Predictive models for fentanyl dose requirement and postoperative pain using clinical and genetic factors in patients undergoing major breast surgery

Abstract

Fentanyl exhibits interindividual variability in its dose requirement due to various nongenetic and genetic factors such as single nucleotide polymorphisms (SNPs). This study aims to develop and cross-validate robust predictive models for postoperative fentanyl analgesic requirement and other related outcomes in patients undergoing major breast surgery. Data regarding genotypes of 10 candidate SNPs, cold pain test (CPT) scores, pupillary response to fentanyl (PRF), and other common clinical characteristics were recorded from 257 patients undergoing major breast surgery. Predictive models for 24-hour fentanyl requirement, 24-hour pain scores, and time for first analgesic (TFA) in the postoperative period were developed using 4 different algorithms: generalised linear regression model, linear support vector machine learning (SVM-Linear), random forest (RF), and Bayesian regularised neural network. The variant genotype of OPRM1 (rs1799971) and higher CPT scores were associated with higher 24-hour postoperative fentanyl consumption, whereas higher PRF and history of hypertension were associated with lower fentanyl requirement. The variant allele of COMT (rs4680) and higher CPT scores were associated with 24-hour postoperative pain scores. The variant genotype of CTSG (rs2070697), higher intraoperative fentanyl use, and higher CPT scores were associated with significantly lower TFA. The predictive models for 24-hour postoperative fentanyl requirement, pain scores, and TFA had R-squared values of 0.313 (SVM-Linear), 0.434 (SVM-Linear), and 0.532 (RF), respectively. We have developed and cross-validated predictive models for 24-hour postoperative fentanyl requirement, 24-hour postoperative pain scores, and TFA with satisfactory performance characteristics and incorporated them in a novel web application.

Artificial Intelligence and Machine Learning in Cancer Related Pain: A Systematic Review

“…These models accurately predicted odynophagia during lung cancer RT, revealing their effectiveness in pain prediction during the course of treatment for lung cancers. Studies conducted by Sun et al (2023) [32], Juwara et al (2020) [33], Lotsch et al (2017) [23], Lotsch et al (2018) [36], Sipila et al (2012) [44], Lotsch et al (2018) [34] and Wang et al (2021) [27], investigated various ML models for persistent/chronic pain prediction and features identification in post-surgery in breast cancer patients. Studies by Sun et al (2023) [32] and Juwara et al (2020) [33] revealed that the novel ML approaches, including RF, GBM, and XGBoost, exhibited higher performance in predicting Chronic Postsurgical Pain (CPSP) over the traditional regression models.…”

“…• Cancer pain prediction • Post cancer treatment pain prediction • Cancer pain analgesics requirement Random Forest (RF) [24], [32], [33], [34], [26], [35], [27], [36], [37], [16], [10], [38], [39], [40], [41] An ensemble approach that combines the output of multiple decision trees to reach a single high accurate result. It provides a good predictive performance, low overfitting, and easy interpretability.…”

“…Generalized linear mixed models (GLMMs), least absolute shrinkage and selection operator (Lasso), Linear regression (LR), Logistic regression (LR), Bayesian regression [42], [32], [43], [33], [34], [35], [27], [44], [45], [46], [39], [41], [31] Model that estimates the relationship between one dependent variable and one or more independent variables using a line.…”

“…Neural Networks (NN) [33], [25], [34], [26], [27], [28], [51], [52], [16], [10], [46], [38], [53], [54], [55], [56], [31] Composed of node layers, containing an input layer, one or multiple hidden layers and an output layer. Could be supervised, unsupervised or semi-supervised.…”

“…This is an opportunity that future studies should take towards more comprehensive models in cancer pain prediction. Five studies used video recording, machine signals or tests results as input data of the models [23, 34,40,52,54]. Text mining and NLP open the opportunities for future studies in cancer pain using texts and different non-structured data, Numerous studies have explored the power of AI and ML to predict pain in various medical diseases.…”