COVID-19 Hastalarında Biyokimyasal ve Hematolojik Değişiklikler: Denetimsiz Kümeleme Analizi

Öz

Amaç: Koronavirüs Hastalığı 2019 (COVID-19), biyokimyasal ve hematolojik belirteçlerde anlamlı değişimlere neden olmaktadır; ancak mevcut çalışmaların büyük çoğunluğu yalnızca tek bir zaman noktasındaki ölçümlere odaklanmaktadır. Bu çalışma, hastanede yatan COVID-19 hastalarında kan biyobelirteçlerindeki zamansal değişimlerin prognostik önemini araştırmakta ve K-means ile Gaussian Mixture Model (GMM) kümeleme algoritmalarının, bu dinamik örüntülere dayanarak klinik açıdan anlamlı hasta alt gruplarını belirlemedeki etkinliğini değerlendirmektedir. Gereç ve Yöntemler: Hastanede yatan 120 COVID-19 hastası üzerinde retrospektif bir analiz yapıldı. Hastalar klinik sonuçlarına göre (şifa ile taburcu veya ileri tedavi için sevk) gruplandırıldı. Seçilen kan parametrelerindeki değişimler (yatış ve taburculuk) hesaplandı ve K-means ile GMM kullanılarak kümeleme analizine tabi tutuldu. İki boyutlu görselleştirme için Temel Bileşenler Analizi (PCA) uygulandı. Kümelemenin iç geçerliliği Silhouette skoru ve Davies-Bouldin indeksi ile değerlendirildi. Bulgular: Her iki kümeleme yöntemi, farklı biyobelirteç seyirleri ve klinik sonuçlara sahip üç hasta alt grubu tanımladı. K-means Küme 2 ve GMM Küme 1, CRP'de belirgin düşüş ve minimal lökosit artışı ile karakterize edilen %100 iyileşme oranı gösterdi. K-means, iç geçerlilik metriklerinde GMM'den daha iyi performans sergiledi (Silhouette: 0,191 vs. 0,161; Davies-Bouldin: 1,61 vs. 1,78). Özellikle, CRP düzeylerinde yetersiz düşüş ile birlikte yüksek lökosit seviyeleri gösteren hasta kümeleri, daha olumsuz iyileşme süreçleriyle ilişkilendirildi. Sonuç: Kan biyobelirteçlerindeki zamansal değişimler, COVID-19'da prognostik değer taşımakta ve sonuç tahminine katkı sağlayabilmektedir. K-means kümelemesi daha iyi performans göstermiş olsa da, GMM karmaşık veri kümeleri için geçerli bir alternatif olmaya devam etmektedir. Biyobelirteç dinamiklerinin denetimsiz kümeleme analizi, COVID-19 tedavisinde daha kişiselleştirilmiş ve bilinçli tedavi stratejilerinin geliştirilmesine yardımcı olabilir.

Anahtar Kelimeler

• COVID-19
• biyobelirteçler
• kümeleme analizi

Biochemical and Hematological Changes in COVID-19 Patients: Unsupervised Clustering Analysis

Öz

Objective: The Coronavirus Disease 2019 (COVID-19) causes notable changes in biochemical and hematological markers, yet most studies have focused on single-timepoint measurements. This study investigates the prognostic significance of temporal changes in blood biomarkers among hospitalized COVID-19 patients and evaluates the effectiveness of K-means and Gaussian Mixture Model (GMM) clustering algorithms in identifying clinically relevant patient subgroups based on these dynamic patterns. Material and methods: A retrospective analysis was conducted on 120 hospitalized COVID-19 patients. Patients were grouped based on clinical outcome—discharge in recovery or referral for advanced care. Changes in selected blood parameters between admission and discharge were calculated and used for unsupervised clustering via K-means and GMM. Principal Component Analysis (PCA) was applied for two-dimensional visualization. Internal validation of clustering was assessed using the Silhouette score and Davies–Bouldin index. Results: Both clustering methods identified three patient subgroups with distinct biomarker trajectories and clinical outcomes. K-means Cluster 2 and GMM Cluster 1 demonstrated 100% recovery rates, marked by a significant drop in CRP and minimal leukocyte increase. K-means showed slightly better internal validity metrics than GMM (Silhouette: 0.191 vs. 0.161; Davies–Bouldin: 1.61 vs. 1.78). Notably, inadequate CRP reduction and elevated leukocyte counts were associated with poorer recovery. Conclusions: Temporal changes in blood biomarkers were associated with clinical outcomes in hospitalized COVID-19 patients and may support patient stratification. K-means clustering showed slightly better performance, though GMM remains a viable alternative for complex datasets. Unsupervised clustering of biomarker dynamics may support more personalized and informed treatment strategies in COVID-19 care.

Anahtar Kelimeler

• COVID-19
• biomarkers
• cluster analysis

Kaynakça

1. Park SE. Epidemiology, virology, and clinical features of severe acute respiratory syndrome - coronavirus-2 (SARS-CoV-2; Coronavirus Disease19). Clin Exp Pediatr. 2020;63(4):119–24. doi:10.3345/cep.2020.00493
2. Zhai P, Ding Y, Wu X, Long J, Zhong Y, Li Y. The epidemiology, diagnosis and treatment of COVID-19. Int J Antimicrob Agents. 2020;55(5):105955. doi:10.1016/j.ijantimicag.2020.105955
3. Wan S, Xiang Y, Fang W, Zheng Y, Li B, Hu Y, et al. Clinical features and treatment of COVID-19 patients in northeast Chongqing. J Med Virol. 2020; 92(7):797–806. doi:10.1002/jmv.25783
4. Zhang G, Zhang J, Wang B, Zhu X, Wang Q, Qiu S. Analysis of clinical characteristics and laboratory findings of 95 cases of 2019 novel coronavirus pneumonia in Wuhan, China: a retrospective analysis. Respir Res. 2020;21:74. doi:10.1186/s12931-020-01338-8
5. Lippi G, Plebani M. Laboratory abnormalities in patients with COVID-2019 infection. Clin Chem Lab Med. 2020;58(7):1131–4. doi:10.1515/cclm-2020-0198
6. Yang R, Gui X, Zhang Y, Xiong Y. The role of essential organ-based comorbidities in the prognosis of COVID-19 infection patients. Expert Rev Respir Med. 2020;14(8):835–8. doi:10.1080/17476348.2020.1761791
7. Cobre AF, Stremel DP, Noleto GR, Fachi MM, Surek M, Wiens A, et al. Diagnosis and prediction of COVID-19 severity: Can biochemical tests and machine learning be used as prognostic indicators? Comput Biol Med. 2021;134:104531. doi:10.1016/j.compbiomed.2021.104531
8. Ji X, Guo Y, Tang L, Gao C. Identifying and validating prognostic hyper-inflammatory and hypo-inflammatory COVID-19 clinical phenotypes using machine learning methods. J Inflamm Res. 2025;18:3009-24. doi:10.2147/JIR.S504028

9. Lau K-Y, Ng K-S, Kwok K-W, Tsia K-M, Sin C-F, Lam C-W, et al. An unsupervised machine learning clustering and prediction of differential clinical phenotypes of COVID-19 patients based on blood tests—A Hong Kong population study. Front Med (Lausanne). 2022;8:764934. doi:10.3389/fmed.2021.764934
10. Nalinthasnai N, Thammasudjarit R, Tassaneyasin T, Eksombatchai D, Sungkanuparph S, Boonsarngsuk V, et al. Unsupervised machine learning clustering approach for hospitalized COVID-19 pneumonia patients. BMC Pulm Med. 2025;25:70. doi:10.1186/s12890-025-03536-w
11. MacQueen J. Some methods for classification and analysis of multivariate observations. In: Proceedings of the Fifth Berkeley Symposium on Mathematical Statistics and Probability. Vol. 1. Berkeley: University of California Press, 1967: 281–97.
12. Davies DL, Bouldin DW. A cluster separation measure. IEEE Trans Pattern Anal Mach Intell. 1979;1(2):224–7. doi:10.1109/TPAMI.1979.4766909
13. Rousseeuw PJ. Silhouettes: A graphical aid to the interpretation and validation of cluster analysis. J Comput Appl Math. 1987;20:53–65. doi:10.1016/0377-0427(87)90125-7
14. Reynolds DA. Gaussian mixture models. In: Li SZ, Jain AK, eds. Encyclopedia of Biometrics. Boston, MA: Springer; 2009. p.659–63. doi:10.1007/978-0-387-73003-5_196
15. Bishop CM. Pattern recognition and machine learning. New York: Springer; 2006.
16. Duda RO, Hart PE, Stork DG. Pattern classification. 2nd ed. Hoboken, NJ: Wiley; 2000.
17. Schwarz GE. Estimating the dimension of a model. Ann Stat. 1978;6(2):461–4. doi:10.1214/aos/1176344136
18. Jolliffe IT, Cadima J. Principal component analysis: a review and recent developments. Philos Trans A Math Phys Eng Sci. 2016;374(2065):20150202. doi:10.1098/rsta.2015.0202
19. Deng G, Yin M, Chen X, Zeng F. Clinical determinants for fatality of 44,672 patients with COVID-19. Crit Care. 2020;24:179. doi:10.1186/s13054-020-02902-w
20. Zhou F, Yu T, Du R, Fan G, Liu Y, Liu Z, et al. Clinical course and risk factors for mortality of adult inpatients with COVID-19 in Wuhan, China: a retrospective cohort study. Lancet. 2020;395:1054–62. doi:10.1016/S0140-6736(20)30566-3
21. Sproston NR, Ashworth JJ. Role of C-reactive protein at sites of inflammation and infection. Front Immunol. 2018;9:754. doi:10.3389/fimmu.2018.00754
22. Tan C, Huang Y, Shi F, Tan K, Ma Q, Chen Y, et al. C-reactive protein correlates with computed tomographic findings and predicts severe COVID-19 early. J Med Virol. 2020;92(7):856–62. doi:10.1002/jmv.25871
23. Liu Y, Du X, Chen J, Jin Y, Peng L, Wang HHX, et al. Neutrophil-to-lymphocyte ratio as an independent risk factor for mortality in hospitalized patients with COVID-19. J Infect. 2020;81(1):e6–12. doi:10.1016/j.jinf.2020.04.002
24. Zhang C, Shi L, Wang FS. Liver injury in COVID-19: management and challenges. Lancet Gastroenterol Hepatol. 2020;5(5):428–30. doi:10.1016/S2468-1253(20)30057-1
25. Olry A, Meunier L, Délire B, Larrey D, Horsmans Y, Le Louët H. Drug-induced liver injury and COVID-19 infection: The rules remain the same. Drug Saf. 2020;43:615–7. doi:10.1007/s40264-020-00954-z
26. Mehta P, McAuley DF, Brown M, Sanchez E, Tattersall RS, Manson JJ. COVID-19: consider cytokine storm syndromes and immunosuppression. Lancet. 2020;395(10229):1033–4. doi:10.1016/S0140-6736(20)30628-0
27. Chen N, Zhou M, Dong X, Qu J, Gong F, Han Y, et al. Epidemiological and clinical characteristics of 99 cases of 2019 novel coronavirus pneumonia in Wuhan, China: a descriptive study. Lancet. 2020;395(10223):507–13. doi:10.1016/S0140-6736(20)30211-7
28. Jain AK. Data clustering: 50 years beyond K-means. Pattern Recognit Lett. 2010;31(8):651–66. doi:10.1016/j.patrec.2009.09.011
29. Fraley C, Raftery AE. Model-based clustering, discriminant analysis, and density estimation. J Am Stat Assoc. 2002;97(458):611–31. doi:10.1198/016214502760047131