of five studies suggests that AI approaches show promise for predicting and diagnosing post-stroke epilepsy, with pooled sensitivity of 88% and specificity of 83%. However, the evidence base is extremely limited with methodological concerns including incomplete search strategy, lack of quality assessment, and no heterogeneity evaluation. The SROC AUC of 0.90 indicates good discriminative ability, but the small number of studies and emerging nature of the technology means these results should be interpreted cautiously. While the diagnostic accuracy appears promising, significant validation studies and regulatory approval would be needed before clinical implementation. Australian clinicians should await larger, higher-quality systematic reviews and local validation studies before considering AI tools for post-stroke epilepsy diagnosis in routine practice.
Key Findings
P Value: Not reported
Effect Size: Sensitivity 88%, Specificity 83%
Primary Outcome: Diagnostic accuracy of AI approaches for post-stroke epilepsy prediction/diagnosis
Nnt Or Sensitivity: SROC AUC 0.90 (95% CI 0.87-0.92) indicating good discriminative ability
Confidence Interval: Sensitivity 95% CI 0.78-0.94, Specificity 95% CI 0.79-0.86
Clinical Application
Limited by current availability of AI diagnostic platforms and need for technical expertise Would require TGA approval for diagnostic AI tools, integration with existing stroke care pathways, and consideration of Medicare funding for AI-assisted diagnostics Stroke patients in hospital settings where AI diagnostic tools are available
Abstract
INTRODUCTION: Post-stroke epilepsy (PSE) is a common complication following a stroke and is a major cause of epilepsy in the elderly. Artificial intelligence (AI) is currently developing rapidly in the medical field and has a promising outlook in disease diagnosis, treatment, and prognosis. METHODS: We screened five studies that fully met the requirements from the PubMed, Web of Science, and EMBASE databases using relevant search terms such as PSE and AI, and analyzed the role of AI in predicting and diagnosing PSE in these studies. RESULTS: The results showed that the sensitivity of AI in predicting and diagnosing PSE was 88% (95% CI 0.78-0.94), and the specificity was 83% (95% CI 0.79-0.86). The area under the summary receiver operating characteristic (SROC) curve was 0.90 (95% CI 0.87-0.92). CONCLUSION: These results indicate that using AI to predict and assist in diagnosing PSE demonstrates high specificity and sensitivity, and has certain prospects in the future auxiliary diagnosis of PSE.
References
- 1.Qu, T., Wu, Y., Lei, J., Dong, J., Wang, Q., & Wang, Y. (2026). Prediction and diagnosis of post-stroke epilepsy using artificial intelligence approaches: a systematic review and meta-analysis. Neurological Sciences. https://doi.org/10.1097/ALN.0000000000002960
Related Research
Sleep medicine reviews
Machine and deep learning in REM sleep behavior disorder: a scoping review and analysis of reporting quality
3 Aug 2026
Neurological research
Federated deep learning model for epilepsy seizure detection using electroencephalogram (EEG) signal
2 Aug 2026
Journal of psychiatric research
Systematic review of machine learning and deep learning models for EEG-based detection of depression.
2 Aug 2026
This content is for educational purposes for healthcare professionals only and does not constitute clinical advice. Clinical decisions should be based on individual patient assessment, current guidelines, and appropriate specialist consultation. Editorial Standards · Privacy Policy · Terms of Service