HYBRID ANOVA-SHAP APPROACH FOR MODEL-AGNOSTIC FEATURE SELECTION IN GALLSTONE DISEASE PREDICTION

Edi Jaya Kusuma; Ririn Nurmandhani; Yusthin Meriantti Manglapy; Ika Pantiawati; Evina Widianawati

doi:10.24252/instek.v11i1.66622

Authors

Edi Jaya Kusuma Universitas Dian Nuswantoro
Ririn Nurmandhani Universitas Dian Nuswantoro
Yusthin Meriantti Manglapy Universitas Dian Nuswantoro
Ika Pantiawati Universitas Dian Nuswantoro
Evina Widianawati Chung Yuan Christian University

DOI:

https://doi.org/10.24252/instek.v11i1.66622

Keywords:

ANOVA, Clinical Prediction, Feature Selection, Gallstone, Model-Agnostic Approach, SHAP

Abstract

Penyakit batu empedu merupakan kondisi yang cukup umum terjadi, namun sering kali tidak terdeteksi hingga muncul komplikasi, sehingga pentingnya deteksi dini menjadi semakin penting. Penelitian ini bertujuan untuk mengusulkan pendekatan seleksi fitur hibrid yang bersifat model-agnostik dengan menggabungkan metode statistik (ANOVA) dan explainable machine learning (multi-model SHAP) untuk meningkatkan kinerja prediksi batu empedu sekaligus tetap menjaga interpretabilitas model. Kerangka yang diusulkan diuji menggunakan dataset klinis yang terdiri dari 319 pasien dengan 38 variabel, yang mencakup data demografis, bioimpedansi, dan parameter laboratorium. Proses seleksi fitur dilakukan melalui beberapa konfigurasi bobot (\alpha=0.3,\ 0.5,\ 0.7) guna melihat keseimbangan antara kontribusi statistik dan berbasis model. Hasil penelitian menunjukkan bahwa fitur seperti C-Reactive Protein (CRP), Vitamin D, Bone Mass, serta parameter fungsi hati secara konsisten berperan dalam prediksi, yang menggambarkan keterkaitan antara inflamasi, metabolisme, dan fungsi hati dalam pembentukan batu empedu. Performa terbaik diperoleh pada \alpha=0.7 khususnya metode Gradient Boosting dengan akurasi 0.8646 dan recall 0.875, sedangkan \alpha=0.5 menghasilkan nilai AUC tertinggi sebesar 0.905. Secara keseluruhan, pendekatan ini memberikan solusi yang cukup andal dan mudah dipahami untuk mendukung deteksi dini.

Downloads

Download data is not yet available.

References

[1] M. K. Azzami, A. Rastono, and B. Adhayati, “A Case Control Study on The Relationship Between Bodey Mass Index as a Risk Factors for Incidence of Cholelithiasis at RSUD Banten,” Int. J. Med. Public Health, vol. 2, no. 1, pp. 31–42, 2025.

[2] J. R. Gallaher and A. Charles, “Acute Cholecystitis A Review,” JAMA, vol. 327, no. 10, p. 965, Mar. 2022, doi: 10.1001/jama.2022.2350.

[3] X. Wang et al., “Global Epidemiology of Gallstones in the 21st Century: A Systematic Review and Meta-Analysis,” Clinical Gastroenterology and Hepatology, vol. 22, no. 8, pp. 1586–1595, Aug. 2024, doi: 10.1016/j.cgh.2024.01.051.

[4] A. Unalp-Arida and C. E. Ruhl, “Increasing gallstone disease prevalence and associations with gallbladder and biliary tract mortality in the US,” Hepatology, vol. 77, no. 6, pp. 1882–1895, Jun. 2023, doi: 10.1097/HEP.0000000000000264.

[5] H. Hendarto, F. N. Akbar, J. B. Muzakki, R. A. Amri, S. N. A. Nugraha, and H. Adlani, “Obesity, dyslipidemia, and diabetes mellitus as risk factors in cholelithiasis,” Electronic Journal of General Medicine, vol. 20, no. 6, p. em549, Nov. 2023, doi: 10.29333/ejgm/13814.

[6] C. W. Candra, I. K. Sudartana, I. W. G. A. E. Putra, K. P. Yasa, I. M. Mahayasa, and K. D. Ariyanta, “Risk Factors Affecting the Occurrence of Gallstones at Prof. Dr. IGNG Visits General Hospital Denpasar Bali,” International Journal Of Scientific Advances, vol. 5, no. 3, 2024, doi: 10.51542/ijscia.v5i3.12.

[7] Alchamdani and A. S. Anas, “Urban Obesity in Transition: Socioeconomic, Lifestyle, and Environmental Drivers in Jakarta, Indonesia,” Medicor: Journal of Health Informatics and Health Policy, vol. 2, no. 2, pp. 113–124, Apr. 2024, doi: 10.61978/medicor.v2i2.748.

[8] I. Esen, H. Arslan, S. Aktürk Esen, M. Gülsen, N. Kültekin, and O. Özdemir, “Early prediction of gallstone disease with a machine learning-based method from bioimpedance and laboratory data,” Medicine, vol. 103, no. 8, p. e37258, Feb. 2024, doi: 10.1097/MD.0000000000037258.

[9] P. Sarker, J.-J. Tiang, and A.-A. Nahid, “Gallstone Classification Using Random Forest Optimized by Sand Cat Swarm Optimization Algorithm with SHAP and DiCE-Based Interpretability,” Sensors, vol. 25, no. 17, p. 5489, Sep. 2025, doi: 10.3390/s25175489.

[10] K. Shrestha, P. Sarker, J.-J. Tiang, and A.-A. Nahid, “Metaheuristic-based gallstone classification using rotational forest explained with SHAP,” Front. Digit. Health, vol. 7, Jan. 2026, doi: 10.3389/fdgth.2025.1727559.

[11] E. J. Kusuma, I. Pantiawati, and S. Handayani, “Melanoma Classification based on Simulated Annealing Optimization in Neural Network,” Knowledge Engineering and Data Science, vol. 4, no. 2, p. 97, Mar. 2022, doi: 10.17977/um018v4i22021p97-104.

[12] A. Das, S. Mollick, H. Mondal, T. J. Bala, A. Nag, and A. Guho, “Meta-heuristic Algorithms for High-Dimensional Feature Selection: A Systematic Review of Methodologies, Applications, and Emerging Challenges with Future Research Directions,” in Feature Fusion for Next-Generation AI, 1st ed., Md. M. Anindya Nag and A. K. B. Hassan, Eds., Springer Cham, 2026, pp. 51–61. doi: 10.1007/978-3-031-94386-7_5.

[13] E. J. Kusuma, R. Nurmandhani, I. Pantiawati, Y. M. Manglapy, and E. Widianawati, “A Random Forest and SMOTE-Based Machine Learning Model for Predicting Recurrence in Papillary Thyroid Carcinoma,” Jurnal Teknik Informatika (Jutif), vol. 6, no. 4, pp. 2019–2034, Aug. 2025, doi: 10.52436/1.jutif.2025.6.4.4854.

[14] V. Bhaskar P, V. R, B. Rao K, E. L. Devi, S. Saripalle, and R. Dhumpati, “An Interpretable Stacking Ensemble Model AB-CBLC for High-Accuracy Lung Cancer Detection Using Clinical and Behavioral Data,” in 2025 8th International Conference on Computing Methodologies and Communication (ICCMC), IEEE, Jul. 2025, pp. 1580–1587. doi: 10.1109/ICCMC65190.2025.11140773.

[15] D. V. Artamonov et al., “Interpretable Machine Learning with SHAP Identifies Key Biomarkers in a Multi-Factorial Spectrum of Age-Related Neurological and Metabolic Conditions,” Int. J. Mol. Sci., vol. 27, no. 4, p. 1805, Feb. 2026, doi: 10.3390/ijms27041805.

[16] E. J. Kusuma, R. Nurmandhani, L. Aryani, I. Pantiawati, and G. F. Shidik, “Optimasi Model Extreme Gradient Boosting Dalam Upaya Penentuan Tingkat Risiko Pada Ibu Hamil Berbasis Bayesian Optimization (BOXGB),” Jurnal Teknologi Informasi dan Ilmu Komputer, vol. 12, no. 1, pp. 111–120, Feb. 2025, doi: 10.25126/jtiik.20251219001.

[17] J. Lotsch and A. Ultsch, “A model-agnostic framework for dataset-specific selection of missing value imputation methods in pain-related numerical data,” Canadian Journal of Pain, vol. 10, no. 1, Dec. 2026, doi: 10.1080/24740527.2025.2595160.

[18] S. S. Ayon et al., “Explainable AI framework for improved Thalassemia mental health classification and feature selection,” PLoS One, vol. 21, no. 1, p. e0341168, Jan. 2026, doi: 10.1371/journal.pone.0341168.

[19] I. Maruotto, F. K. Ciliberti, P. Gargiulo, and M. Recenti, “Feature Selection in Healthcare Datasets: Towards a Generalizable Solution,” Comput. Biol. Med., vol. 196, p. 110812, Sep. 2025, doi: 10.1016/j.compbiomed.2025.110812.

[20] K. Zhang et al., “Machine learning model for differentiating xanthogranulomatous cholecystitis and gallbladder cancer in multicenter largescale study,” NPJ Digit. Med., vol. 8, no. 1, p. 590, Oct. 2025, doi: 10.1038/s41746-025-01991-7.

[21] J. Chen, Z.-T. Liu, J.-T. Lyu, and G.-P. Jiang, “Impact of metabolic disorders on gallstone disease and perioperative recovery after laparoscopic cholecystectomy,” Hepatobiliary & Pancreatic Diseases International, vol. 23, no. 6, pp. 604–612, Dec. 2024, doi: 10.1016/j.hbpd.2024.08.001.

[22] I. F. Amri, F. H. N. Rohim, M. I. Nurul Azka, and M. S. Rakhmawati, “Analysis of Gallstone Incidence Factors Using a Logistic Regression Model.,” EKSAKTA: Journal of Sciences and Data Analysis, Oct. 2025, doi: 10.20885/EKSAKTA.vol6.iss2.art3.

[23] X. Feng et al., “MAFLD mediates the association between CHR and gallstones in the U.S. adults: evidence from NHANES 2021–2023,” BMC Gastroenterol., vol. 25, no. 1, p. 268, Apr. 2025, doi: 10.1186/s12876-025-03805-2.