Machine learning based prediction of perioperative blood loss in orthognathic surgery

State of the art and prospects for artificial intelligence in orthognathic surgery: A systematic review with meta-analysis

OBJECTIVE

To present a systematic review of the state of the art regarding clinical applications, main features, and outcomes of artificial intelligence (AI) in orthognathic surgery.

METHODS

The PICOS strategy was performed on a systematic review (SR) to answer the following question: "What are the state of the art, characteristics and outcomes of applications with artificial intelligence for orthognathic surgery?" After registering in PROSPERO (CRD42021270789) a systematic search was performed in the databases: PubMed (including MedLine), Scopus, Embase, LILACS, MEDLINE EBSCOHOST and Cochrane Library. 195 studies were selected, after screening titles and abstracts, of which thirteen manuscripts were included in the qualitative analysis and six in the quantitative analysis. The treatment effects were plotted in a Forest-plot. JBI questionnaire for observational studies was used to asses the risk of bias. The quality of the SR evidence was assessed using the GRADE tool.

RESULTS

AI studies on 2D cephalometry for orthognathic surgery, the Tau2 = 0.00, Chi2 = 3.78, p = 1.00 and I² of 0 %, indicating low heterogeneity, AI did not differ statistically from control (p = 0.79). AI studies in the diagnosis of the decision of whether or not to perform orthognathic surgery showed heterogeneity, and therefore meta-analysis was not peformed.

CONCLUSION

The outcome of AI is similar to the control group, with a low degree of bias, highlighting its potential for use in various applications.

Using machine learning to predict bleeding after cardiac surgery

OBJECTIVES

The primary objective was to predict bleeding after cardiac surgery with machine learning using the data from the Australia New Zealand Society of Cardiac and Thoracic Surgeons Cardiac Surgery Database, cardiopulmonary bypass perfusion database, intensive care unit database and laboratory results.

METHODS

We obtained surgical, perfusion, intensive care unit and laboratory data from a single Australian tertiary cardiac surgical hospital from February 2015 to March 2022 and included 2000 patients undergoing cardiac surgery. We trained our models to predict either the Papworth definition or Dyke et al.'s universal definition of perioperative bleeding. Our primary outcome was the performance of our machine learning algorithms using sensitivity, specificity, positive and negative predictive values, accuracy, area under receiver operating characteristics curve (AUROC) and area under precision-recall curve (AUPRC).

RESULTS

Of the 2000 patients undergoing cardiac surgery, 13.3% (226/2000) had bleeding using the Papworth definition and 17.2% (343/2000) had moderate to massive bleeding using Dyke et al.'s definition. The best-performing model based on AUPRC was the Ensemble Voting Classifier model for both Papworth (AUPRC 0.310, AUROC 0.738) and Dyke definitions of bleeding (AUPRC 0.452, AUROC 0.797).

CONCLUSIONS

Machine learning can incorporate routinely collected data from various datasets to predict bleeding after cardiac surgery.

Con: Artificial Intelligence-Derived Algorithms to Guide Perioperative Blood Management Decision Making

Artificial intelligence has the potential to improve the care that is given to patients; however, the predictive models created are only as good as the base data used in their design. Perioperative blood management presents a complex clinical conundrum in which significant variability and the unstructured nature of the required data make it difficult to develop precise prediction models. There is a potential need for training clinicians to ensure they can interrogate the system and override when errors occur. Current systems created to predict perioperative blood transfusion are not generalizable across clinical settings, and there is a considerable cost implication required to research and develop artificial intelligence systems that would disadvantage resource-poor health systems. In addition, a lack of strong regulation currently means it is difficult to prevent bias.

Applications of artificial intelligence and machine learning in orthognathic surgery: A scoping review

Over the recent years, Artificial Intelligence (AI) has been progressing rapidly with its ability to mimic human cognitive functions. The potential applications of AI technology in diagnosis, treatment planning, and prognosis prediction have been demonstrated in various studies. The present scoping review aimed to provide an overview of AI and Machine Learning (ML) algorithms and their applications in orthognathic surgery. A comprehensive search was conducted in databases including PubMed, Embase, Scopus, Web of Science and OVID Medline until November 2021. This scoping review was conducted following the PRISMA-ScR guidelines. After applying the inclusion and exclusion criteria, a total of 19 studies were included for final review. AI has profoundly impacted the diagnosis and prediction of orthognathic surgeries with a clinically acceptable accuracy range. Furthermore, AI reduces the work burden of the clinician by eliminating the tedious registration procedures, thereby helping in efficient and automated planning. However, focussing on the research gaps, there is a need to foster the AI models/algorithms to contemporize their efficiency in clinical decision making, diagnosis and surgical planning in future studies.

Performance of Artificial Intelligence Models Designed for Diagnosis, Treatment Planning and Predicting Prognosis of Orthognathic Surgery (OGS)—A Scoping Review

The technological advancements in the field of medical science have led to an escalation in the development of artificial intelligence (AI) applications, which are being extensively used in health sciences. This scoping review aims to outline the application and performance of artificial intelligence models used for diagnosing, treatment planning and predicting the prognosis of orthognathic surgery (OGS). Data for this paper was searched through renowned electronic databases such as PubMed, Google Scholar, Scopus, Web of science, Embase and Cochrane for articles related to the research topic that have been published between January 2000 and February 2022. Eighteen articles that met the eligibility criteria were critically analyzed based on QUADAS-2 guidelines and the certainty of evidence of the included studies was assessed using the GRADE approach. AI has been applied for predicting the post-operative facial profiles and facial symmetry, deciding on the need for OGS, predicting perioperative blood loss, planning OGS, segmentation of maxillofacial structures for OGS, and differential diagnosis of OGS. AI models have proven to be efficient and have outperformed the conventional methods. These models are reported to be reliable and reproducible, hence they can be very useful for less experienced practitioners in clinical decision making and in achieving better clinical outcomes.

Development and Validation of a Nomogram to Predict the Risk of Blood Transfusion in Orthognathic Patients

OBJECTIVE

This study aims to establish a nomogram to predict the probability of blood transfusion in patients with preoperative autologous blood donation before orthognathic surgery.

METHODS

The authors conducted a retrospective case-control study on consecutive orthognathic patients with preoperative autologous blood donation from January 2014 to December 2020. The outcome variable was the actual transfusion of autologous blood (ATAB). Predictors included patients' demographics, preoperative blood cell test, vital signs, American Society of Anesthesiologists classification, surgical procedure, operation duration, and blood loss. Univariable and multivariable logistic regressions were performed to identify independent risk factors associated with ATAB. A nomogram was constructed to predict the risk for ATAB. The performance of the nomogram was evaluated using the area under the receiver operating characteristic curve, calibration curve and the consistency index.

RESULTS

A total of 142 patients (75 males and 67 females) with an average age of 22.72 ± 5.34 years donated autologous blood before their orthognathic surgery. Patients in the transfusion group (n = 56) had significantly lower preoperative red blood cell counts (4.74 ± 0.55 × 109/L versus 4.98 ± 0.45 × 109/L, P = 0.0063), hemoglobin (141.48 ± 15.18 g/dL versus 150.33 ± 14.73 g/dL, P = 0.0008), and hematocrit (41.05% ± 4.03% versus 43.32% ± 3.42%, P = 0.0006), more bimaxillary osteotomies (92.86% versus 56.98%, P < 0.001), longer operation duration (348.4 ± 111.10 minutes versus 261.6 ± 115.44 minutes, P < 0.001), and more intraoperative blood loss (629.23 ± 273.06 ml versus 359.53 ± 222.84 ml, P < 0.001) than their counterparts (n = 86) in the non-transfusion group. Univariable and multivariable logistic regression demonstrated that only hemoglobin (adjusted odds ratio [OR] 0.864, 95% confidence interval [CI]:0.76-0.98, P = 0.026), operation procedures (adjusted OR 8.14, 95% CI:1.69-39.16, P = 0.009), and blood loss (adjusted OR 1.006, 95% CI:1.002-1.009, P < 0.001) were independent risk factors for ATAB. The area under the receiver operating characteristic curve of the nomogram was 0.823. The consistency index of the nomogram was 0.823. The calibration curve illustrated that the nomogram was highly consistent with the actual observation.

CONCLUSIONS

The nomogram is a simple and useful tool with good accuracy and performance in predicting the risk for blood transfusion.

Gender-specific differences in haemostatic parameters and their influence on blood loss in bimaxillary surgery

Objective

The objectives of this prospective cohort study were to establish gender-related differences in blood loss and haemostatic profiles associated with bimaxillary surgery. In addition, we aimed to identify if any gender differences could be established which might help predict blood loss volume.

Materials and methods

Fifty-four patients (22 males; 32 females) undergoing bimaxillary surgery for skeletal dentofacial deformities were eligible for inclusion. Blood samples were taken 1 day preoperatively and 48 h postoperatively for detailed gender-specific coagulation analysis incorporating global coagulation assays (endogenous thrombin potential) and specific coagulation parameters. Blood loss was measured at two different time points: (1) the end of surgery, visible intraoperative blood loss (IOB) using ‘subtraction method’; and (2) 48 h postoperatively perioperative bleeding volume (CBL-48 h) using ‘haemoglobin-balance method’ and Nadler’s formula. Correlation and regression analyses were performed to identify relevant parameters affecting the amount of blood loss.

Results

Significant differences in IOB and CBL-48 h were observed (p < 0.001). Men had higher IOB versus women, lacking statistical significance (p = 0.056). In contrast, men had significantly higher CLB-48 h (p = 0.019). Reduced CBL-48 h was shown to be most closely associated with the level of Antithrombin-III being decreased in females.

Conclusions

Male gender is associated with higher IOB and CBL-48 compared with females. Gender does not affect IOB regarding haemostatic profile but does correlate strongly with procedure length. Conversely, CBL-48 is closely associated with gender-specific imbalances in the anticoagulant system.

Clinical relevance

Knowledge of gender-related differences will help clinicians establish predictive factors regarding excessive blood loss in orthognathic surgery and identify at-risk patients.

Bone marrow space volume of the mandible influencing intraoperative blood loss in bilateral sagittal split osteotomy: A pilot Study

The aim of this study was to investigate whether the bone marrow space volume of the mandible affects blood loss during bilateral sagittal split osteotomy (BSSO). Sixteen patients who underwent BSSO in our hospital were included in this study. Bone marrow space volume of the mandible was measured by analyzing images from computed tomography. Blood loss during BSSO was measured by weighing gauze, measuring suctioned blood, and adjusting for the volume of irrigation solution used during BSSO. Mean blood loss during BSSO for the 16 patients was 200.5 ml, and patients were divided into: Group I, with less than mean blood loss; and Group II, with greater than mean blood loss. Total bone marrow space volume was significantly greater in Group II (12,450.7 ± 2644.3 mm³) than in Group I (9130.3 ± 3005.8 mm³; P<0.05). A correlation between bone marrow space volume and blood loss during BSSO was suggested, and these results are beneficial for surgeons planning and preparing the orthognathic surgery.

Prediction model of laparoendoscopic single-site surgery in gynecology using machine learning algorithm

Introduction

Minimally invasive surgery has been widely used in gynecology. The laparoendoscopic single-site surgery (LESS) risk prediction model can provide evidence-based references for preoperative surgical procedure selection.

Aim

To determine whether the patients are suitable for LESS and to provide guidance for the clinical operation plan, we aimed to compare the clinical outcomes of LESS and conventional laparoscopic surgery (CLS) in gynecology. We constructed a LESS risk prediction model and predicted surgical conditions for the preoperative evaluation system.

Material and methods

A retrospective analysis was carried out among patients undergoing LESS (n = 1019) and CLS (n = 1055). Various clinical indicators were compared. Multiple machine model algorithms were evaluated. The optimal results were chosen as the model to form the risk prediction model.

Results

The LESS group showed advantages in the postoperative 12/24 h visual analog scale and Vancouver scar score compared with the CLS group (p < 0.05). The comparisons in other clinical indicators between the two groups showed that each group had advantages and the difference was statistically significant (p < 0.05), including operative time, estimated blood loss, and hospital stay. We evaluated the predictive value for various models using AUC values of 0.77, 0.77, 0.76, and 0.67 for XGBoost, random forest, GBDT, and logistic regression, respectively. The decision tree model was shown to be the optimal model.

Conclusions

LESS can reduce postoperative pain, shorten hospital stay and make scars acceptable. The risk prediction model based on a machine learning algorithm has manifested a high degree of accuracy and can satisfy the doctors’ demand for individualized preoperative evaluation and surgical safety in LESS.

Perioperative Coagulation Monitoring

The main goal of perioperative coagulation monitoring is to improve safety of patients undergoing surgical procedures. Various conditions can affect the coagulation system during surgery and bleeding. The value of traditional standard coagulation tests is limited in detecting hemostatic dysfunctions and they are particularly ineffective in diagnosing hyperfibrinolysis. This article reports on key issues and pathophysiologic changes that affect the hemostatic system in the perioperative setting. Values of preoperative coagulation tests are discussed and the basic principles for point-of-care coagulation devices, including platelet analyzers and their clinical use, are evaluated.

Determination of blood loss in bimaxillary surgery: does the formula and the time point affect results?

The amount of blood loss determined in orthognathic surgery differs greatly among studies. This can be attributed to the inhomogeneity in study cohorts analysed, but may also be a result of the varying methodologies used for blood loss determination. However, this has yet to be explored. Thus, the aim of this study was to investigate the extent to which the formula and time point used to measure blood loss affect the blood loss volume, determined in a homogeneous cohort undergoing bimaxillary surgery. Blood loss was calculated at 24 and 48 hours postoperatively using the haemoglobin balance method and the formula of Hurle et al. The estimated total blood volume was established based on the formulae of Nadler et al. and Choi et al. Differences in blood loss volume with respect to time point and formula were analysed and compared. Fifty-four patients were included in the final analysis. Statistically significant differences in blood loss were observed: a significant increase in the blood loss volume from 24 hours to 48 hours postoperatively was detected. When comparing the formulae used, blood loss differed significantly at 24 hours after surgery; however no such difference resulted at 48 hours postoperatively. These findings imply that the time point of measuring blood loss is highly relevant, whereas the formulae applied seem to have less of an impact on the blood loss volumes calculated.

Machine learning and orthodontics, current trends and the future opportunities: A scoping review

INTRODUCTION

In recent years, artificial intelligence (AI) has been applied in various ways in medicine and dentistry. Advancements in AI technology show promising results in the practice of orthodontics. This scoping review aimed to investigate the effectiveness of AI-based models employed in orthodontic landmark detection, diagnosis, and treatment planning.

METHODS

A precise search of electronic databases was conducted, including PubMed, Google Scholar, Scopus, and Embase (English publications from January 2010 to July 2020). Quality Assessment and Diagnostic Accuracy Tool 2 (QUADAS-2) was used to assess the quality of the articles included in this review.

RESULTS

After applying inclusion and exclusion criteria, 49 articles were included in the final review. AI technology has achieved state-of-the-art results in various orthodontic applications, including automated landmark detection on lateral cephalograms and photography images, cervical vertebra maturation degree determination, skeletal classification, orthodontic tooth extraction decisions, predicting the need for orthodontic treatment or orthognathic surgery, and facial attractiveness. Most of the AI models used in these applications are based on artificial neural networks.

CONCLUSIONS

AI can help orthodontists save time and provide accuracy comparable to the trained dentists in diagnostic assessments and prognostic predictions. These systems aim to boost performance and enhance the quality of care in orthodontics. However, based on current studies, the most promising application was cephalometry landmark detection, skeletal classification, and decision making on tooth extractions.

Machine Learning and Surgical Outcomes Prediction: A Systematic Review

BACKGROUND

Machine learning (ML) has garnered increasing attention as a means to quantitatively analyze the growing and complex medical data to improve individualized patient care. We herein aim to critically examine the current state of ML in predicting surgical outcomes, evaluate the quality of currently available research, and propose areas of improvement for future uses of ML in surgery.

METHODS

A systematic review was conducted in accordance with the Preferred Reporting Items for a Systematic Review and Meta-Analysis (PRISMA) checklist. PubMed, MEDLINE, and Embase databases were reviewed under search syntax "machine learning" and "surgery" for papers published between 2015 and 2020.

RESULTS

Of the initial 2677 studies, 45 papers met inclusion and exclusion criteria. Fourteen different subspecialties were represented with neurosurgery being most common. The most frequently used ML algorithms were random forest (n = 19), artificial neural network (n = 17), and logistic regression (n = 17). Common outcomes included postoperative mortality, complications, patient reported quality of life and pain improvement. All studies which compared ML algorithms to conventional studies which used area under the curve (AUC) to measure accuracy found improved outcome prediction with ML models.

CONCLUSIONS

While still in its early stages, ML models offer surgeons an opportunity to capitalize on the myriad of clinical data available and improve individualized patient care. Limitations included heterogeneous outcome and imperfect quality of some of the papers. We therefore urge future research to agree upon methods of outcome reporting and require basic quality standards.

How Do Machines Learn? Artificial Intelligence as a New Era in Medicine

With an increased number of medical data generated every day, there is a strong need for reliable, automated evaluation tools. With high hopes and expectations, machine learning has the potential to revolutionize many fields of medicine, helping to make faster and more correct decisions and improving current standards of treatment. Today, machines can analyze, learn, communicate, and understand processed data and are used in health care increasingly. This review explains different models and the general process of machine learning and training the algorithms. Furthermore, it summarizes the most useful machine learning applications and tools in different branches of medicine and health care (radiology, pathology, pharmacology, infectious diseases, personalized decision making, and many others). The review also addresses the futuristic prospects and threats of applying artificial intelligence as an advanced, automated medicine tool.

Deep learning based discrimination of soft tissue profiles requiring orthognathic surgery by facial photographs

Facial photographs of the subjects are often used in the diagnosis process of orthognathic surgery. The aim of this study was to determine whether convolutional neural networks (CNNs) can judge soft tissue profiles requiring orthognathic surgery using facial photographs alone. 822 subjects with dentofacial dysmorphosis and / or malocclusion were included. Facial photographs of front and right side were taken from all patients. Subjects who did not need orthognathic surgery were classified as Group I (411 subjects). Group II (411 subjects) was set up for cases requiring surgery. CNNs of VGG19 was used for machine learning. 366 of the total 410 data were correctly classified, yielding 89.3% accuracy. The values of accuracy, precision, recall, and F1 scores were 0.893, 0.912, 0.867, and 0.889, respectively. As a result of this study, it was found that CNNs can judge soft tissue profiles requiring orthognathic surgery relatively accurately with the photographs alone.

Is there a hidden blood loss in orthognathic surgery and should it be considered? Results of a prospective cohort study

The aim of this prospective observational study was to investigate the parameter 'hidden blood loss' (HBL) in the context of orthognathic surgery, incorporating undetected bleeding volumes occurring intra- and postoperatively. Orthognathic bleeding volumes were recorded at three different time points. At the end of the operation the visible intraoperative blood loss (VBL) was measured. Additionally, the perioperative blood loss was calculated 24 h and 48 h postoperatively using the 'haemoglobin balance method'. Analysis of the HBL was based on the difference between the visible intraoperative blood loss (VBL) and calculated blood loss (CBL), determined 48 h after surgery. 82 patients (male 33, female 49) were included in this study, of whom 41 underwent bimaxillary surgery and of whom 41 underwent Bilateral Sagittal Split Osteotomy (BSSO). Statistically significant differences with reference to the absolute bleeding volumes were found when comparing the two treatment modalities. In terms of HBL, a bleeding volume of 287.2 ml (±265.9) in the bimaxillary group and 346.9 ml (±271.3) in the BSSO cohort was recorded. This accounted for 32.2% (bimaxillary surgery) and 62.6% (BSSO) of the CBL after 48 h (BIMAX vs. BSSO, p < 0.001). HBL is a valuable adjunct to record within the perioperative management of orthognathic surgery to further improve patient safety and postoperative outcomes.