Predicting Bone Cement Implantation Syndrome Using Artificial Intelligence Methods

Condition(s) studied

Investigational drug(s) / intervention(s)

Collecting data from patients who underwent arthroplasty and cement was used: Collecting data that may change in cement reaction before, during and after surgery from patients who underwent arthroplasty and cement use and taught it to artificial intelligence.

Study summary

This study aims to predict the development of bone cement syndrome, which may develop due to polymethylmethacrylate cement used to adhere the prosthesis to the bone in arthroplasty surgeries, which may cause intraoperative and postoperative mortality and morbidity, using artificial intelligence methods and to provide a sustainable life comfort to patients in the postoperative period with the standardization predicted in the long term.

Eligibility

Primary outcome measure(s)

• collection of data-1 — 2 months
  General examination findings of the patient in the perioperative period, ASA score, whether the patient is female whether male or female, body mass index, age, whether mobility is decreased, whether assisted living, basic life parameters (operation blood pressure, pulse, respiratory rate, saturation, temperature, consciousness, orientation, etc.), the cause of the fracture (trauma, arthrosis, infection, necrosis, osteolysis, intertrochanteric fracture), type of surgery (total hip replacement, partial hip replacement, total knee replacement or revision surgery), medical history and medical history (cardiovascular disease, kidney disease, diabetes, stroke, peripheral vascular disease, arteriosclerosis, pulmonary hypertension, angina pectoris, congestive heart failure, COPD, cancer, dementia, arrhythmia, nicotine dependence, and the presence of previous myocardial infarction.
• collection of data-2 — 2 months
  β-adrenergic blockers, diuretics, antiplatelet drugs, organic nitrates, calcium antagonists, ACE inhibitors, ARBs, acetylsalicylic acid, insulin, and warfarin and statin use), radiological evaluations (chest radiography), laboratory parameters (D-Dimer, IMA, IL-6, C-Peeptit, HbA1c, glucose, insulin, Ig E, NT Pro BNP, homocysteine, serotonin, adrenaline, noradrenaline, D vit, CK, CKMB, troponin, myoglobin, urea, uric acid, creatinine, AST, ALT, BIL, ALB, GGT, ALP, LDH, PT, PTT, INR, Sedim, CRP, fibrinogen, procalcitonin, TFT, HDL, LDL, TG, cholesterol, electrolytes and calcium, Mg, CA125, CA15-3, CA19 -9, afp, CEA, hemogram, 5HT-2A receptor, GFR) and anaesthesia and surgery-related parameters (type of anaesthesia; general anaesthesia, local anesthesia) anaesthesia, volume of epidural anaesthetic used, type of local anaesthetic used, opioid use, duration of surgery, duration of anaesthesia, total blood loss
• collection of data-3 — 2 months
  IMA measured at certain intervals after cementing. troponin-T measured at intervals, duration of hypotension, duration of hypoxia, medullary lavage, use of cement gun, vacuum cement Mixing and cementing pressure have been reported to be risk factors for the development of BCIS; these data will be collected throughout the study. When and which parameters will be performed will be determined by experienced physicians, and their results will be analysed according to their clinical experience. are interpreted by the algorithms they create.
• artificial intelligence modeling — 1 months
  Artificial Intelligence model process steps The collected data will be made ready for artificial intelligence algorithms by going through the data preprocessing stage. In the feature extraction stage, simple feature extraction algorithms, kurtosis, skewness, local maximum, local minimum, hyperparameters, and principal component analysis processes will be applied. In the classification stage, a dataset is assigned to one of the predetermined classes that are different from each other. Classification algorithms learn which data to assign to which class from the given training set. Then it tries to assign the test data to the correct classes. Logistic regression, linear discriminant analysis, decision trees, Naive Bayes, the K-nearest neighbour algorithm, support vector machines, and random forest algorithms will be used in the classification phase. In the evaluation phase, the best model will be selected by creating an error matrix and interpreting the accuracy value of the model.

Trial sites (1)

More Ataturk University trials in Turkey