I.A. Kuznetsov1

1 Design Information Technologies Center Russian Academy of Sciences (Odintsovo, Moscow oblast, Russia)
1 info@ditc.ras.ru

The emergence of the concept of personalized medicine imposes new requirements for working with patients, which are associated with prediction, personalization and precision. As a result, individualization of the approach to determining treatment tactics to achieve the best result and reduce the risk of complications comes to the fore. The construction of predictive models for medical decision support systems is based on the use of machine learning and artificial intelligence methods, where an important criterion is the presence of trusted decisions that can be interpreted and not presented as a “black box” model. Conducting a comparative analysis of the most effective machine learning methods used in the medical field for subsequent assessment of the possibility of their interpretation for decision-making in predicting perioperative results. The work provides a review and analysis of scientific research in the field of constructing predictive models for implementing the principles of personalized medicine, considers the most popular and relevant machine learning methods, and the metrics used to assess the quality of these methods. A comparative analysis of the interpretability of various machine learning methods was carried out based on data from organ-preserving operations with localized formations of the kidney parenchyma. The ability to substantiate the predicted values from machine learning models significantly increases the confidence of specialists in the results of artificial intelligence, which will make it possible to further integrate this type of decisions into medical decision support systems when determining surgical treatment tactics for patients with localized formations of the renal parenchyma.

Kuznetsov I.A. Modern mathematical methods for building trusted medical decision support systems. Highly Available Systems. 2023. V. 19. № 4. P. 63−72. DOI: https://doi.org/ 10.18127/j20729472-202304-05 (in Russian)

1. Harbediya E.H. et al. Primenenie intellektual'nogo analiza v urologii [Application of intelligent analysis in urology]. Urologiya [Urology]. 3:162-166.
2. Behrad F., Saniee Abadeh M. An overview of deep learning methods for multimodal medical data mining. Expert Systems with Applications. 200:117006.
3. Rebrova O.Yu. ZHiznennyj cikl sistem podderzhki prinyatiya vrachebnyh reshenij kak medicinskih tekhnologij [Life cycle of medical decision support systems as medical technologies]. Vrach i informacionnye tekhnologii [Doctor and information technology]. 1.
4. Sirota, E.S. et al. Analysis of the learning curve in laparoscopic partial nephrectomy in patients with localized renal parenchymal lesions depending on the nephrometric score. Urologiia. 2020. № 6. Р. 11–18. DOI: https://dx.doi.org/10.18565/urology.2020.6.11-18
5. Grebenev F. V. et al. 2022. Prognozirovanie funkcional'nogo iskhoda posle mikrohirurgicheskogo lecheniya nervavshihsya anevrizm mozga s pomoshch'yu mashinnogo obucheniya [Prediction of functional outcome after microsurgical treatment of ruptured brain aneurysms using machine learning]. III Vserossijskaya konferenciya molodyh nejrohirurgov v ramkah Vserossijskogo nejrohirurgiches­kogo foruma: sbornik tezisov: pod redakciej k.m.n. Poluninoj N.A. [III All-Russian Conference of Young Neurosurgeons within the framework of the All-Russian Neurosurgical Forum: collection of abstracts: edited by Ph.D. Polunina N.A.]. Moscow. 64.
6. Bektaş M. et al. Machine Learning Algorithms for Predicting Surgical Outcomes after Colorectal Surgery: A Systematic Reviewю World Journal of Surgery. 2022. 12(46):3100–3110.
7. Nair A.A. et al. Machine learning approach to predict postoperative opioid requirements in ambulatory surgery patients. PLOS ONE. 2020. 7(15):e0236833.
8. Zanier O. et al. Machine learning-based clinical outcome prediction in surgery for acromegaly. Endocrine. 2022. 2(75): 508–515.
9. Jiang H. et al. Machine Learning for the Prediction of Complications in Patients After Mitral Valve Surgery. Frontiers in Cardiovascular Medicine. 2021. (8).
10. Sahara K. et al. Machine learning predicts unpredicted deaths with high accuracy following hepatopancreatic surgery. Hepatobiliary surgery and nutrition. 2021. 1(10):20–30.
11. Ejmalian A. et al. Prediction of Acute Kidney Injury After Cardiac Surgery Using Interpretable Machine Learning. Anesthesiology and Pain Medicine. 2022. 4(12):e127140.
12. Hong L. et al. Prediction of low cardiac output syndrome in patients following cardiac surgery using machine learning. Frontiers in Medicine. 2022. (9).
13. Mekhaldi R. N. et al. Using Machine Learning Models to Predict the Length of Stay in a Hospital Setting by Á. Rocha et al., Cham: Springer International Publishing, 2020. 202–211.
14. Farrokhi F. et al. Investigating Risk Factors and Predicting Complications in Deep Brain Stimulation Surgery with Machine Learning Algorithms. World Neurosurgery. 2020. (134):e325–e338.
15. Gridin V.N. et al. Prediction of perioperative parameters of laparoscopic organ-sparing interventions on the kidney taking into account the surgeon's "learning curve". Biomedical Radioelectronics. 2021. V. 24. № 2. Р. 13–20. DOI: 10.18127/j15604136-202102-02
16. Gebreyesus Y. et al. Machine Learning for Data Center Optimizations: Feature Selection Using Shapley Additive exPlanation (SHAP). Future Internet. 2023; 15(3):88. https://doi.org/10.3390/fi15030088