Tackling the Antimicrobial Resistance "Pandemic" with Machine Learning Tools: A Summary of Available Evidence.

Antimicrobial resistance is recognised as one of the top threats healthcare is bound to face in the future. There have been various attempts to preserve the efficacy of existing antimicrobials, develop new and efficient antimicrobials, manage infections with multi-drug resistant strains, and improve patient outcomes, resulting in a growing mass of routinely available data, including electronic health records and microbiological information that can be employed to develop individualised antimicrobial stewardship. Machine learning methods have been developed to predict antimicrobial resistance from whole-genome sequencing data, forecast medication susceptibility, recognise epidemic patterns for surveillance purposes, or propose new antibacterial treatments and accelerate scientific discovery. Unfortunately, there is an evident gap between the number of machine learning applications in science and the effective implementation of these systems. This narrative review highlights some of the outstanding opportunities that machine learning offers when applied in research related to antimicrobial resistance. In the future, machine learning tools may prove to be superbugs' kryptonite. This review aims to provide an overview of available publications to aid researchers that are looking to expand their work with new approaches and to acquaint them with the current application of machine learning techniques in this field.

Evaluation of Reporting Quality of Glaucoma Randomized Controlled Trial Abstracts: Current Status and Future Perspectives.

The aim of this study was to explore adherence to the Consolidated Standards of Reporting Trials (CONSORT) reporting standards in abstracts of randomized controlled trials on glaucoma. A cross-sectional observational study was conducted on the aforementioned abstracts, indexed in MEDLINE/PubMed between the years 2017 and 2021. In total, 302 abstracts met the inclusion criteria and were further analyzed. The median score of CONSORT-A items was 8 (interquartile range, 7-10) out of 17 (47.0%). Most analyzed studies were conducted in a single center (80.5%) and the abstracts were predominantly structured (95.0%). Only 20.5% of the abstracts adequately described the trial design, while randomization and funding were described by 6.0% of the abstracts. Higher overall scores were associated with structured abstracts, a multicenter setting, statistically significant results, funding by industry, a higher number of participants, and having been published in journals with impact factors above four (p < 0.001, respectively). The results of this study indicate a suboptimal adherence to CONSORT-A reporting standards, especially in particular items such as randomization and funding. Since these factors could contribute to the overall quality of the trials and further translation of trial results into clinical practice, an improvement in glaucoma research reporting transparency is needed.

DEBI-NN: Distance-encoding biomorphic-informational neural networks for minimizing the number of trainable parameters.

Modern artificial intelligence (AI) approaches mainly rely on neural network (NN) or deep NN methodologies. However, these approaches require large amounts of data to train, given, that the number of their trainable parameters has a polynomial relationship to their neuron counts. This property renders deep NNs challenging to apply in fields operating with small, albeit representative datasets such as healthcare. In this paper, we propose a novel neural network architecture which trains spatial positions of neural soma and axon pairs, where weights are calculated by axon-soma distances of connected neurons. We refer to this method as distance-encoding biomorphic-informational (DEBI) neural network. This concept significantly minimizes the number of trainable parameters compared to conventional neural networks. We demonstrate that DEBI models can yield comparable predictive performance in tabular and imaging datasets, where they require a fraction of trainable parameters compared to conventional NNs, resulting in a highly scalable solution.

Obesity and Clonal Hematopoiesis of Indeterminate Potential: Allies in Cardiovascular Diseases and Malignancies.

The clonal hematopoiesis of indeterminate potential (CHIP) is a term used to describe individuals who have detectable somatic mutations in genes commonly found in individuals with hematologic cancers but without any apparent evidence of such conditions. The mortality rate in individuals with CHIP is remarkably higher than the influence ascribed to hematologic malignancies, and it is plausible that cardiovascular diseases (CVD) could elucidate the apparent disparity. Studies have shown that the most frequently altered genes in CHIP are associated with the increased incidence of CVDs, type 2 diabetes mellitus (T2DM) and myeloid malignancies, as well as obesity. Additionally, multiple research studies have confirmed that obesity is also independently associated with these conditions, particularly the development and progression of atherosclerotic CVD. Considering the shared pathogenetic mechanisms of obesity and CHIP, our objective in this review was to investigate both preclinical and clinical evidence regarding the correlation between obesity and CHIP and the resulting implications of this interaction on the pathophysiology of CVDs and malignancies. The pro-inflammatory condition induced by obesity and CHIP enhances the probability of developing both diseases and increases the likelihood of developing CVDs, T2DM and malignancies, suggesting that a dangerous vicious loop may exist. However, it is vital to conduct additional research that will suggest targeted treatment options for obese individuals with CHIP in order to reduce harmful effects connected to these conditions.

The Role of AI in Breast Cancer Lymph Node Classification: A Comprehensive Review.

Breast cancer is a significant health issue affecting women worldwide, and accurately detecting lymph node metastasis is critical in determining treatment and prognosis. While traditional diagnostic methods have limitations and complications, artificial intelligence (AI) techniques such as machine learning (ML) and deep learning (DL) offer promising solutions for improving and supplementing diagnostic procedures. Current research has explored state-of-the-art DL models for breast cancer lymph node classification from radiological images, achieving high performances (AUC: 0.71-0.99). AI models trained on clinicopathological features also show promise in predicting metastasis status (AUC: 0.74-0.77), whereas multimodal (radiomics + clinicopathological features) models combine the best from both approaches and also achieve good results (AUC: 0.82-0.94). Once properly validated, such models could greatly improve cancer care, especially in areas with limited medical resources. This comprehensive review aims to compile knowledge about state-of-the-art AI models used for breast cancer lymph node metastasis detection, discusses proper validation techniques and potential pitfalls and limitations, and presents future directions and best practices to achieve high usability in real-world clinical settings.

Radiogenomic markers enable risk stratification and inference of mutational pathway states in head and neck cancer.

PURPOSE: Head and neck squamous cell carcinomas (HNSCCs) are a molecularly, histologically, and clinically heterogeneous set of tumors originating from the mucosal epithelium of the oral cavity, pharynx, and larynx. This heterogeneous nature of HNSCC is one of the main contributing factors to the lack of prognostic markers for personalized treatment. The aim of this study was to develop and identify multi-omics markers capable of improved risk stratification in this highly heterogeneous patient population. METHODS: In this retrospective study, we approached this issue by establishing radiogenomics markers to identify high-risk individuals in a cohort of 127 HNSCC patients. Hybrid in vivo imaging and whole-exome sequencing were employed to identify quantitative imaging markers as well as genetic markers on pathway-level prognostic in HNSCC. We investigated the deductibility of the prognostic genetic markers using anatomical and metabolic imaging using positron emission tomography combined with computed tomography. Moreover, we used statistical and machine learning modeling to investigate whether a multi-omics approach can be used to derive prognostic markers for HNSCC. RESULTS: Radiogenomic analysis revealed a significant influence of genetic pathway alterations on imaging markers. A highly prognostic radiogenomic marker based on cellular senescence was identified. Furthermore, the radiogenomic biomarkers designed in this study vastly outperformed the prognostic value of markers derived from genetics and imaging alone. CONCLUSION: Using the identified markers, a clinically meaningful stratification of patients is possible, guiding the identification of high-risk patients and potentially aiding in the development of effective targeted therapies.

Automated data preparation for in vivo tumor characterization with machine learning.

Background: This study proposes machine learning-driven data preparation (MLDP) for optimal data preparation (DP) prior to building prediction models for cancer cohorts. Methods: A collection of well-established DP methods were incorporated for building the DP pipelines for various clinical cohorts prior to machine learning. Evolutionary algorithm principles combined with hyperparameter optimization were employed to iteratively select the best fitting subset of data preparation algorithms for the given dataset. The proposed method was validated for glioma and prostate single center cohorts by 100-fold Monte Carlo (MC) cross-validation scheme with 80-20% training-validation split ratio. In addition, a dual-center diffuse large B-cell lymphoma (DLBCL) cohort was utilized with Center 1 as training and Center 2 as independent validation datasets to predict cohort-specific clinical endpoints. Five machine learning (ML) classifiers were employed for building prediction models across all analyzed cohorts. Predictive performance was estimated by confusion matrix analytics over the validation sets of each cohort. The performance of each model with and without MLDP, as well as with manually-defined DP were compared in each of the four cohorts. Results: Sixteen of twenty established predictive models demonstrated area under the receiver operator characteristics curve (AUC) performance increase utilizing the MLDP. The MLDP resulted in the highest performance increase for random forest (RF) (+0.16 AUC) and support vector machine (SVM) (+0.13 AUC) model schemes for predicting 36-months survival in the glioma cohort. Single center cohorts resulted in complex (6-7 DP steps) DP pipelines, with a high occurrence of outlier detection, feature selection and synthetic majority oversampling technique (SMOTE). In contrast, the optimal DP pipeline for the dual-center DLBCL cohort only included outlier detection and SMOTE DP steps. Conclusions: This study demonstrates that data preparation prior to ML prediction model building in cancer cohorts shall be ML-driven itself, yielding optimal prediction models in both single and multi-centric settings.

Multi-lesion radiomics of PET/CT for non-invasive survival stratification and histologic tumor risk profiling in patients with lung adenocarcinoma.

OBJECTIVES: This study investigates the ability of machine learning (ML) models trained on clinical data and 2-deoxy-2-[18F]fluoro-D-glucose(FDG) positron emission tomography/computed tomography (PET/CT) radiomics to predict overall survival (OS), tumor grade (TG), and histologic growth pattern risk (GPR) in lung adenocarcinoma (LUAD) patients. METHODS: A total of 421 treatment-naive patients with histologically-proven LUAD and available FDG PET/CT imaging were retrospectively included. Four cohorts were assessed for predicting 4-year OS (n = 276), 3-year OS (n = 280), TG (n = 298), and GPR (n = 265). FDG-avid lesions were delineated, and 2082 radiomics features were extracted and combined with endpoint-specific clinical parameters. ML models were built for the prediction of 4-year OS (M4OS), 3-year OS (M3OS), tumor grading (MTG), and histologic growth pattern risk (MGPR). A 100-fold Monte Carlo cross-validation with 80:20 training to validation split was employed as a performance evaluation for all models. The association between the M4OS and M3OS predictions with OS was assessed by the Kaplan-Meier survival analysis. RESULTS: The area under the receiver operator characteristics curve (AUC) was the highest for M4OS (AUC 0.88, 95% confidence interval (CI) 86.7-88.7), followed by M3OS (AUC 0.84, CI 82.9-84.9), while MTG and MGPR performed equally well (AUC 0.76, CI 74.4-77.9, CI 74.6-78, respectively). Predictions of M4OS (hazard ratio (HR) -2.4, CI -2.47 to -1.64, p < 0.05) and M3OS (HR -2.36, CI -2.79 to -1.93, p < 0.05) were independently associated with OS. CONCLUSION: ML models are able to predict long-term survival outcomes in LUAD patients with high accuracy. Furthermore, histologic grade and predominant growth pattern risk can be predicted with satisfactory accuracy. KEY POINTS: • Machine learning models trained on pre-therapeutic PET/CT radiomics enable highly accurate long-term survival prediction of patients with lung adenocarcinoma. • Highly accurate survival predictions are achieved in lung adenocarcinoma patients despite heterogenous histologies and treatment regimens. • Radiomic machine learning models are able to predict lung adenocarcinoma tumor grade and histologic growth pattern risk with satisfactory accuracy.

Adherence to Mediterranean diet and advanced glycation endproducts in patients with diabetes.

BACKGROUND: In recent years, American Diabetes Association started to strongly advocate the Mediterranean diet (MD) over other diets in patients with diabetes mellitus (DM) because of its beneficial effects on glycemic control and cardiovascular (CV) risk factors. Tissue levels of advanced glycation endproducts (AGEs) emerged as an indicator of CV risk in DM. Skin biopsy being invasive, the use of AGE Reader has been shown to reflect tissue AGEs reliably. AIM: To examine the association between adherence to MD and AGEs in patients with DM type II. METHODS: This cross-sectional study was conducted on 273 patients with DM type II. A survey questionnaire was composed of 3 separate sections. The first part of the questionnaire included general data and the habits of the participants. The second part aimed to assess the basic parameters of participants' diseases and associated conditions. The third part of the questionnaire was the Croatian version of the 14-item MD service score (MDSS). AGEs levels and associated CV risk were measured using AGE Reader (DiagnOptics Technologies BV, Groningen, The Netherlands). RESULTS: A total of 27 (9.9%) patients fulfilled criteria for adherence to MD, with a median score of 8.0 (6.0-10.0). Patients with none/limited CV risk had significantly higher percentage of MD adherence in comparison to patients with increased/definite CV risk (15.2% vs 6.9%, P = 0.028), as well as better adherence to guidelines for nuts (23.2% vs 12.6%, P = 0.023) and legumes (40.4% vs 25.9%, P = 0.013) consumption. Higher number of patients with glycated hemoglobin (HbA1c) < 7% adhered to MD when compared to patients with HbA1c > 7% (14.9% vs 7.3%, P = 0.045). Moreover, those patients followed the MDSS guidelines for eggs (33.0% vs 46.8%, P = 0.025) and wine (15.6% vs 29.8%, P = 0.006) consumption more frequently. MDSS score had significant positive correlation with disease duration (r = 0.179, P = 0.003) and negative correlation with body mass index (BMI) values (r = -0.159, P = 0.008). In the multiple linear regression model, BMI (ß ± SE, -0.09 ± 0.04, P = 0.037) and disease duration (ß ± SE, 0.07 ± 0.02, P < 0.001) remained significant independent correlates of the MDSS score. Patients with HbA1c > 7% think that educational programs on nutrition would be useful for patients in significantly more cases than patients with HbA1c < 7% (98.9% vs 92.6%, P = 0.009). CONCLUSION: Although adherence to MD was very low among people with diabetes, we demonstrated that adherence to MD is greater in patients with lower CV risk, longer disease duration, and well-controlled glycaemia.

A Sneak-Peek into the Physician's Brain: A Retrospective Machine Learning-Driven Investigation of Decision-Making in TAVR versus SAVR for Young High-Risk Patients with Severe Symptomatic Aortic Stenosis.

Transcatheter aortic valve replacement (TAVR) has rapidly become a viable alternative to the conventional isolated surgical aortic valve replacement (iSAVR) for treating severe symptomatic aortic stenosis. However, data on younger patients is scarce and a gap exists between data-based recommendations and the clinical use of TAVR. In our study, we utilized a machine learning (ML) driven approach to model the complex decision-making process of Heart Teams when treating young patients with severe symptomatic aortic stenosis with either TAVR or iSAVR and to identify the relevant considerations. Out of the considered factors, the variables most prominently featured in our ML model were congestive heart failure, established risk assessment scores, previous cardiac surgeries, a reduced left ventricular ejection fraction and peripheral vascular disease. Our study demonstrates a viable application of ML-based approaches for studying and understanding complex clinical decision-making processes.

Breast Tumor Characterization Using [18F]FDG-PET/CT Imaging Combined with Data Preprocessing and Radiomics.

Background: This study investigated the performance of ensemble learning holomic models for the detection of breast cancer, receptor status, proliferation rate, and molecular subtypes from [18F]FDG-PET/CT images with and without incorporating data pre-processing algorithms. Additionally, machine learning (ML) models were compared with conventional data analysis using standard uptake value lesion classification. Methods: A cohort of 170 patients with 173 breast cancer tumors (132 malignant, 38 benign) was examined with [18F]FDG-PET/CT. Breast tumors were segmented and radiomic features were extracted following the imaging biomarker standardization initiative (IBSI) guidelines combined with optimized feature extraction. Ensemble learning including five supervised ML algorithms was utilized in a 100-fold Monte Carlo (MC) cross-validation scheme. Data pre-processing methods were incorporated prior to machine learning, including outlier and borderline noisy sample detection, feature selection, and class imbalance correction. Feature importance in each model was assessed by calculating feature occurrence by the R-squared method across MC folds. Results: Cross validation demonstrated high performance of the cancer detection model (80% sensitivity, 78% specificity, 80% accuracy, 0.81 area under the curve (AUC)), and of the triple negative tumor identification model (85% sensitivity, 78% specificity, 82% accuracy, 0.82 AUC). The individual receptor status and luminal A/B subtype models yielded low performance (0.46-0.68 AUC). SUVmax model yielded 0.76 AUC in cancer detection and 0.70 AUC in predicting triple negative subtype. Conclusions: Predictive models based on [18F]FDG-PET/CT images in combination with advanced data pre-processing steps aid in breast cancer diagnosis and in ML-based prediction of the aggressive triple negative breast cancer subtype.

3D ultrasound guided navigation system with hybrid image fusion.

A prototype of a navigation system to fuse two image modalities is presented. The standard inter-modality registration is replaced with a tracker-based image registration of calibrated imaging devices. Intra-procedure transrectal US (TRUS) images were merged with pre-procedure magnetic resonance (MR) images for prostate biopsy. The registration between MR and TRUS images was performed by an additional abdominal 3D-US (ab-3D-US), which enables replacing the inter-modal MR/TRUS registration by an intra-modal ab-3D-US/3D-TRUS registration. Calibration procedures were carried out using an optical tracking system (OTS) for the pre-procedure image fusion of the ab-3D-US with the MR. Inter-modal ab-3D-US/MR image fusion was evaluated using a multi-cone phantom for the target registration error (TRE) and a prostate phantom for the Dice score and the Hausdorff distance of lesions . Finally, the pre-procedure ab- 3D-US was registered with the TRUS images and the errors for the transformation from the MR to the TRUS were determined. The TRE of the ab-3D-US/MR image registration was 1.81 mm. The Dice-score and the Hausdorff distance for ab-3D-US and MR were found to be 0.67 and 3.19 mm. The Dice score and the Hausdorff distance for TRUS and MR were 0.67 and 3.18 mm. The hybrid navigation system showed sufficient accuracy for fusion guided biopsy procedures with prostate phantoms. The system might provide intra-procedure fusion for most US-guided biopsy and ablation interventions.

Optimized Feature Extraction for Radiomics Analysis of 18F-FDG PET Imaging.

Radiomics analysis of 18F-FDG PET/CT images promises well for an improved in vivo disease characterization. To date, several studies have reported significant variations in textural features due to differences in patient preparation, imaging protocols, lesion delineation, and feature extraction. Our objective was to study variations in features before a radiomics analysis of 18F-FDG PET data and to identify those feature extraction and imaging protocol parameters that minimize radiomic feature variations across PET imaging systems. Methods: A whole-body National Electrical Manufacturers Association image-quality phantom was imaged with 13 PET/CT systems at 12 different sites following local protocols. We selected 37 radiomic features related to the 4 largest spheres (17-37 mm) in the phantom. On the basis of a combined analysis of voxel size, bin size, and lesion volume changes, feature and imaging system ranks were established. A 1-way ANOVA was performed over voxel size, bin size, and lesion volume subgroups to identify the dependency and the trend change in feature variations across these parameters. Results: Feature ranking revealed that the gray-level cooccurrence matrix and shape features are the least sensitive to PET imaging system variations. Imaging system ranking illustrated that the use of point-spread function, small voxel sizes, and narrow gaussian postfiltering helped minimize feature variations. ANOVA subgroup analysis indicated that variations in each of the 37 features and for a given voxel size and bin size can be minimized. Conclusion: Our results provide guidance to selecting optimized features from 18F-FDG PET/CT studies. We were able to demonstrate that feature variations can be minimized for selected image parameters and imaging systems. These results can help imaging specialists and feature engineers in increasing the quality of future radiomics studies involving PET/CT.

PSMA Ligand PET/MRI for Primary Prostate Cancer: Staging Performance and Clinical Impact.

PURPOSE: Primary staging of prostate cancer relies on modalities, which are limited. We evaluate simultaneous [68Ga]Ga-PSMA-11 PET (PSMA-PET)/MRI as a new diagnostic method for primary tumor-node-metastasis staging compared with histology and its impact on therapeutic decisions. EXPERIMENTAL DESIGN: We investigated 122 patients with PSMA-PET/MRI prior to planned radical prostatectomy (RP). Primary endpoint was the accuracy of PSMA-PET/MRI in tumor staging as compared with staging-relevant histology. In addition, a multidisciplinary team reassessed the initial therapeutic approach to evaluate its impact on the therapeutic management. RESULTS: PSMA-PET/MRI correctly identified prostate cancer in 119 of 122 patients (97.5%). Eighty-one patients were treated with RP and pelvic lymphadenectomy. The accuracy for T staging was 82.5% [95% confidence interval (CI), 73-90; P < 0.001], for T2 stage was 85% (95% CI, 71-94; P < 0.001), for T3a stage was 79% (95% CI, 43-85; P < 0.001), for T3b stage was 94% (95% CI, 73-100; P < 0.001), and for N1 stage was 93% (95% CI, 84-98; P < 0.001). PSMA-PET/MRI changed the therapeutic strategy in 28.7% of the patients with either the onset of systemic therapy/radiotherapy (n = 16) or active surveillance (n = 19). CONCLUSIONS: PSMA-PET/MRI can provide an accurate staging of newly diagnosed prostate cancer. In addition, treatment strategies were changed in almost a third of the patients due to the information of this hybrid imaging technique.

Glioma Survival Prediction with Combined Analysis of In Vivo 11C-MET PET Features, Ex Vivo Features, and Patient Features by Supervised Machine Learning.

Gliomas are the most common type of tumor in the brain. Although the definite diagnosis is routinely made ex vivo by histopathologic and molecular examination, diagnostic work-up of patients with suspected glioma is mainly done using MRI. Nevertheless, l-S-methyl-11C-methionine (11C-MET) PET holds great potential in the characterization of gliomas. The aim of this study was to establish machine-learning-driven survival models for glioma built on in vivo 11C-MET PET characteristics, ex vivo characteristics, and patient characteristics. Methods: The study included 70 patients with a treatment-naïve glioma that was 11C-MET-positive and had histopathology-derived ex vivo feature extraction, such as World Health Organization 2007 tumor grade, histology, and isocitrate dehydrogenase 1 R132H mutational status. The 11C-MET-positive primary tumors were delineated semiautomatically on PET images, followed by the extraction of tumor-to-background-based general and higher-order textural features by applying 5 different binning approaches. In vivo and ex vivo features, as well as patient characteristics (age, weight, height, body mass index, Karnofsky score), were merged to characterize the tumors. Machine-learning approaches were used to identify relevant in vivo, ex vivo, and patient features and their relative weights for predicting 36-mo survival. The resulting feature weights were used to establish 3 predictive models per binning configuration: one model based on a combination of in vivo, ex vivo, and clinical patient information (M36IEP); another based on in vivo and patient information only (M36IP); and a third based on in vivo information only (M36I). In addition, a binning-independent model based on ex vivo and patient information only (M36EP) was created. The established models were validated in a Monte Carlo cross-validation scheme. Results: The most prominent machine-learning-selected and -weighted features were patient-based and ex vivo-based, followed by in vivo-based. The highest areas under the curve for our models as revealed by the Monte Carlo cross-validation were 0.9 for M36IEP, 0.87 for M36EP, 0.77 for M36IP, and 0.72 for M36IConclusion: Prediction of survival in amino acid PET-positive glioma patients was highly accurate using computer-supported predictive models based on in vivo, ex vivo, and patient features.