Artificial intelligence and automation in endoscopy and surgery

A surgical activity model of laparoscopic cholecystectomy for co-operation with collaborative robots

BACKGROUND

Laparoscopic cholecystectomy is a very frequent surgical procedure. However, in an ageing society, less surgical staff will need to perform surgery on patients. Collaborative surgical robots (cobots) could address surgical staff shortages and workload. To achieve context-awareness for surgeon-robot collaboration, the intraoperative action workflow recognition is a key challenge.

METHODS

A surgical process model was developed for intraoperative surgical activities including actor, instrument, action and target in laparoscopic cholecystectomy (excluding camera guidance). These activities, as well as instrument presence and surgical phases were annotated in videos of laparoscopic cholecystectomy performed on human patients (n = 10) and on explanted porcine livers (n = 10). The machine learning algorithm Distilled-Swin was trained on our own annotated dataset and the CholecT45 dataset. The validation of the model was conducted using a fivefold cross-validation approach.

RESULTS

In total, 22,351 activities were annotated with a cumulative duration of 24.9 h of video segments. The machine learning algorithm trained and validated on our own dataset scored a mean average precision (mAP) of 25.7% and a top K = 5 accuracy of 85.3%. With training and validation on our dataset and CholecT45, the algorithm scored a mAP of 37.9%.

CONCLUSIONS

An activity model was developed and applied for the fine-granular annotation of laparoscopic cholecystectomies in two surgical settings. A machine recognition algorithm trained on our own annotated dataset and CholecT45 achieved a higher performance than training only on CholecT45 and can recognize frequently occurring activities well, but not infrequent activities. The analysis of an annotated dataset allowed for the quantification of the potential of collaborative surgical robots to address the workload of surgical staff. If collaborative surgical robots could grasp and hold tissue, up to 83.5% of the assistant's tissue interacting tasks (i.e. excluding camera guidance) could be performed by robots.

SimCol3D - 3D reconstruction during colonoscopy challenge

Colorectal cancer is one of the most common cancers in the world. While colonoscopy is an effective screening technique, navigating an endoscope through the colon to detect polyps is challenging. A 3D map of the observed surfaces could enhance the identification of unscreened colon tissue and serve as a training platform. However, reconstructing the colon from video footage remains difficult. Learning-based approaches hold promise as robust alternatives, but necessitate extensive datasets. Establishing a benchmark dataset, the 2022 EndoVis sub-challenge SimCol3D aimed to facilitate data-driven depth and pose prediction during colonoscopy. The challenge was hosted as part of MICCAI 2022 in Singapore. Six teams from around the world and representatives from academia and industry participated in the three sub-challenges: synthetic depth prediction, synthetic pose prediction, and real pose prediction. This paper describes the challenge, the submitted methods, and their results. We show that depth prediction from synthetic colonoscopy images is robustly solvable, while pose estimation remains an open research question.

Optimizing cystoscopy and TURBT: enhanced imaging and artificial intelligence

Diagnostic cystoscopy in combination with transurethral resection of the bladder tumour are the standard for the diagnosis, surgical treatment and surveillance of bladder cancer. The ability to inspect the bladder in its current form stems from a long chain of advances in imaging science and endoscopy. Despite these advances, bladder cancer recurrence and progression rates remain high after endoscopic resection. This stagnation is a result of the heterogeneity of cancer biology as well as limitations in surgical techniques and tools, as incomplete resection and provider-specific differences affect cancer persistence and early recurrence. An unmet clinical need remains for solutions that can improve tumour delineation and resection. Translational advances in enhanced cystoscopy technologies and artificial intelligence offer promising avenues to overcoming the progress plateau.

Recent Advances in Digital Technology in Implant Dentistry

Digital technology has emerged as a transformative tool in dental implantation, profoundly enhancing accuracy and effectiveness across multiple facets, such as diagnosis, preoperative treatment planning, surgical procedures, and restoration delivery. The multiple integration of radiographic data and intraoral data, sometimes with facial scan data or electronic facebow through virtual planning software, enables comprehensive 3-dimensional visualization of the hard and soft tissue and the position of future restoration, resulting in heightened diagnostic precision. In virtual surgery design, the incorporation of both prosthetic arrangement and individual anatomical details enables the virtual execution of critical procedures (e.g., implant placement, extended applications, etc.) through analysis of cross-sectional images and the reconstruction of 3-dimensional surface models. After verification, the utilization of digital technology including templates, navigation, combined techniques, and implant robots achieved seamless transfer of the virtual treatment plan to the actual surgical sites, ultimately leading to enhanced surgical outcomes with highly improved accuracy. In restoration delivery, digital techniques for impression, shade matching, and prosthesis fabrication have advanced, enabling seamless digital data conversion and efficient communication among clinicians and technicians. Compared with clinical medicine, artificial intelligence (AI) technology in dental implantology primarily focuses on diagnosis and prediction. AI-supported preoperative planning and surgery remain in developmental phases, impeded by the complexity of clinical cases and ethical considerations, thereby constraining widespread adoption.

Role of endoscopy in hepatology

The growing and evolving field of EUS and advanced hepatobiliary endoscopy has amplified traditional upper gastrointestinal endoscopy and unveiled novel options for remaining unsolved hepatobiliary issues, both diagnostically and therapeutically. This conceptually appealing and fascinating integration of endoscopy within the practice of hepatology is referred to as 'endo-hepatology'. Endo-hepatology focuses on the one hand on disorders of the liver parenchyma and liver vasculature and of the hepatobiliary tract on the other hand. Applications hanging under the umbrella of endohepatology involve amongst others EUS-guided liver biopsy, EUS-guided portal pressure measurement, EUS-guided portal venous blood sampling, EUS-guided coil & glue embolization of gastric varices and spontaneous portosystemic shunts as well as ERCP in the challenging context of (decompensated cirrhosis) and intraductal cholangioscopy for primary sclerosing cholangitis. Although endoscopic proficiency however does not necessarily equal in an actual straightforward end-solution for currently persisting (complex) hepatobiliary situations. Therefore, endohepatology continues to generate high-quality data to validate and standardize procedures against currently considered (best available) "golden standards" while continuing to search and trying to provide novel minimally invasive solutions for persisting hepatological stalemate situations. In the current review, we aim to critically appraise the status and potential future directions of endo-hepatology.

Systematic analysis of levels of evidence supporting Chinese clinical practice guidelines for gastrointestinal disease

BACKGROUND

Clinical practice guidelines (CPGs) inform healthcare decisions and improve patient care. However, an evaluation of guidelines on gastrointestinal diseases (GIDs) is lacking. This study aimed to systematically analyze the level of evidence (LOE) supporting Chinese CPGs for GIDs.

METHODS

CPGs for GIDs were identified by systematically searching major databases. Data on LOEs and classes of recommendations (CORs) were extracted. According to the Grades of Recommendation, Assessment, Development, and Evaluation system, LOEs were categorized as high, moderate, low, or very low, whereas CORs were classified as strong or weak. Statistical analyses were conducted to determine the distribution of LOEs and CORs across different subtopics and assess changes in evidence quality over time.

FINDINGS

Only 27.9% of these recommendations were supported by a high LOE, whereas approximately 70% were strong recommendations. There was a significant disparity among different subtopics in the proportion of strong recommendations supported by a high LOE. The number of guidelines has increased in the past 5 years, but there has been a concomitant decline in the proportion of recommendations supported by a high LOE.

CONCLUSIONS

There is a general lack of high-quality evidence supporting Chinese CPGs for GIDs, and there are inconsistencies in strong recommendations that have not improved. This study identified areas requiring further research, emphasizing the need to bridge these gaps and promote the conduct of high-quality clinical trials.

FUNDING

This study was supported by the National Key R&D Program of China (2022YFC2503604 and 2022YFC2503605) and Special Topics in Military Health Care (22BJZ25).

Tracking and mapping in medical computer vision: A review

As computer vision algorithms increase in capability, their applications in clinical systems will become more pervasive. These applications include: diagnostics, such as colonoscopy and bronchoscopy; guiding biopsies, minimally invasive interventions, and surgery; automating instrument motion; and providing image guidance using pre-operative scans. Many of these applications depend on the specific visual nature of medical scenes and require designing algorithms to perform in this environment. In this review, we provide an update to the field of camera-based tracking and scene mapping in surgery and diagnostics in medical computer vision. We begin with describing our review process, which results in a final list of 515 papers that we cover. We then give a high-level summary of the state of the art and provide relevant background for those who need tracking and mapping for their clinical applications. After which, we review datasets provided in the field and the clinical needs that motivate their design. Then, we delve into the algorithmic side, and summarize recent developments. This summary should be especially useful for algorithm designers and to those looking to understand the capability of off-the-shelf methods. We maintain focus on algorithms for deformable environments while also reviewing the essential building blocks in rigid tracking and mapping since there is a large amount of crossover in methods. With the field summarized, we discuss the current state of the tracking and mapping methods along with needs for future algorithms, needs for quantification, and the viability of clinical applications. We then provide some research directions and questions. We conclude that new methods need to be designed or combined to support clinical applications in deformable environments, and more focus needs to be put into collecting datasets for training and evaluation.

Endoscopic procedures in hepatology: Current trends and new developments

Gastrointestinal endoscopy has long been a reliable backbone in the diagnosis and management of hepatobilary disorders and their complications. However, with evolving non-invasive testing, personalised medicine has reframed the utility and necessity of endoscopic screening. Conversely, the growing interest and use of endoscopic ultrasound (EUS) and advanced endoscopy within gastrointestinal units has also opened novel diagnostic and therapeutic avenues for patients with various hepatobiliary diseases. The integration of "advanced endoscopy" within the practice of hepatology is nowadays referred to as "endo-hepatology". In essence, endo-hepatology consists of two pillars: one focusing primarily on disorders of the liver parenchyma, vascular disorders, and portal hypertension, which is mainly captured via EUS, while the other targets the hepatobiliary tract via endoscopic retrograde cholangiopancreatography and advanced imaging. Applications under the umbrella of endo-hepatology include, amongst others, EUS-guided liver biopsy, EUS-guided portal pressure gradient measurement, coil and glue embolisation of gastric varices as well as cholangioscopy. As such endo-hepatology could become an attractive concept wherein advanced endoscopy might reinforce the medical management of patients with hepatobiliary disorders and their complications after initial basic work-up. In this review, we discuss current trends and future developments within endo-hepatology and the remaining hurdles to overcome.

Pediatric endoscopy: how can we improve patient outcomes and ensure best practices?

INTRODUCTION

Strategies to promote high-quality endoscopy in children require consensus around pediatric-specific quality standards and indicators. Using a rigorous guideline development process, the international Pediatric Endoscopy Quality Improvement Network (PEnQuIN) was developed to support continuous quality improvement efforts within and across pediatric endoscopy services.

AREAS COVERED

This review presents a framework, informed by the PEnQuIN guidelines, for assessing endoscopist competence, granting procedural privileges, audit and feedback, and for skill remediation, when required. As is critical for promoting quality, PEnQuIN indicators can be benchmarked at the individual endoscopist, endoscopy facility, and endoscopy community levels. Furthermore, efforts to incorporate technologies, including electronic medical records and artificial intelligence, into endoscopic quality improvement processes can aid in creation of large-scale networks to facilitate comparison and standardization of quality indicator reporting across sites.

EXPERT OPINION

PEnQuIN quality standards and indicators provide a framework for continuous quality improvement in pediatric endoscopy, benefiting individual endoscopists, endoscopy facilities, and the broader endoscopy community. Routine and reliable measurement of data, facilitated by technology, is required to identify and drive improvements in care. Engaging all stakeholders in endoscopy quality improvement processes is crucial to enhancing patient outcomes and establishing best practices for safe, efficient, and effective pediatric endoscopic care.

Intelligent Phase Contrast Meta-Microscope System

Phase contrast imaging techniques enable the visualization of disparities in the refractive index among various materials. However, these techniques usually come with a cost: the need for bulky, inflexible, and complicated configurations. Here, we propose and experimentally demonstrate an ultracompact meta-microscope, a novel imaging platform designed to accomplish both optical and digital phase contrast imaging. The optical phase contrast imaging system is composed of a pair of metalenses and an intermediate spiral phase metasurface located at the Fourier plane. The performance of the system in generating edge-enhanced images is validated by imaging a variety of human cells, including lung cell lines BEAS-2B, CLY1, and H1299 and other types. Additionally, we integrate the ResNet deep learning model into the meta-microscope to transform bright-field images into edge-enhanced images with high contrast accuracy. This technology promises to aid in the development of innovative miniature optical systems for biomedical and clinical applications.

A Temporal Learning Approach to Inpainting Endoscopic Specularities and Its Effect on Image Correspondence

Video streams are utilised to guide minimally-invasive surgery and diagnosis in a wide range of procedures, and many computer-assisted techniques have been developed to automatically analyse them. These approaches can provide additional information to the surgeon such as lesion detection, instrument navigation, or anatomy 3D shape modelling. However, the necessary image features to recognise these patterns are not always reliably detected due to the presence of irregular light patterns such as specular highlight reflections. In this paper, we aim at removing specular highlights from endoscopic videos using machine learning. We propose using a temporal generative adversarial network (GAN) to inpaint the hidden anatomy under specularities, inferring its appearance spatially and from neighbouring frames, where they are not present in the same location. This is achieved using in-vivo data from gastric endoscopy (Hyper Kvasir) in a fully unsupervised manner that relies on the automatic detection of specular highlights. System evaluations show significant improvements to other methods through direct comparison and ablation studies that depict the importance of the network's temporal and transfer learning components. The generalisability of our system to different surgical setups and procedures was also evaluated qualitatively on in-vivo data of gastric endoscopy and ex-vivo porcine data (SERV-CT, SCARED). We also assess the effect of our method in comparison to other methods on computer vision tasks that underpin 3D reconstruction and camera motion estimation, namely stereo disparity, optical flow, and sparse point feature matching. These are evaluated quantitatively and qualitatively and results show a positive effect of our specular inpainting method on these tasks in a novel comprehensive analysis. Our code and dataset are made available at https://github.com/endomapper/Endo-STTN.

Colonoscopy 3D video dataset with paired depth from 2D-3D registration

Screening colonoscopy is an important clinical application for several 3D computer vision techniques, including depth estimation, surface reconstruction, and missing region detection. However, the development, evaluation, and comparison of these techniques in real colonoscopy videos remain largely qualitative due to the difficulty of acquiring ground truth data. In this work, we present a Colonoscopy 3D Video Dataset (C3VD) acquired with a high definition clinical colonoscope and high-fidelity colon models for benchmarking computer vision methods in colonoscopy. We introduce a novel multimodal 2D-3D registration technique to register optical video sequences with ground truth rendered views of a known 3D model. The different modalities are registered by transforming optical images to depth maps with a Generative Adversarial Network and aligning edge features with an evolutionary optimizer. This registration method achieves an average translation error of 0.321 millimeters and an average rotation error of 0.159 degrees in simulation experiments where error-free ground truth is available. The method also leverages video information, improving registration accuracy by 55.6% for translation and 60.4% for rotation compared to single frame registration. 22 short video sequences were registered to generate 10,015 total frames with paired ground truth depth, surface normals, optical flow, occlusion, six degree-of-freedom pose, coverage maps, and 3D models. The dataset also includes screening videos acquired by a gastroenterologist with paired ground truth pose and 3D surface models. The dataset and registration source code are available at https://durr.jhu.edu/C3VD.

Role of surgery in T4N0-3M0 esophageal cancer

BACKGROUND

This study aimed to investigate an unsettled issue that whether T4 esophageal cancer could benefit from surgery.

METHODS

Patients with T4N0-3M0 esophageal cancer from 2004 to 2015 from the Surveillance, Epidemiology, and End Results (SEER) database were included in this study. Kaplan-Meier method, Cox proportional hazard regression, and propensity score matching (PSM) were used to compare overall survival (OS) between the surgery and no-surgery group.

RESULTS

A total of 1822 patients were analyzed. The multivariable Cox regression showed the HR (95% CI) for surgery vs. no surgery was 0.492 (0.427-0.567) (P < 0.001) in T4N0-3M0 cohort, 0.471 (0.354-0.627) (P < 0.001) in T4aN0-3M0 cohort, and 0.480 (0.335-0.689) (P < 0.001) in T4bN0-3M0 cohort. The HR (95% CI) for neoadjuvant therapy plus surgery vs. no surgery and surgery without neoadjuvant therapy vs. no surgery were 0.548 (0.461-0.650) (P < 0.001) and 0.464 (0.375-0.574) (P < 0.001), respectively. No significant OS difference was observed between neoadjuvant therapy plus surgery and surgery without neoadjuvant therapy: 0.966 (0.686-1.360) (P = 0.843). Subgroup analyses and PSM-adjusted analyses showed consistent results.

CONCLUSION

Surgery might bring OS improvement for T4N0-3M0 esophageal cancer patients, no matter in T4a disease or in T4b disease. Surgery with and without neoadjuvant therapy might both achieve better OS than no surgery.

Active learning for extracting surgomic features in robot-assisted minimally invasive esophagectomy: a prospective annotation study

BACKGROUND

With Surgomics, we aim for personalized prediction of the patient's surgical outcome using machine-learning (ML) on multimodal intraoperative data to extract surgomic features as surgical process characteristics. As high-quality annotations by medical experts are crucial, but still a bottleneck, we prospectively investigate active learning (AL) to reduce annotation effort and present automatic recognition of surgomic features.

METHODS

To establish a process for development of surgomic features, ten video-based features related to bleeding, as highly relevant intraoperative complication, were chosen. They comprise the amount of blood and smoke in the surgical field, six instruments, and two anatomic structures. Annotation of selected frames from robot-assisted minimally invasive esophagectomies was performed by at least three independent medical experts. To test whether AL reduces annotation effort, we performed a prospective annotation study comparing AL with equidistant sampling (EQS) for frame selection. Multiple Bayesian ResNet18 architectures were trained on a multicentric dataset, consisting of 22 videos from two centers.

RESULTS

In total, 14,004 frames were tag annotated. A mean F1-score of 0.75 ± 0.16 was achieved for all features. The highest F1-score was achieved for the instruments (mean 0.80 ± 0.17). This result is also reflected in the inter-rater-agreement (1-rater-kappa > 0.82). Compared to EQS, AL showed better recognition results for the instruments with a significant difference in the McNemar test comparing correctness of predictions. Moreover, in contrast to EQS, AL selected more frames of the four less common instruments (1512 vs. 607 frames) and achieved higher F1-scores for common instruments while requiring less training frames.

CONCLUSION

We presented ten surgomic features relevant for bleeding events in esophageal surgery automatically extracted from surgical video using ML. AL showed the potential to reduce annotation effort while keeping ML performance high for selected features. The source code and the trained models are published open source.

A super-high brightness and excellent colour quality laser-driven white light source enables miniaturized endoscopy

A laser-driven white light source promises intrinsic advantages for miniaturized endoscopic illumination. However, it remains a great challenge to simultaneously achieve high brightness and excellent colour rendition due to the shortage of highly efficient and thermally robust red-emitting laser phosphor converters. Here, we designed CaAlSiN3:Eu@Al (CASN@Al) converters with neglectable efficiency loss by tightly bonding all-inorganic phosphor films on an aluminium substrate. A layer-by-layer phosphor converter (LuAG/CASN@Al), i.e., stacking a green-emitting Lu3Al5O12:Ce (LuAG) layer on CASN@Al, was constructed to enhance light conversion efficiency and reduce reabsorption loss under blue laser excitation, which thus produces an excellent white light source with a luminous efficacy of 258 lm W-1 and a colour rendering index of 91. A miniaturized endoscopy with a coupling efficiency twice that of the commercial white LEDs was demonstrated by using the laser-driven white light and showed a central illuminance as high as 52 730 lx, more vivid images and long-term reliability.

Enhancing head and neck tumor management with artificial intelligence: Integration and perspectives

Head and neck tumors (HNTs) constitute a multifaceted ensemble of pathologies that primarily involve regions such as the oral cavity, pharynx, and nasal cavity. The intricate anatomical structure of these regions poses considerable challenges to efficacious treatment strategies. Despite the availability of myriad treatment modalities, the overall therapeutic efficacy for HNTs continues to remain subdued. In recent years, the deployment of artificial intelligence (AI) in healthcare practices has garnered noteworthy attention. AI modalities, inclusive of machine learning (ML), neural networks (NNs), and deep learning (DL), when amalgamated into the holistic management of HNTs, promise to augment the precision, safety, and efficacy of treatment regimens. The integration of AI within HNT management is intricately intertwined with domains such as medical imaging, bioinformatics, and medical robotics. This article intends to scrutinize the cutting-edge advancements and prospective applications of AI in the realm of HNTs, elucidating AI's indispensable role in prevention, diagnosis, treatment, prognostication, research, and inter-sectoral integration. The overarching objective is to stimulate scholarly discourse and invigorate insights among medical practitioners and researchers to propel further exploration, thereby facilitating superior therapeutic alternatives for patients.

Laparoscopic right hemi-hepatectomy plus total caudate lobectomy for perihilar cholangiocarcinoma via anterior approach with augmented reality navigation: a feasibility study

BACKGROUND

Right hemi-hepatectomy plus total caudate lobectomy is the appropriate procedure for type IIIa or partial type II pCCA. However, the laparoscopic implementation of this procedure remains technically challenging, especially hilar vascular dissection and en bloc resection of the total caudate lobe. Augmented reality navigation can provide intraoperative navigation to enhance visualization of invisible hilar blood vessels and guide the parenchymal transection plane.

METHODS

Eleven patients who underwent laparoscopic right hemi-hepatectomy plus total caudate lobectomy from January 2021 to January 2023 were enrolled in this study. Augmented reality navigation technology and the anterior approach were utilized in this operation. Routine operative and short-term postoperative outcomes were assessed to evaluate the feasibility of the novel navigation method in this operation.

RESULTS

Right hemi-hepatectomy plus total caudate lobectomy was successfully performed in all 11 enrolled patients. Among the 11 patients, the mean operation time was 454.5 ± 25.0 min and the mean estimated blood loss was 209.1 ± 56.1 ml. Negative surgical margins were achieved in all patients. The postoperative course of all the patients was uneventful, and the mean length of postoperative hospital stay was 10.5 ± 1.2 days.

CONCLUSION

Laparoscopic right hemi-hepatectomy plus total caudate lobectomy via the anterior approach may be feasible and safe for pCCA with the assistance of augmented reality navigation.

Endoscopic image classification algorithm based on Poolformer

Image desmoking is a significant aspect of endoscopic image processing, effectively mitigating visual field obstructions without the need for additional surgical interventions. However, current smoke removal techniques tend to apply comprehensive video enhancement to all frames, encompassing both smoke-free and smoke-affected images, which not only escalates computational costs but also introduces potential noise during the enhancement of smoke-free images. In response to this challenge, this paper introduces an approach for classifying images that contain surgical smoke within endoscopic scenes. This classification method provides crucial target frame information for enhancing surgical smoke removal, improving the scientific robustness, and enhancing the real-time processing capabilities of image-based smoke removal method. The proposed endoscopic smoke image classification algorithm based on the improved Poolformer model, augments the model's capacity for endoscopic image feature extraction. This enhancement is achieved by transforming the Token Mixer within the encoder into a multi-branch structure akin to ConvNeXt, a pure convolutional neural network. Moreover, the conversion to a single-path topology during the prediction phase elevates processing speed. Experiments use the endoscopic dataset sourced from the Hamlyn Centre Laparoscopic/Endoscopic Video Dataset, augmented by Blender software rendering. The dataset comprises 3,800 training images and 1,200 test images, distributed in a 4:1 ratio of smoke-free to smoke-containing images. The outcomes affirm the superior performance of this paper's approach across multiple parameters. Comparative assessments against existing models, such as mobilenet_v3, efficientnet_b7, and ViT-B/16, substantiate that the proposed method excels in accuracy, sensitivity, and inference speed. Notably, when contrasted with the Poolformer_s12 network, the proposed method achieves a 2.3% enhancement in accuracy, an 8.2% boost in sensitivity, while incurring a mere 6.4 frames per second reduction in processing speed, maintaining 87 frames per second. The results authenticate the improved performance of the refined Poolformer model in endoscopic smoke image classification tasks. This advancement presents a lightweight yet effective solution for the automatic detection of smoke-containing images in endoscopy. This approach strikes a balance between the accuracy and real-time processing requirements of endoscopic image analysis, offering valuable insights for targeted desmoking process.

Predicting mortality in patients with nonvariceal upper gastrointestinal bleeding using machine-learning

Background

Non-endoscopic risk scores, Glasgow Blatchford (GBS) and admission Rockall (Rock), are limited by poor specificity. The aim of this study was to develop an Artificial Neural Network (ANN) for the non-endoscopic triage of nonvariceal upper gastrointestinal bleeding (NVUGIB), with mortality as a primary outcome.

Methods

Four machine learning algorithms, namely, Linear Discriminant Analysis (LDA), Quadratic Discriminant Analysis (QDA), logistic regression (LR), K-Nearest Neighbor (K-NN), were performed with GBS, Rock, Beylor Bleeding score (BBS), AIM65, and T-score.

Results

A total of 1,096 NVUGIB hospitalized in the Gastroenterology Department of the County Clinical Emergency Hospital of Craiova, Romania, randomly divided into training and testing groups, were included retrospectively in our study. The machine learning models were more accurate at identifying patients who met the endpoint of mortality than any of the existing risk scores. AIM65 was the most important score in the detection of whether a NVUGIB would die or not, whereas BBS had no influence on this. Also, the greater AIM65 and GBS, and the lower Rock and T-score, the higher mortality will be.

Conclusion

The best accuracy was obtained by the hyperparameter-tuned K-NN classifier (98%), giving the highest precision and recall on the training and testing datasets among all developed models, showing that machine learning can accurately predict mortality in patients with NVUGIB.