YOĞUN BAKIM HASTALARINDA SEPSİS ENFEKSİYONU TAHMİNİ: DENETİMSİZ MAKİNE ÖĞRENMESİ MODELİ

Yoğun bakım servislerinde yaşanan ölümlerin en önemli sebeplerinden biri olan sepsis enfeksiyonu ciddi bir küresel sağlık krizi olarak görülmektedir. Sepsis enfeksiyonunun erken teşhisi yapılamaz ve hızla tedaviye başlanmaz ise çoklu organ yetmezliğine ve ölüme neden olabilmektedir. Bu nedenle hızlı sepsis tanısı ve tedavisi hayati önem taşımaktadır. Bu çalışmada amaç, sepsis enfeksiyonunun gelişimini etkileyen önemli parametrelerden olan laktat ve Ph değerlerini kullanarak yeni bir denetimsiz makine öğrenmesi modeli gerçekleştirmektir. Çalışma kapsamında kullanılan veriler MIMIC-III klinik veri tabanından elde edilmiştir. Çalışma genelinde sepsis tanısı konmuş ve sepsis tanısı konmamış hastalar üzerinde Bulanık-C ortalamalar algoritması ile denetimsiz makine öğrenmesi gerçekleştirilmiştir. Makine sepsis olan ve olamayan hastaları beşi sepsis pozitif, beşi sepsis negatif olacak şekilde 10 ayrı etiketle işaretlemiştir. Etiketlenen küme temsilcileri öğrenmenin monitorize edilebilmesi için Temel Bileşenler Analizi yöntemiyle iki boyuta indirgenmiştir. Çalışma, iki parametre özelinde (laktat ve Ph) değerlendirilerek denetimsiz öğrenme gerçekleştirmiş olması ve çok parametreli çalışmalara öncülük etmesi açısından literatüre katkı sağlamaktadır. Ayrıca, çalışma Lactat ve Ph değerleri bakımından beş ayrı kümede hasta bulunduğunu rapor etmektedir.

AN EARLY PREDICTION AND DIAGNOSIS OF SEPSIS IN INTENSIVE CARE UNITS: AN UNSUPERVİSED MACHINE LEARNING MODEL

Sepsis infection, which is one of the most important causes of death in intensive care units, is seen as a severe globalhealth crisis. If an early diagnosis of sepsis infection cannot be made, and treatment is not started rapidly, septic shockmay result in multiple organ failure and death is almost inevitable. Therefore, it is vital to establish an early diagnosisand start the treatment at once. This study aims to accomplish a new model of unsupervised machine learning usinglactate and Ph laboratory test values, which are considered to be important parameters to diagnose sepsis infection. Thedata used in the study have been obtained from MIMIC-III international clinical database. Unsupervised machine learninghas been performed via the Fuzzy-C algorithm along with validity indexes like Xie Beni on patients’ data diagnosed sepsisand non-sepsis. The machine-generated ten labels at the end of the training session considering-designed validity indexes.The labelled cluster representatives have been reduced to two dimensions by Principal Component Analysis method inorder to monitor the learning in a two-dimensional space. The study contributes to the literature by conductingunsupervised learning through two parameters (Lactate and Ph) and leading to multi-parameter studies. In addition, thestudy reports that there are five types of sepsis patterns in terms of Lactate and PH laboratory tests.

PDF

___

[1] The Global Competitiveness Report 2017-2018 | World Economic Forum.
[2] Silahtaroğlu, G. and Yılmaztürk, N. (2019) Data analysis in health and big data: A machine learning medical diagnosis model based on patients’ complaints. Communications in Statistics - Theory and Methods, Taylor & Francis. 1–10. https://doi.org/10.1080/03610926.2019.16227 28
[3] Yadav, P., Steinbach, M., Kumar, V. and Simon, G. (2018) Mining electronic health records (EHRs): A survey. ACM Comput. Surv. https://doi.org/10.1145/3127881
[4] Kocbek, S., Cavedon, L., Martinez, D., Bain, C., Manus, C. Mac, Haffari, G. et al. (2016) Text mining electronic hospital records to automatically classify admissions against disease: Measuring the impact of linking data sources. Journal of Biomedical Informatics, 64. https://doi.org/10.1016/j.jbi.2016.10.008
[5] de Bruijne, M. (2016) Machine learning approaches in medical image analysis: From detection to diagnosis. Med. Image Anal. https://doi.org/10.1016/j.media.2016.06.032
[6] Sakar, B.E., Isenkul, M.E., Sakar, C.O., Sertbas, A., Gurgen, F., Delil, S. et al. (2013) Collection and analysis of a Parkinson speech dataset with multiple types of sound recordings. IEEE Journal of Biomedical and Health Informatics,. https://doi.org/10.1109/JBHI.2013.2245674
[7] Sepsis — Global Sepsis Alliance.
[8] Bone, R. C., Balk, R. A., Cerra, F. B., Dellinger, R. P., Fein, A. M., Knaus, W. A., ... & Sibbald, W. (1992) Definitions for sepsis and organ failure and guidelines for the use of innovative therapies in sepsis. Chest, 101, 1644–55. https://doi.org/10.1378/chest.101.6.1644
[9] Com, I. (2018) (No Title).
[10] Szakmany, T., Lundin, R.M., Sharif, B., Ellis, G., Morgan, P., Kopczynska, M. et al. (2016) Sepsis prevalence and outcome on the general wards and emergency departments in Wales: Results of a multi-centre, observational, point prevalence study. PLoS ONE,. https://doi.org/10.1371/journal.pone.0167230
[11] Lovejoy, C.A., Buch, V. and Maruthappu, M. (2019) Artificial intelligence in the intensive care unit. Crit. Care. https://doi.org/10.1186/s13054-018-2301-9
[12] Nemati, S., Holder, A., Razmi, F., Stanley, M.D., Clifford, G.D. and Buchman, T.G. (2018) An Interpretable Machine Learning Model for Accurate Prediction of Sepsis in the ICU. Critical Care Medicine,. https://doi.org/10.1097/CCM.00000000000029 36
[13] Broder, G., Science, M.W.- and 1964, undefined. Excess lactate: an index of reversibility of shock in human patients. ScienceSciencemagOrg,.
[14] Callaway, D.W., Shapiro, N.I., Donnino, M.W., Baker, C. and Rosen, C.L. (2009) Serum lactate and base deficit as predictors of mortality in normotensive elderly blunt trauma patients. Journal of Trauma - Injury, Infection and Critical Care, 66, 1040–4. https://doi.org/10.1097/TA.0b013e3181895e9 e
[15] Vandromme, M.J., Griffin, R.L., Weinberg, J.A., Rue, L.W. and Kerby, J.D. (2010) Lactate is a better predictor than systolic blood pressure for determining blood requirement and mortality: could prehospital measures improve trauma triage? Journal of the American College of Surgeons, 210, 861–7, 867–9. https://doi.org/10.1016/j.jamcollsurg.2010.01. 012
[16] Dede, G., Şahan, L., Dede, B. and Demirbilek, S. Araştırma Makalesi Kan Laktat Seviyesi Yoğun Bakım Hastalarında Mortaliteyi Tahmin Etmede Ne Kadar Etkilidir? Blood Lactate Levels Intensive Care Patients Mortality Estimating, How much?
[17] Ağır Sepsiste Santral Venöz, Arteriyel ve Periferik Venöz Kan Gazı Değerlerinin Karşılaştırılması | Makale | Türkiye Klinikleri.
[18] Nguyen, H.B., Rivers, E.P., Knoblich, B.P., Jacobsen, G., Muzzin, A., Ressler, J.A. et al. (2004) Early lactate clearance is associated with improved outcome in severe sepsis and septic shock. Critical Care Medicine,. https://doi.org/10.1097/01.CCM.0000132904.3 5713.A7
[19] Bouch, C.D. and Thompson, J.P. (2008) Severity scoring systems in the critically ill. Continuing Education in Anaesthesia, Critical Care and Pain, Oxford University Press. 8, 181–5. https://doi.org/10.1093/bjaceaccp/mkn033
[20] Delahanty, R.J., Alvarez, J.A., Flynn, L.M., Sherwin, R.L. and Jones, S.S. (2019) Development and Evaluation of a Machine Learning Model for the Early Identification of Patients at Risk for Sepsis. Annals of Emergency Medicine, Mosby Inc. 73, 334–44. https://doi.org/10.1016/j.annemergmed.2018.1 1.036
[21] Johnson, A.E.W., Pollard, T.J., Shen, L., Lehman, L.W.H., Feng, M., Ghassemi, M. et al. (2016) MIMIC-III, a freely accessible critical care database. Scientific Data, Nature Publishing Groups. 3. https://doi.org/10.1038/sdata.2016.35
[22] Dunn, J.C. (1973) A fuzzy relative of the ISODATA process and its use in detecting compact well-separated clusters. Journal of Cybernetics,. https://doi.org/10.1080/01969727308546046
[23] Bezdek, J.C., Ehrlich, R. and Full, W. (1984) FCM: The fuzzy c-means clustering algorithm. Computers and Geosciences,. https://doi.org/10.1016/0098-3004(84)90020- 7
[24] Flores-Sintas, A., Cadenas, J., Systems, F.M.-F.S. and and 1999, undefined. Membership functions in the fuzzy C-means algorithm. Elsevier,.
[25] Xie, X.L. and Beni, G. (1991) A validity measure for fuzzy clustering. IEEE Transactions on Pattern Analysis and Machine Intelligence, 13, 841–7. https://doi.org/10.1109/34.85677