Suna KOÇ, Mehmet DOKUR, Türkan ÖZER, Betül BÖRKÜ UYSAL, Mehmet Sami İSLAMOĞLU, Nilgün AÇIKGÖZ, İlke KÜPELİ, Sena Gül KOÇ, Sema Nur DOKUR, İsmail Tuncer DEGİM

Yoğun bakım ünitesindeki COVID-19 hastalarının yapay sinir ağı ile tedavi maliyetinin tahmini

Amaç: Yapay sinir ağları (YSA), memeli beyinlerini oluşturan biyolojik sinir ağlarından esinlenen bilgi işlem sistemleridir. Bir YSA, biyolojik bir beyindeki nöronları kabaca modelleyen, yapay nöronlar adı verilen bağlantılı birimler veya düğümler koleksiyonuna dayanır. Her bağlantı, biyolojik bir beyindeki sinapslar gibi, diğer nöronlara bir sinyal iletebilir. Bağlantılara kenar denir. Nöronlar ve kenarlar tipik olarak öğrenme ilerledikçe ayarlanan bir ağırlığa sahiptir. Ağırlık, bir bağlantıdaki sinyalin gücünü artırır veya azaltır. Doğrusal olmayan süreçleri yeniden üretme ve modelleme yetenekleri nedeniyle yapay sinir ağları, birçok disiplinde uygulama bulmuştur. Uygulama alanları arasında sistem tanımlama ve kontrol, tıbbi teşhis, veri madenciliği, görselleştirme, makine çevirisi, yalnızca hücre şekli bilgilerini kullanarak yüksek düzeyde istilacı kanser hücre dizilerini daha az istilacı hücre dizilerinden ayırt etmek ve diğer birçok alan yer alır. Bu çalışmada, yoğun bakım ünitesi (YBÜ)’nde COVID-19 hastasının toplam tedavi maliyetini veya prognozunu tahmin etmek için ANN’den yararlanması amaçlanmıştır. Yöntem: COVID-19 ile enfekte ağır hastaların YBÜ’de kalış sürelerini (gün) etkileyen yaş gibi parametreler ve diğer biyokimyasal parametreler YSA analizi kullanılarak değerlendirilmiştir. Bu çalışmada, bir bilgisayar programı olan Pythia® YSA modellerini geliştirmek için seçilen hastalara ait veriler kullandık. Bulgular: YBÜ’den elde edilen ve başlangıç parametreleri olarak bilgisayara verilen gerçek veriler elde edilmiştir. Bilgisayar programı tahmin için en uygun model olarak birinci seviye için 15 nöron, ikinci seviye için bir nöron vermiştir (SSD = 0,000995). Bu program, gerçek hasta için 2019 yılındaki gerçek maliyeti 142.234,06 TL iken toplam maliyeti 144.930,94 TL olarak tahmin edilmiştir. Bu ilişkinin, kalış süreleri üzerindeki olası etkilenen parametreleri tahmin etmede iyi olduğu görülmüştür. Sonuç: Bu çalışmada, geliştirilen ve yayınlanan YSA modeli herhangi bir deneysel parametre gerektirmemektedir. Ayrıca YSA, YBÜ’deki COVID-19 hastalarının masraflarıyla ilgili yararlı ve kesin tahmin veya bilgi sağlama yeteneğine sahiptir.

Prediction of treatment cost by artificial neural network of patients with COVID-19 in intensive care unit

Objective: Artificial neural networks (ANNs) are computer systems that are inspired by the biological neural networks that make up mammalian brains. An ANN is built from a network of linked units or nodes known as artificial neurons, which are roughly modeled after the neurons in the human brain. Each link, like synapses in a human brain, has the ability to send a signal to other neurons. The connections are referred to as edges. Neurons and edges usually have a weight that changes as learning progresses. The weight changes the intensity of the signal at a connection. Artificial neural networks have found applications in a wide range of fields due to their capacity to recreate and simulate nonlinear phenomena. System identification and control, medical diagnostics, data mining, visualization, machine translation, distinguishing highly invasive cancer cell lines from less invasive lines using simply cell shape information, and many more domains are examples of application areas. In this study, ANN analysis was utilized by us to forecast the total cost of therapy or the prognosis of severe COVID-19 the patients in the intensive care unit (ICU). Methods: The parameters such as ages, and the other biochemical parameters that affect the staying periods (days) of COVID-19 infected patients in ICU were evaluated by using an ANN analysis. For this a computer program, Pythia®, was used to develop ANN models. Real data was used for that selected patients in this study. Results: The real data obtained from the ICU and gave to the computer as initial parameters. The computer program gave 15 neurons for the first level, one neurons for the second level as the most suitable model for the prediction (SSD = 0.000995). This program predicts a total cost 144.930,94 Turkish Lira (27.300 USD) where the real cost 142.234,06 Turkish Lira (26.792 USD) for the real patient in 2019. This relation was found to be good to predict the possible affected parameters on staying times. Conclusion: The ANN model developed and released in this research does not necessitate any experimental parameters. Besides, ANN has the ability to deliver helpful and exact prediction or information regarding the expense of COVID-19 patients in ICU.

PDF

___

1. Değim T, Hadgraft J, Ilbasmiş S, Ozkan Y. Prediction of skin penetration using artificial neural network (ANN) modeling. J Pharm Sci, 2003;92(3):656-64.
2. Aslan C, Çelebi N, Değim IT, Atak A, Özer Ç. Development of interleukin-2 loaded chitosan- based nanogels using artificial neural networks and investigating the effects on wound healing in rats. AAPS Pharm Sci Tech, 2017;18(4):1019-30.
3. Tuncer Degim I. Understanding skin penetration: computer aided modeling and data interpretation. Curr Comput Aided Drug Des, 2005;1(1):11-9.
4. Takayama K, Fujikawa M, Obata Y, Morishita M. Neural network based optimization of drug formulations. Adv Drug Deliv Rev, 2003;55(9):1217-31.
5. Wan Z, Xu Y, Šavija B. On the use of machine learning models for prediction of compressive strength of concrete: influence of dimensionality reduction on the model performance. Materials (Basel). 2021;14(4):713.
6. Artificial neural network. https://en.wikipedia.org/ wiki/Artificial_neural_network#Applications, Date Accessed: 18.03.2022).
7. Patro R, Norel R, Prill RJ, Saez-Rodriguez J, Lorenz P, Steinbeck F, et al. A computational method for designing diverse linear epitopes including citrullinated peptides with desired binding affinities to intravenous immunoglobulin. BMC Bioinform, 2016;17:155.
8. Elhag AA, Aloafi TA, Jawa TM, Sayed-Ahmed N, Bayones FS, Bouslimi J. Artificial neural networks and statistical models for optimization studying COVID-19. Results Phys, 2021;25:104274.
9. Galata DL, Könyves Z, Nagy B, Novák M, Mészáros LA, Szabó E, et al. Real-time release testing of dissolution based on surrogate models developed by machine learning algorithms using NIR spectra, compression force and particle size distribution as input data. Int J Pharm, 2021;597:120338.
10. Larentzakis A, Lygeros N. Artificial intelligence (AI) in medicine as a strategic valuable tool. Pan Afr Med J, 2021;38:184.
11. Karnuta JM, Churchill JL, Haeberle HS, Nwachukwu BU, Taylor SA, Ricchetti ET, et al. The value of artificial neural networks for predicting length of stay, discharge disposition, and inpatient costs after anatomic and reverse shoulder arthroplasty. J Shoulder Elbow Surg, 2020;29(11):2385-94.
12. Kulkarni H, Thangam M, Amin AP. Artificial neural network-based prediction of prolonged length of stay and need for post-acute care in acute coronary syndrome patients undergoing percutaneous coronary intervention. Eur J Clin Invest, 2021;51(3):e13406.
13. Zhao Z, Chen A, Hou W, Graham JM, Li H, Richman PS, et al. Prediction model and risk scores of ICU admission and mortality in COVID-19. PLoS ONE, 2020; 15(7): e0236618.
14. Rigo-Bonnin R, Gumucio-Sanguino VD, Pérez- Fernández XL, Corral-Ansa L, Fuset-Cabanes M, Pons-Serra M, et al. Individual outcome prediction models for patients with COVID-19 based on their first day of admission to the intensive care unit. Clin Biochem, 2022;100:13-21.
15. Corey KM, Kashyap S, Lorenzi E, Lagoo Deenadayalan SA, Heller K, Whalen K, et al. Development and validation of machine learning models to identify high-risk surgical patients using automatically curated electronic health record data (Pythia): a retrospective, single-site study. PLoS Med, 2018; 15(11): e1002701.
16. Arriaga-Pizano LA, Gonzalez-Olvera MA, Ferat- Osorio EA, Escobar J, Hernandez-Perez AL, Revilla- Monsalve C, et al. Accurate diagnosis of sepsis using a neural network: Pilot study using routine clinical variables. Comput Methods Programs Biomed, 2021;210:106366.
17. Chelazzi C, Villa G, Manno A, Ranfagni V, Gemmi E, Romagnoli S. The new SUMPOT to predict postoperative complications using an Artificial Neural Network. Sci Rep, 2021;11(1):22692.
18. Jiang X, Wang Y, Pan Y, Zhang W. Prediction Models for sepsis-associated thrombocytopenia risk in intensive care units based on a machine learning algorithm. Front Med, 2022; 9:837382.