PERSONALIZED MATHEMATICAL MODEL OF OUTCOME IN PATIENTS WITH TRAUMATIC BRAIN INJURIES
Introduction. Previously, determining the prognosis based on epidemiological data was the key to informing about the treatment of patients [1,2], but modern prognostic models based on demographic data, clinical examination, radiological imaging have limited prognostic ability [3,4]. At the same time, a reliable prognosis of the outcome of the disease with modern highly specific and sensitive markers is of great clinical importance [5,6].
Aim. Developing a method for mathematical modeling of the outcome of acute traumatic cerebral injuries based on complex clinical, laboratory, and neuroimaging studies with integral scales for assessing neurological status.
Materials and methods. The studies were conducted in 79 patients with various acute traumatic brain injuries. All patients underwent a detailed clinical and neurological examination using the Glasgow Coma Scale (GCS), computed tomography, X-ray, ultrasound, hematological and biochemical examinations. To identify dependent and independent risk factors for death in the acute period of injury, a one-dimensional and multidimensional regression analysis was performed. To determine the predictive variables, an analysis of the receiver's performance characteristics (ROC) was performed with the calculation of sensitivity and specificity.
The results obtained. We found that the strongest predictors of a poor outcome were - AVDO2 > 52% of the left side (OR) - 9.01 (95% СI: 3.45 - 23.51), p<0.0001; AVDO2 >52%, right side (OR) - 5.71 (95% СI: 2.31-14.16), p=0.0002. Lactate level >3.3 mmol/l (OR) - 4.30 (95% СI: 1.61-11.51), p=0.0036. With an increase in S100ß 0.1 mcg/l> (OR) - 3.77 (95% СI:1.63-8.73), p=0.0020 and NSE ng/ml>12.5 (OR) - 2.69 (95% СI:1.14-6.36), p=0.0240; SAD>169 mmHg (OR) - 3.27 (95% СI:1.26-8.48), p=0.0146; at the age of > 65 years (OR) - 2.43 (95% СI: 1.04-5.68), p=0.0406. The measure of reliability of the model obtained by the criterion of pseudo R2, Nagelkerke - 627.3% and logLikelihood - 112.7.
Сonclusion. These data were used to develop a mathematical model that allows predicting the outcome of the disease. The best predicted value of the model had a cut-off point of 99.13%, AuROC-0.912; Se-93.26%; Sp-80.00%; NPV-94.55%; PPV-76.15%.
Количество просмотров: 1109
Категория статей:
Оригинальное исследование
Библиографическая ссылка
Sarkulova Zh.N., Tokshilykova A.B., Sarkulov M.N., Zhankulov M.B., Kalieva B.M., Daniyarova K.R., Zhuldyzbaeva Zh.Zh. Personalized mathematical model of outcome in patients with traumatic brain injuries // Nauka i Zdravookhranenie [Science & Healthcare]. 2024. Vol.26 (4), pp. 92-98. doi 10.34689/SH.2024.26.4.012Похожие публикации:
EPIDEMIOLOGY, RESOURCE DISTRIBUTION, AND MANAGEMENT CHALLENGES OF CHRONIC HEART FAILURE IN KAZAKHSTAN: A NATIONAL RETROSPECTIVE ANALYSIS
PREVALENCE OF POLYMORPHISMS IN WARFARIN METABOLISM-RELATED GENES AMONG THE KAZAKH POPULATION
NEUROLOGICAL COMPLICATIONS IN EXTRACORPOREAL MEMBRANE OXYGENATION: RISK FACTORS FOR THE DEVELOPMENT OF ACUTE CEREBROVASCULAR EVENTS
THE EFFECT OF REPAGLINIDE AND METFORMIN ON EXTERNAL RESPIRATORY FUNCTION IN PATIENTS WITH DIABETES AND ISCHEMIC HEART DISEASE ASSOCIATED WITH OBESITY
RETROSPECTIVE ANALYSIS OF PATIENTS WITH OBLITERATING ATHEROSCLEROSIS OF THE LOWER EXTREMITY ARTERIES IN THE ABAI REGION OF THE REPUBLIC OF KAZAKHSTAN