Explaining Support Vector Machines: A Color Based Nomogram

User-Centered Design and Implementation of an Interoperable FHIR Application for Pediatric Pneumonia Prognostication in a Randomized Trial

OBJECTIVES

 To support a pragmatic, electronic health record (EHR)-based randomized controlled trial, we applied user-centered design (UCD) principles, evidence-based risk communication strategies, and interoperable software architecture to design, test, and deploy a prognostic tool for children in emergency departments (EDs) with pneumonia.

METHODS

 Risk for severe in-hospital outcomes was estimated using a validated ordinal logistic regression model to classify pneumonia severity. To render the results usable for ED clinicians, we created an integrated SMART on Fast Healthcare Interoperability Resources (FHIR) web application built for interoperable use in two pediatric EDs using different EHR vendors: Epic and Cerner. We followed a UCD framework, including problem analysis and user research, conceptual design and early prototyping, user interface development, formative evaluation, and postdeployment summative evaluation.

RESULTS

 Problem analysis and user research from 39 clinicians and nurses revealed user preferences for risk aversion, accessibility, and timing of risk communication. Early prototyping and iterative design incorporated evidence-based design principles, including numeracy, risk framing, and best-practice visualization techniques. After rigorous unit and end-to-end testing, the application was successfully deployed in both EDs, which facilitated enrollment, randomization, model visualization, data capture, and reporting for trial purposes.

CONCLUSION

 The successful implementation of a custom application for pneumonia prognosis and clinical trial support in two health systems on different EHRs demonstrates the importance of UCD, adherence to modern clinical data standards, and rigorous testing. Key lessons included the need for understanding users' real-world needs, regular knowledge management, application maintenance, and the recognition that FHIR applications require careful configuration for interoperability.

Single-cell RNA-seq reveals T cell exhaustion and immune response landscape in osteosarcoma

Background

T cell exhaustion in the tumor microenvironment has been demonstrated as a substantial contributor to tumor immunosuppression and progression. However, the correlation between T cell exhaustion and osteosarcoma (OS) remains unclear.

Methods

In our present study, single-cell RNA-seq data for OS from the GEO database was analysed to identify CD8+ T cells and discern CD8+ T cell subsets objectively. Subgroup differentiation trajectory was then used to pinpoint genes altered in response to T cell exhaustion. Subsequently, six machine learning algorithms were applied to develop a prognostic model linked with T cell exhaustion. This model was subsequently validated in the TARGETs and Meta cohorts. Finally, we examined disparities in immune cell infiltration, immune checkpoints, immune-related pathways, and the efficacy of immunotherapy between high and low TEX score groups.

Results

The findings unveiled differential exhaustion in CD8+ T cells within the OS microenvironment. Three genes related to T cell exhaustion (RAD23A, SAC3D1, PSIP1) were identified and employed to formulate a T cell exhaustion model. This model exhibited robust predictive capabilities for OS prognosis, with patients in the low TEX score group demonstrating a more favorable prognosis, increased immune cell infiltration, and heightened responsiveness to treatment compared to those in the high TEX score group.

Conclusion

In summary, our research elucidates the role of T cell exhaustion in the immunotherapy and progression of OS, the prognostic model constructed based on T cell exhaustion-related genes holds promise as a potential method for prognostication in the management and treatment of OS patients.

Explainability as the key ingredient for AI adoption in Industry 5.0 settings

Explainable Artificial Intelligence (XAI) has gained significant attention as a means to address the transparency and interpretability challenges posed by black box AI models. In the context of the manufacturing industry, where complex problems and decision-making processes are widespread, the XMANAI platform emerges as a solution to enable transparent and trustworthy collaboration between humans and machines. By leveraging advancements in XAI and catering the prompt collaboration between data scientists and domain experts, the platform enables the construction of interpretable AI models that offer high transparency without compromising performance. This paper introduces the approach to building the XMANAI platform and highlights its potential to resolve the "transparency paradox" of AI. The platform not only addresses technical challenges related to transparency but also caters to the specific needs of the manufacturing industry, including lifecycle management, security, and trusted sharing of AI assets. The paper provides an overview of the XMANAI platform main functionalities, addressing the challenges faced during the development and presenting the evaluation framework to measure the performance of the delivered XAI solutions. It also demonstrates the benefits of the XMANAI approach in achieving transparency in manufacturing decision-making, fostering trust and collaboration between humans and machines, improving operational efficiency, and optimizing business value.

Multiclass Support Vector Machine improves the Pivot-shift grading from Gerdy's acceleration resultant prior to the acute Anterior Cruciate Ligament surgery

INTRODUCTION

Rotatory laxity acceleration still lacks objective classification due to interval grading superposition, resulting in a biased pivot shift grading prior to the Anterior Cruciate Ligament (ACL) reconstruction. However, data analysis might help improve data grading in the operative room. Therefore, we described the improvement of the pivot-shift categorization in Gerdy's acceleration under anesthesia prior to ACL surgery using a support vector machine (SVM) classification, surgeon, and literature reference.

METHODS

Seventy-five patients (aged 30.3 ± 10.2 years, and IKDC 52.0 ± 16.5 points) with acute ACL rupture under anesthesia prior to ACL surgery were analyzed. Patients were graded with pivot-shift sign glide (+), clunk (++), and (+++) gross by senior orthopedic surgeons. At the same time, the tri-axial tibial plateau acceleration was measured. Categorical data were statistically described, and the accelerometry and categorical data were associated (α = 5%). A multiclass SVM kernel with the best accuracy trained by orthopedic surgeons and assisted from literature for missing data was compared with experienced surgeons and literature interval grading. The cubic SVM classifier achieved the best grading.

RESULTS

The intra-group proportions were different for each grading in the three compared strategies (p < 0.001). The inter-group proportions were different for all comparisons (p < 0.001). There were significant (p < 0.001) associations (Tau: 0.69, -0.28, and -0.50) between the surgeon and SVM, the surgeon and interval grading, and the interval and SVM, respectively.

CONCLUSION

The multiclass SVM classifier improves the acceleration categorization of the (+), (++), and (+++) pivot shift sign prior to the ACL surgery in agreement with surgeon criteria.

A comparative study between deep learning and radiomics models in grading liver tumors using hepatobiliary phase contrast-enhanced MR images

Purpose

To compare a deep learning model with a radiomics model in differentiating high-grade (LR-3, LR-4, LR-5) liver imaging reporting and data system (LI-RADS) liver tumors from low-grade (LR-1, LR-2) LI-RADS tumors based on the contrast-enhanced magnetic resonance images.

Methods

Magnetic resonance imaging scans of 361 suspected hepatocellular carcinoma patients were retrospectively reviewed. Lesion volume segmentation was manually performed by two radiologists, resulting in 426 lesions from the training set and 83 lesions from the test set. The radiomics model was constructed using a support vector machine (SVM) with pre-defined features, which was first selected using Chi-square test, followed by refining using binary least absolute shrinkage and selection operator (LASSO) regression. The deep learning model was established based on the DenseNet. Performance of the models was quantified by area under the receiver-operating characteristic curve (AUC), accuracy, sensitivity, specificity and F1-score.

Results

A set of 8 most informative features was selected from 1049 features to train the SVM classifier. The AUCs of the radiomics model were 0.857 (95% confidence interval [CI] 0.816–0.888) for the training set and 0.879 (95% CI 0.779–0.935) for the test set. The deep learning method achieved AUCs of 0.838 (95% CI 0.799–0.871) for the training set and 0.717 (95% CI 0.601–0.814) for the test set. The performance difference between these two models was assessed by t-test, which showed the results in both training and test sets were statistically significant.

Conclusion

The deep learning based model can be trained end-to-end with little extra domain knowledge, while the radiomics model requires complex feature selection. However, this process makes the radiomics model achieve better performance in this study with smaller computational cost and more potential on model interpretability.

Comprehensive Pan-Cancer Analyses of Immunogenic Cell Death as a Biomarker in Predicting Prognosis and Therapeutic Response

Immunogenic cell death (ICD), a form of regulated cell death, is related to anticancer therapy. Due to the absence of widely accepted markers, characterizing ICD-related phenotypes across cancer types remained unexplored. Here, we defined the ICD score to delineate the ICD landscape across 33 cancerous types and 31 normal tissue types based on transcriptomic, proteomic and epigenetics data from multiple databases. We found that ICD score showed cancer type-specific association with genomic and immune features. Importantly, the ICD score had the potential to predict therapy response and patient prognosis in multiple cancer types. We also developed an ICD-related prognostic model by machine learning and cox regression analysis. Single-cell level analysis revealed intra-tumor ICD state heterogeneity and communication between ICD-based clusters of T cells and other immune cells in the tumor microenvironment in colon cancer. For the first time, we identified IGF2BP3 as a potential ICD regulator in colon cancer. In conclusion, our study provides a comprehensive framework for evaluating the relation between ICD and clinical relevance, gaining insights into identification of ICD as a potential cancer-related biomarker and therapeutic target.

Machine learning-based tumor-infiltrating immune cell-associated lncRNAs for predicting prognosis and immunotherapy response in patients with glioblastoma

Long noncoding ribonucleic acids (RNAs; lncRNAs) have been associated with cancer immunity regulation. However, the roles of immune cell-specific lncRNAs in glioblastoma (GBM) remain largely unknown. In this study, a novel computational framework was constructed to screen the tumor-infiltrating immune cell-associated lncRNAs (TIIClnc) for developing TIIClnc signature by integratively analyzing the transcriptome data of purified immune cells, GBM cell lines and bulk GBM tissues using six machine learning algorithms. As a result, TIIClnc signature could distinguish survival outcomes of GBM patients across four independent datasets, including the Xiangya in-house dataset, and more importantly, showed superior performance than 95 previously established signatures in gliomas. TIIClnc signature was revealed to be an indicator of the infiltration level of immune cells and predicted the response outcomes of immunotherapy. The positive correlation between TIIClnc signature and CD8, PD-1 and PD-L1 was verified in the Xiangya in-house dataset. As a newly demonstrated predictive biomarker, the TIIClnc signature enabled a more precise selection of the GBM population who would benefit from immunotherapy and should be validated and applied in the near future.

Identification of Diagnostic CpG Signatures in Patients with Gestational Diabetes Mellitus via Epigenome-Wide Association Study Integrated with Machine Learning

Background

Gestational diabetes mellitus (GDM) is the most prevalent metabolic disease during pregnancy, but the diagnosis is controversial and lagging partly due to the lack of useful biomarkers. CpG methylation is involved in the development of GDM. However, the specific CpG methylation sites serving as diagnostic biomarkers of GDM remain unclear. Here, we aimed to explore CpG signatures and establish the predicting model for the GDM diagnosis.

Methods

DNA methylation data of GSE88929 and GSE102177 were obtained from the GEO database, followed by the epigenome-wide association study (EWAS). GO and KEGG pathway analyses were performed by using the clusterProfiler package of R. The PPI network was constructed in the STRING database and Cytoscape software. The SVM model was established, in which the β-values of selected CpG sites were the predictor variable and the occurrence of GDM was the outcome variable.

Results

We identified 62 significant CpG methylation sites in the GDM samples compared with the control samples. GO and KEGG analyses based on the 62 CpG sites demonstrated that several essential cellular processes and signaling pathways were enriched in the system. A total of 12 hub genes related to the identified CpG sites were found in the PPI network. The SVM model based on the selected CpGs within the promoter region, including cg00922748, cg05216211, cg05376185, cg06617468, cg17097119, and cg22385669, was established, and the AUC values of the training set and testing set in the model were 0.8138 and 0.7576. The AUC value of the independent validation set of GSE102177 was 0.6667.

Conclusion

We identified potential diagnostic CpG signatures by EWAS integrated with the SVM model. The SVM model based on the identified 6 CpG sites reliably predicted the GDM occurrence, contributing to the diagnosis of GDM. Our finding provides new insights into the cross-application of EWAS and machine learning in GDM investigation.

Machine learning-assisted decision-support models to better predict patients with calculous pyonephrosis

Background

To develop a machine learning (ML)-assisted model capable of accurately identifying patients with calculous pyonephrosis before making treatment decisions by integrating multiple clinical characteristics.

Methods

We retrospectively collected data from patients with obstructed hydronephrosis who underwent retrograde ureteral stent insertion, percutaneous nephrostomy (PCN), or percutaneous nephrolithotomy (PCNL). The study cohort was divided into training and testing datasets in a 70:30 ratio for further analysis. We developed 5 ML-assisted models from 22 clinical features using logistic regression (LR), LR optimized by least absolute shrinkage and selection operator (Lasso) regularization (Lasso-LR), support vector machine (SVM), extreme gradient boosting (XGBoost), and random forest (RF). The area under the curve (AUC) was applied to determine the model with the highest discrimination. Decision curve analysis (DCA) was used to investigate the clinical net benefit associated with using the predictive models.

Results

A total of 322 patients were included, with 225 patients in the training dataset, and 97 patients in the testing dataset. The XGBoost model showed good discrimination with the AUC, accuracy, sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) of 0.981, 0.991, 0.962, 1.000, 1.000, and 0.989, respectively, followed by SVM [AUC =0.985, 95% confidence interval (CI): 0.970–1.000], Lasso-LR (AUC =0.977, 95% CI: 0.958–0.996), LR (AUC =0.936, 95% CI: 0.905–0.968), and RF (AUC =0.920, 95% CI: 0.870–0.970). Validation of the model showed that SVM yielded the highest AUC (0.977, 95% CI: 0.952–1.000), followed by Lasso-LR (AUC =0.959, 95% CI: 0.921–0.997), XGBoost (AUC =0.958, 95% CI: 0.902–1.000), LR (AUC =0.932, 95% CI: 0.878–0.987), and RF (AUC =0.868, 95% CI: 0.779–0.958) in the testing dataset.

Conclusions

Our ML-based models had good discrimination in predicting patients with obstructed hydronephrosis at high risk of harboring pyonephrosis, and the use of these models may be greatly beneficial to urologists in treatment planning, patient selection, and decision-making.

Balancing accuracy and interpretability of machine learning approaches for radiation treatment outcomes modeling

Radiation outcomes prediction (ROP) plays an important role in personalized prescription and adaptive radiotherapy. A clinical decision may not only depend on an accurate radiation outcomes’ prediction, but also needs to be made based on an informed understanding of the relationship among patients’ characteristics, radiation response and treatment plans. As more patients’ biophysical information become available, machine learning (ML) techniques will have a great potential for improving ROP. Creating explainable ML methods is an ultimate task for clinical practice but remains a challenging one. Towards complete explainability, the interpretability of ML approaches needs to be first explored. Hence, this review focuses on the application of ML techniques for clinical adoption in radiation oncology by balancing accuracy with interpretability of the predictive model of interest. An ML algorithm can be generally classified into an interpretable (IP) or non-interpretable (NIP) (“black box”) technique. While the former may provide a clearer explanation to aid clinical decision-making, its prediction performance is generally outperformed by the latter. Therefore, great efforts and resources have been dedicated towards balancing the accuracy and the interpretability of ML approaches in ROP, but more still needs to be done. In this review, current progress to increase the accuracy for IP ML approaches is introduced, and major trends to improve the interpretability and alleviate the “black box” stigma of ML in radiation outcomes modeling are summarized. Efforts to integrate IP and NIP ML approaches to produce predictive models with higher accuracy and interpretability for ROP are also discussed.

Investigating Maize Yield-Related Genes in Multiple Omics Interaction Network Data

Zea mays (maize) is the highest yielding food crop globally, feeding large numbers of people across the planet. It is thus especially important to explore the key genes that affect maize production with prior knowledge. Merging multiple datasets of different types can improve the accuracy of candidate genes prediction results, so we constructed interaction networks using gene, mRNA, protein, and expression profile datasets. A network propagation schedule was used considering combined scores obtained by integrating both network scores and significance scores for each candidate gene based on the guilt-by-association principle. An SVM model was used to optimize the weighted parameters to achieve more reliable results, according to the accuracy of label classification. We found that integrating multiple omics data with more data types improves the reliability of the results. We investigated the GO terms particularly associated with the top 100 candidate genes and the known genes, and analyzed the roles that these genes play in determining the phenotype of maize. We hope that the candidate genes identified here will provide a biological perspective and contribute to maize breeding research.

Develop Nomogram to Predict Malignancy of Intraductal Papillary Mucinous Neoplasm

A nomogram is a useful graphical tool for presenting a risk prediction and prognosis prediction in medical research. Intraductal papillary mucinous neoplasm (IPMN) is the premalignant lesions of the pancreas. Among the IPMN, branch duct (BD) IPMN is hard to determine whether progress to an invasive tumor or not. Surgery on the pancreas part is likely to lower the quality of life of the patient, so avoiding surgery to remove IPMN tissue of the patients with low risk should be carefully decided. In this study, we introduce the process of constructing a nomogram and illustrate it with a prediction model to predict malignancy of IPMN.

Making Expert Decisions Easier to Fathom: On the Explainability of Visual Object Recognition Expertise

In everyday life, we rely on human experts to make a variety of complex decisions, such as medical diagnoses. These decisions are typically made through some form of weakly guided learning, a form of learning in which decision expertise is gained through labeled examples rather than explicit instructions. Expert decisions can significantly affect people other than the decision-maker (for example, teammates, clients, or patients), but may seem cryptic and mysterious to them. It is therefore desirable for the decision-maker to explain the rationale behind these decisions to others. This, however, can be difficult to do. Often, the expert has a “gut feeling” for what the correct decision is, but may have difficulty giving an objective set of criteria for arriving at it. Explainability of human expert decisions, i.e., the extent to which experts can make their decisions understandable to others, has not been studied systematically. Here, we characterize the explainability of human decision-making, using binary categorical decisions about visual objects as an illustrative example. We trained a group of “expert” subjects to categorize novel, naturalistic 3-D objects called “digital embryos” into one of two hitherto unknown categories, using a weakly guided learning paradigm. We then asked the expert subjects to provide a written explanation for each binary decision they made. These experiments generated several intriguing findings. First, the expert’s explanations modestly improve the categorization performance of naïve users (paired t-tests, p < 0.05). Second, this improvement differed significantly between explanations. In particular, explanations that pointed to a spatially localized region of the object improved the user’s performance much better than explanations that referred to global features. Third, neither experts nor naïve subjects were able to reliably predict the degree of improvement for a given explanation. Finally, significant bias effects were observed, where naïve subjects rated an explanation significantly higher when told it comes from an expert user, compared to the rating of the same explanation when told it comes from another non-expert, suggesting a variant of the Asch conformity effect. Together, our results characterize, for the first time, the various issues, both methodological and conceptual, underlying the explainability of human decisions.

Societal Issues Concerning the Application of Artificial Intelligence in Medicine

Background

Medicine is becoming an increasingly data-centred discipline and, beyond classical statistical approaches, artificial intelligence (AI) and, in particular, machine learning (ML) are attracting much interest for the analysis of medical data. It has been argued that AI is experiencing a fast process of commodification. This characterization correctly reflects the current process of industrialization of AI and its reach into society. Therefore, societal issues related to the use of AI and ML should not be ignored any longer and certainly not in the medical domain. These societal issues may take many forms, but they all entail the design of models from a human-centred perspective, incorporating human-relevant requirements and constraints. In this brief paper, we discuss a number of specific issues affecting the use of AI and ML in medicine, such as fairness, privacy and anonymity, explainability and interpretability, but also some broader societal issues, such as ethics and legislation. We reckon that all of these are relevant aspects to consider in order to achieve the objective of fostering acceptance of AI- and ML-based technologies, as well as to comply with an evolving legislation concerning the impact of digital technologies on ethically and privacy sensitive matters. Our specific goal here is to reflect on how all these topics affect medical applications of AI and ML. This paper includes some of the contents of the "2nd Meeting of Science and Dialysis: Artificial Intelligence," organized in the Bellvitge University Hospital, Barcelona, Spain.

Summary and Key Messages

AI and ML are attracting much interest from the medical community as key approaches to knowledge extraction from data. These approaches are increasingly colonizing ambits of social impact, such as medicine and healthcare. Issues of social relevance with an impact on medicine and healthcare include (although they are not limited to) fairness, explainability, privacy, ethics and legislation.

Machine Learning and Radiogenomics: Lessons Learned and Future Directions

Due to the rapid increase in the availability of patient data, there is significant interest in precision medicine that could facilitate the development of a personalized treatment plan for each patient on an individual basis. Radiation oncology is particularly suited for predictive machine learning (ML) models due to the enormous amount of diagnostic data used as input and therapeutic data generated as output. An emerging field in precision radiation oncology that can take advantage of ML approaches is radiogenomics, which is the study of the impact of genomic variations on the sensitivity of normal and tumor tissue to radiation. Currently, patients undergoing radiotherapy are treated using uniform dose constraints specific to the tumor and surrounding normal tissues. This is suboptimal in many ways. First, the dose that can be delivered to the target volume may be insufficient for control but is constrained by the surrounding normal tissue, as dose escalation can lead to significant morbidity and rare. Second, two patients with nearly identical dose distributions can have substantially different acute and late toxicities, resulting in lengthy treatment breaks and suboptimal control, or chronic morbidities leading to poor quality of life. Despite significant advances in radiogenomics, the magnitude of the genetic contribution to radiation response far exceeds our current understanding of individual risk variants. In the field of genomics, ML methods are being used to extract harder-to-detect knowledge, but these methods have yet to fully penetrate radiogenomics. Hence, the goal of this publication is to provide an overview of ML as it applies to radiogenomics. We begin with a brief history of radiogenomics and its relationship to precision medicine. We then introduce ML and compare it to statistical hypothesis testing to reflect on shared lessons and to avoid common pitfalls. Current ML approaches to genome-wide association studies are examined. The application of ML specifically to radiogenomics is next presented. We end with important lessons for the proper integration of ML into radiogenomics.

Identification of transformer fault based on dissolved gas analysis using hybrid support vector machine-modified evolutionary particle swarm optimisation

Early detection of power transformer fault is important because it can reduce the maintenance cost of the transformer and it can ensure continuous electricity supply in power systems. Dissolved Gas Analysis (DGA) technique is commonly used to identify oil-filled power transformer fault type but utilisation of artificial intelligence method with optimisation methods has shown convincing results. In this work, a hybrid support vector machine (SVM) with modified evolutionary particle swarm optimisation (EPSO) algorithm was proposed to determine the transformer fault type. The superiority of the modified PSO technique with SVM was evaluated by comparing the results with the actual fault diagnosis, unoptimised SVM and previous reported works. Data reduction was also applied using stepwise regression prior to the training process of SVM to reduce the training time. It was found that the proposed hybrid SVM-Modified EPSO (MEPSO)-Time Varying Acceleration Coefficient (TVAC) technique results in the highest correct identification percentage of faults in a power transformer compared to other PSO algorithms. Thus, the proposed technique can be one of the potential solutions to identify the transformer fault type based on DGA data on site.

Discovery and application of immune biomarkers for hematological malignancies

INTRODUCTION

Hematological malignancies originate and progress in primary and secondary lymphoid organs, where they establish a uniquely immune-suppressive tumour microenvironment. Although high-throughput transcriptomic and proteomic approaches are being employed to interrogate immune surveillance and escape mechanisms in patients with solid tumours, and to identify actionable targets for immunotherapy, our knowledge of the immunological landscape of hematological malignancies, as well as our understanding of the molecular circuits that underpin the establishment of immune tolerance, is not comprehensive. Areas covered: This article will discuss how multiplexed immunohistochemistry, flow cytometry/mass cytometry, proteomic and genomic techniques can be used to dynamically capture the complexity of tumour-immune interactions. Moreover, the analysis of multi-dimensional, clinically annotated data sets obtained from public repositories such as Array Express, TCGA and GEO is crucial to identify immune biomarkers, to inform the rational design of immune therapies and to predict clinical benefit in individual patients. We will also highlight how artificial neural network models and alternative methodologies integrating other algorithms can support the identification of key molecular drivers of immune dysfunction. Expert commentary: High-dimensional technologies have the potential to enhance our understanding of immune-cancer interactions and will support clinical decision making and the prediction of therapeutic benefit from immune-based interventions.