Application of Stacking Ensemble Learning in Clinical Fitting of Orthokeratology Lens for Myopia Correction

Abstract

Aiming at the problems of a large difficulty coefficient and tedious process in the clinical fitting of the orthokeratology(OK) lens, a stacking ensemble learning model is proposed to predict the parameters of the OK lens and realize its intelligent fitting. The feature set that is most suitable for the target variables is constructed by feature derivation based on F-test and feature selection under the variance-improved Boruta algorithm. A stacking ensemble learning prediction model is studied. The model uses random forest(RF), gradient boosting decision tree(GBDT) and support vector regression(SVR) as the first layer basic learners and linear regression(LR) as the second layer meta-learner. The experimental results show that the prediction indexes of the model are highly consistent with the clinical diagnosis results, which verifies that the model can be used as an effective auxiliary diagnosis method.

Keywords

orthokeratology(OK)lens / feature engineering / stacking ensemble model / parameter prediction / intelligent fitting

References

[1] LOUGHEED T. Myopia:the evidence for environmental factors[J]. Environmental Health Perspectives, 2014, 122(1):47-63.

[2] MRUGACZ M, GAJECKA M, MRUKWA-KOMINEK E, et al. Myopia:risk factors, disease mechanisms, diagnostic modalities,and therapeutic options 2019[J]. Journal of Ophthalmology, 2020,2020:1-2.

[3] LEE S S Y, LINGHAM G, SANFILIPPO P G, et al. Incidence and progression of myopia in early adulthood[J]. JAMA Ophthalmology, 2022, 140(2):162-169.

[4] GAO H, MILES T P, TROCHE R, et al. Quality of vision following LASIK and PRK-MMC for treatment of myopia[J]. Military Medicine, 2022, 187(9/10):e1051-e1058.

[5] NUCCI P, DRACK A V. Refractive surgery for unilateral high myopia in children[J]. Journal of American Association for Pediatric Ophthalmology and Strabismus, 2001, 5(6):348-351.

[6] KERSHNER R M. Clear corneal cataract surgery and the correction of myopia,hyperopia,and astigmatism[J]. Ophthalmology, 1997, 104(3):381-389.

[7] PREPAS S B. Light,literacy and the absence of ultraviolet radiation in the development of myopia[J]. Medical Hypotheses, 2008, 70(3):635-637.

[8] LIN H J, WEI C C, CHANG C Y, et al. Role of chronic inflammation in myopia progression:clinical evidence and experimental validation[J]. EBioMedicine, 2016,10:269-281.

[9] NICKLA D L, ZHU X Y, WALLMAN J. Effects of muscarinic agents on chick choroids in intact eyes and eyecups:evidence for a muscarinic mechanism in choroidal thinning[J]. Ophthalmic and Physiological Optics, 2013, 33(3):245-256.

[10] GONG Q W, JANOWSKI M, LUO M, et al. Efficacy and adverse effects of atropine in childhood myopia:a meta-analysis[J]. JAMA Ophthalmology, 2017, 135(6):624-630.

[11] CHO P, CHEUNG S W, EDWARDS M. The longitudinal orthokeratology research in children (LORIC) in Hong Kong:a pilot study on refractive changes and myopic control[J]. Current Eye Research, 2005, 30(1):71-80.

[12] CHO P, CHEUNG S W. Retardation of myopia in orthokeratology (ROMIO) study:a 2-year randomized clinical trial[J]. Investigative Ophthalmology & Visual Science, 2012, 53(11):7077-7085.

[13] WALLINE J J, JONES L A, SINNOTT L T. Corneal reshaping and myopia progression[J]. The British Journal of Ophthalmology, 2009, 93(9):1181-1185.

[14] MORGAN P B, EFRON N, WOODS C A, et al. International survey of orthokeratology contact lens fitting[J]. Contact Lens and Anterior Eye, 2019, 42(4):450-454.

[15] FU A C, QIN J, RONG J B, et al. Effects of orthokeratology lens on axial length elongation in unilateral myopia and bilateral myopia with anisometropia children[J]. Contact Lens and Anterior Eye, 2020, 43(1):73-77.

[16] WONG T Y, FERREIRA A, HUGHES R, et al. Epidemiology and disease burden of pathologic myopia and myopic choroidal neovascularization:an evidence-based systematic review[J]. American Journal of Ophthalmology, 2014, 157(1):9-25.

[17] ROSUERO MLG, BOMBIN AA, ROMERO R, et al. Clinical tool to measure fluorescein patterns in orthokeratology[J]. PeerJ. 2022,10:1-15.

[18] RAJABI M T, HOSSEINI S S, GHORBANI Z, et al. Utility of orthokeratology contact lenses;efficacy of myopia correction and level of patient satisfaction in Iranian myopic/myope-astigmatic patients[J]. Journal of Current Ophthalmology, 2015, 27(3/4):99-102.

[19] MANICKAM P, MARIAPPAN S A, MURUGESAN S M, et al. Artificial intelligence (AI) and internet of medical things (IoMT) assisted biomedical systems for intelligent healthcare[J]. Biosensors, 2022, 12(8):562.

[20] HOGARTY D T, MACKEY D A, HEWITT A W. Current state and future prospects of artificial intelligence in ophthalmology:a review[J]. Clinical & Experimental Ophthalmology, 2019, 47(1):128-139.

[21] ABDULLAH T A A, ZAHID M S M, ALI W. A review of interpretable ML in healthcare:taxonomy,applications,challenges,and future directions[J]. Symmetry, 2021, 13(12):2439.

[22] FAN Y Z, YU Z K, TANG T, et al. Machine learning algorithm improves accuracy of ortho-K lens fitting in vision shaping treatment[J]. Contact Lens and Anterior Eye, 2022, 45(3):101474.

[23] ZHANG Q T, XIE P Y, YANG L N, et al. A machine learning model on orthokeratology lens fitting based on the data of optometry examination[J]. Chinese Journal of Ophthalmology, 2019, 55(2):105-110. (in Chinese)

[24] ZHANG L, ZHANG Y, LIU Y L, et al. A novel fitting algorithm for alignment curve radius estimation using corneal elevation data in orthokeratology lens trial[J]. Contact Lens and Anterior Eye, 2017, 40(6):401-407.

[25] SANDERSON E, WINDMEIJER F. A weak instrument F-test in linear IV models with multiple endogenous variables[J]. Journal of Econometrics, 2016, 190(2):212-221.

[26] ŠAŠIĆ S, VERIOTTI T, KOTECKI T, et al. Comparing the predictions by NIR spectroscopy based multivariate models for distillation fractions of crude oils by F-test[J]. Spectrochimica Acta Part A:Molecular and Biomolecular Spectroscopy, 2023,286:122023.

[27] HE H Y, HUANG G Y, ZHANG B, et al. Research on boruta-ET-based anomalous traffic detection model[J]. Security and Communication Networks, 2022,2022:1-8.

[28] HEIDARI A A, AKHOONDZADEH M, CHEN H. A wavelet PM2.5 prediction system using optimized kernel rxtreme learning with Boruta-XGBoost feature selection[J]. Mathematics, 2022, 10(19):3566.

[29] DEVORE G R. Computing the Z score and centiles for cross-sectional analysis:a practical approach[J]. Journal of Ultrasound in Medicine, 2017, 36(3):459-473.

[30] DEMIR S, SAHIN E K. An investigation of feature selection methods for soil liquefaction prediction based on tree-based ensemble algorithms using AdaBoost,gradient boosting,and XGBoost[J]. Neural Computing and Applications, 2023, 35(4):3173-3190.

[31] YIN F M, DU J, XU X Z, et al. Depression detection in speech using transformer and parallel convolutional neural networks[J]. Electronics, 2023, 12(2):328.

[32] CHUNG D, YUN J, LEE J, et al. Predictive model of employee attrition based on stacking ensemble learning[J]. Expert Systems with Applications, 2023,215:119364.

[33] GOH V, CHOU Y J, LEE C C, et al. Predicting bacteremia among septic patients based on ED information by machine learning methods:a comparative study[J]. Diagnostics, 2022, 12(10):2498.

[34] XIA M, WU Y Z, WANG Z J. Time delay identification in dynamical systems based on interpretable machine learning[J]. Journal of Donghua University (English Edition), 2022, 39(4):332-339.

[35] WU J H, WANG X K, WANG J L, et al. Method of soft-sensor modeling for fermentation process based on geometric support vector regression[J]. Journal of Donghua University (English Edition), 2013, 30(1):1-6.

[36] ELHADIDY A A, EL-BADAWY S M, ELBELTAGI E E. A simplified pavement condition index regression model for pavement evaluation[J]. International Journal of Pavement Engineering, 2021, 22(5):643-652.

[37] THELWALL M, WILSON P. Regression for citation data:an evaluation of different methods[J]. Journal of Informetrics, 2014, 8(4):963-971.

[38] PLEBANKIEWICZ E, GRACKI J. Analysis and prediction of universities’ buildings’ renovation costs using a regression model[J]. Applied Sciences, 2022, 13(1):401.

[39] LI G Z, ZHANG A N, ZHANG Q Z, et al. Pearson correlation coefficient-based performance enhancement of broad learning system for stock price prediction[J]. IEEE Transactions on Circuits and Systems II:Express Briefs, 2022, 69(5):2413-2417.

Funding

Shanghai Science and Technology Project, China(20DZ2251400)