Deep Learning–Based Prediction of Diabetes Enhanced by a Five-Factor Peripheral Arterial Disease Risk Score
DOI:
https://doi.org/10.31838/NJAP/08.02.23Keywords:
Multi-layer authentication, Confidential passcode, Role Integrated Certificate- Single Sign-On, Graphical user interfaceAbstract
People with diabetes commonly also have peripheral artery disease (PAD), which makes their cardiac and metabolic issues worse. The study utilized NHANES data from 2011 to 2018 to develop a five-factor PAD risk score for this investigation. The score was based on high blood pressure, smoking, high cholesterol, and having chronic renal disease. To find out whether someone had diabetes, their PAD score, age, BMI, medical test findings (lipids, blood sugar, inflammatory markers), smoking habits, and kidney test results were all considered. The researcher evaluated feed-forward, LSTM, CNN–LSTM hybrid, broad and deep, and autoencoder-classifier models. They were all utilized on a train-test split of 80% to 20%. The data was clean since it was pre-processed using both median imputation for missing values and outlier capping based on IQR criteria. The accuracy and AUC of the autoencoder-classifier on the test set went up when the PAD score was included. In the end, sequence-based approaches were still the best. When they utilized values that better revealed PAD risk, the diabetes estimates for a sample that represented the full nation were more accurate. Autoencoder approaches found the proper mix between how useful the models were and how hard they were to use.
References
1. E. T. H. Yeh and C. L. Bickford, "Cardiovascular complications of cancer therapy: incidence, pathogenesis, detection, and management,"J. Am. Coll. Cardiol., vol. 53, no. 24, pp. 2231–2247, 2009.
2. J. J. Moslehi, "Cardiovascular toxic effects of targeted cancer therapies,"N. Engl: J. Med., vol. 375, no. 15, pp. 1457–1467, 2016.
3. R. Dhingra, R. S. Vasan, and T. M. Lee, "A new paradigm in risk prediction for cardiovascular diseases using deep learning,"Circ: Res., vol. 125, no. 4, pp. 421–423, 2019.
4. C. Krittanawong et al., "Deep learning for cardiovascular medicine: a review of applications, opportunities, and challenges,"Nat. Rev. Cardiol., vol. 16, pp. 75–84, 2019.
5. S. Yoon et al., "LSTM‐based patient monitoring for cardiovascular risk prediction,"IEEE J. Biomed. Health Inform., vol. 23, no. 5, pp. 1995–2003, 2019.
6. C. Cheng et al., "An autoencoder‐based framework for classification of cardiovascular disease from electronic health records,"IEEE J. Biomed. Health Inform., vol. 24, no. 10, pp. 2899–2908, 2020.
7. X. Huang et al., "Deep learning for diabetes risk stratification using electronic health records,"IEEE Trans. Emerg. Top. Comput., vol. 8, no. 2, pp. 280–287, 2020.
8. L. Zhang et al., "A multilayer perceptron model for diabetes classification based on NHANES dataset,"IEEE Access, vol. 8, pp. 13589–13598, 2020.
9. J. D. Wilson et al., "Random forest versus logistic regression for predicting incident diabetes,"Int. J. Med. Inform., vol. 138, p. 104139, 2020.
10. R. M. Salazar et al., "Comparison of logistic regression and gradient boosting machine for diabetes prediction in NHANES,"IEEE J. Biomed. Health Inform., vol. 24, no. 4, pp. 1025–1032, 2020.
11. J. Q. Lin et al., "Interpretable machine learning models for diabetes prediction: a survey,"IEEE Rev. Biomed. Eng., vol. 14, pp. 217–231, 2021.
12. G. Esteva et al., "A guide to deep learning in healthcare,"Nat. Med., vol. 25, pp. 24–29, 2019.
13. A. Zamorano et al., "2016 ESC position paper on cancer treatments and cardiovascular toxicity,"Eur. Heart J., vol. 37, no: 36, pp. 2768–2801, 2016.
14. R. Moslehi et al., "Temporal patterns of cardiovascular risk in cancer survivors,"J. Oncol. Pract., vol. 15, no. 6, pp. 328–336, 2019.
15. P. Rajpurkar et al., "CheXNet: radiologist‐level pneumonia detection on chest X‐rays with deep learning," arXiv:1711.05225, 2017.
16. G. Litjens et al., "A survey on deep learning in medical image analysis,"Med. Image Anal., vol. 42, pp. 60–88, 2017.
17. S. Y. Lee et al., "Autoencoder‐based feature extraction for early detection of diabetes,"IEEE Trans. Clin. Informatics, vol. 5, no. 3, pp. 476–485, 2021.
