Machine learning and network medicine: a novel approach for precision medicine and personalized therapy in cardiomyopathies

Prediction of pharmacological response in OCD using machine learning techniques and clinical and neuropsychological variables

INTRODUCTION

Obsessive compulsive disorder is associated with affected executive functioning, including memory, cognitive flexibility, and organizational strategies. As it was reported in previous studies, patients with preserved executive functions respond better to pharmacological treatment, while others need to keep trying different pharmacological strategies.

MATERIAL AND METHODS

In this work we used machine learning techniques to predict pharmacological response (OCD patients' symptomatology reduction) based on executive functioning and clinical variables. Among those variables we used anxiety, depression and obsessive-compulsive symptoms scores by applying State-Trait Anxiety Inventory, Hamilton Depression Rating Scale and Yale-Brown Obsessive Compulsive Scale respectively, while Rey-Osterrieth Complex Figure Test was used to assess organisation skills and non-verbal memory; Digits' subtests from Wechsler Adult Intelligence Scale-IV were used to assess short-term memory and working memory; and Raven's Progressive Matrices were applied to assess problem solving and abstract reasoning.

RESULTS

As a result of our analyses, we created a reliable algorithm that predicts Y-BOCS score after 12 weeks based on patients' clinical characteristics (sex at birth, age, pharmacological strategy, depressive and obsessive-compulsive symptoms, years passed since diagnostic and Raven's Progressive Matrices score) and Digits' scores. A high correlation (0.846) was achieved in predicted and true values.

CONCLUSIONS

The present study proves the viability to predict if a patient would respond or not to a certain pharmacological strategy with high reliability based on sociodemographics, clinical variables and cognitive functions as short-term memory and working memory. These results are promising to develop future prediction models to help clinical decision making.

Next-Generation Cardiac Magnetic Resonance Imaging Techniques for Characterization of Myocardial Disease

Purpose of the Review

Many novel cardiac magnetic resonance imaging (cMR) techniques have been developed for diagnosis, risk stratification, and monitoring of myocardial disease. The field is changing rapidly with advances in imaging technology. The purpose of this review is to give an update on next-generation cMR techniques with promising developments for clinical translation in the last two years, and to outline clinical applications.

Recent Findings

There has been increasing widespread clinical adoption of T1/T2 mapping into standard of care clinical practice. Development of auto segmentation has enabled clinical integration, with potential applications to minimize the use of contrast. Advances in diffusion tensor imaging, multiparametric mapping with cardiac MRI fingerprinting, automated quantitative perfusion mapping, metabolic imaging, elastography, and 4D flow are advancing the ability of cMR to provide further quantitative characterization to enable deep myocardial disease phenotyping. Together these advanced imaging features further augment the ability of cMR to contribute to novel disease characterization and may provide an important platform for personalized medicine.

Summary

Next-generation cMR techniques provide unique quantitative imaging features that can enable the identification of imaging biomarkers that may further refine disease classification and risk prediction. However, widespread clinical application continues to be limited by ground truth validation, reproducibility of the techniques across vendor platforms, increased scan time, and lack of widespread availability of advanced cardiac MRI physicists and expert readers. However, these techniques show great promise in minimizing the need for invasive testing, may elucidate novel pathophysiology, and may provide the ability for more accurate diagnosis of myocardial disease.

CDK5R1, GSE1, HSPG2 and WDFY3 as indirect epigenetic-sensitive genes in atrial fibrillation

BACKGROUND

Although mounting evidence supports that aberrant DNA methylation occurs in the hearts of patients with atrial fibrillation (AF), noninvasive epigenetic characterization of AF has not yet been defined.

METHODS

We investigated DNA methylome changes in peripheral blood CD4+ T cells isolated from 10 patients with AF relative to 11 healthy subjects (HS) who were enrolled in the DIANA clinical trial (NCT04371809) via reduced-representation bisulfite sequencing (RRBS).

RESULTS

An atrial-specific PPI network revealed 18 hub differentially methylated genes (DMGs), wherein ROC curve analysis revealed reasonable diagnostic performance of DNA methylation levels found within CDK5R1 (AUC = 0.76; p = 0.049), HSPG2 (AUC = 0.77; p = 0.038), WDFY3 (AUC = 0.78; p = 0.029), USP49 (AUC = 0.76; p = 0.049), GSE1 (AUC = 0.76; p = 0.049), AIFM1 (AUC = 0.76; p = 0.041), CDK5RAP2 (AUC = 0.81; p = 0.017), COL4A1 (AUC = 0.86; p < 0.001), SEPT8 (AUC = 0.90; p < 0.001), PFDN1 (AUC = 0.90; p < 0.01) and ACOT7 (AUC = 0.78; p = 0.032). Transcriptional profiling of the hub DMGs provided a significant overexpression of PSDM6 (p = 0.004), TFRC (p = 0.01), CDK5R1 (p < 0.001), HSPG2 (p = 0.01), WDFY3 (p < 0.001), USP49 (p = 0.004) and GSE1 (p = 0.021) in AF patients vs HS.

CONCLUSIONS

CDK5R1, GSE1, HSPG2 and WDFY3 resulted the best discriminatory genes both at methylation and gene expression level. Our results provide several candidate diagnostic biomarkers with the potential to advance precision medicine in AF.

Patient-level explainable machine learning to predict major adverse cardiovascular events from SPECT MPI and CCTA imaging

BACKGROUND

Machine learning (ML) has shown promise in improving the risk prediction in non-invasive cardiovascular imaging, including SPECT MPI and coronary CT angiography. However, most algorithms used remain black boxes to clinicians in how they compute their predictions. Furthermore, objective consideration of the multitude of available clinical data, along with the visual and quantitative assessments from CCTA and SPECT, are critical for optimal patient risk stratification. We aim to provide an explainable ML approach to predict MACE using clinical, CCTA, and SPECT data.

METHODS

Consecutive patients who underwent clinically indicated CCTA and SPECT myocardial imaging for suspected CAD were included and followed up for MACEs. A MACE was defined as a composite outcome that included all-cause mortality, myocardial infarction, or late revascularization. We employed an Automated Machine Learning (AutoML) approach to predict MACE using clinical, CCTA, and SPECT data. Various mainstream models with different sets of hyperparameters have been explored, and critical predictors of risk are obtained using explainable techniques on the global and patient levels. Ten-fold cross-validation was used in training and evaluating the AutoML model.

RESULTS

A total of 956 patients were included (mean age 61.1 ±14.2 years, 54% men, 89% hypertension, 81% diabetes, 84% dyslipidemia). Obstructive CAD on CCTA and ischemia on SPECT were observed in 14% of patients, and 11% experienced MACE. ML prediction's sensitivity, specificity, and accuracy in predicting a MACE were 69.61%, 99.77%, and 96.54%, respectively. The top 10 global predictive features included 8 CCTA attributes (segment involvement score, number of vessels with severe plaque ≥70, ≥50% stenosis in the left marginal coronary artery, calcified plaque, ≥50% stenosis in the left circumflex coronary artery, plaque type in the left marginal coronary artery, stenosis degree in the second obtuse marginal of the left circumflex artery, and stenosis category in the marginals of the left circumflex artery) and 2 clinical features (past medical history of MI or left bundle branch block, being an ever smoker).

CONCLUSION

ML can accurately predict risk of developing a MACE in patients suspected of CAD undergoing SPECT MPI and CCTA. ML feature-ranking can also show, at a sample- as well as at a patient-level, which features are key in making such a prediction.

Artificial intelligence, big data and heart transplantation: Actualities

BACKGROUND

As diagnostic and prognostic models developed by traditional statistics perform poorly in real-world, artificial intelligence (AI) and Big Data (BD) may improve the supply chain of heart transplantation (HTx), allocation opportunities, correct treatments, and finally optimize HTx outcome. We explored available studies, and discussed opportunities and limits of medical application of AI to the field of HTx.

METHOD

A systematic overview of studies published up to December 31st, 2022, in English on peer-revied journals, have been identified through PUBMED-MEDLINE-WEB of Science, referring to HTx, AI, BD. Studies were grouped in 4 domains based on main studies' objectives and results: etiology, diagnosis, prognosis, treatment. A systematic attempt was made to evaluate studies by the Prediction model Risk Of Bias ASsessment Tool (PROBAST) and the Transparent Reporting of a multivariable prediction model for Individual Prognosis Or Diagnosis (TRIPOD).

RESULTS

Among the 27 publications selected, none used AI applied to BD. Of the selected studies, 4 fell in the domain of etiology, 6 in the domain of diagnosis, 3 in the domain of treatment, and 17 in that of prognosis, as AI was most frequently used for algorithmic prediction and discrimination of survival, but in retrospective cohorts and registries. AI-based algorithms appeared superior to probabilistic functions to predict patterns, but external validation was rarely employed. Indeed, based on PROBAST, selected studies showed, to some extent, significant risk of bias (especially in the domain of predictors and analysis). In addition, as example of applicability in the real-world, a free-use prediction algorithm developed through AI failed to predict 1-year mortality post-HTx in cases from our center.

CONCLUSIONS

While AI-based prognostic and diagnostic functions performed better than those developed by traditional statistics, risk of bias, lack of external validation, and relatively poor applicability, may affect AI-based tools. More unbiased research with high quality BD meant for AI, transparency and external validations, are needed to have medical AI as a systematic aid to clinical decision making in HTx.

"Transplantomics" for predicting allograft rejection: real-life applications and new strategies from Network Medicine

Although decades of the reductionist approach achieved great milestones in optimizing the immunosuppression therapy, traditional clinical parameters still fail in predicting both acute and chronic (mainly) rejection events leading to higher rates across all solid organ transplants. To clarify the underlying immune-related cellular and molecular mechanisms, current biomedical research is increasingly focusing on "transplantomics" which relies on a huge quantity of big data deriving from genomics, transcriptomics, epigenomics, proteomics, and metabolomics platforms. The AlloMap (gene expression) and the AlloSure (donor-derived cell-free DNA) tests represent two successful examples of how omics and liquid biopsy can really improve the precision medicine of heart and kidney transplantation. One of the major challenges in translating big data in clinically useful biomarkers is the integration and interpretation of the different layers of omics datasets. Network Medicine offers advanced bioinformatic-molecular strategies which were widely used to integrate large omics datasets and clinical information in end-stage patients to prioritize potential biomarkers and drug targets. The application of network-oriented approaches to clarify the complex nature of graft rejection is still in its infancy. Here, we briefly discuss the real-life clinical applications derived from omics datasets as well as novel opportunities for establishing predictive tests in solid organ transplantation. Also, we provide an original "graft rejection interactome" and propose network-oriented strategies which can be useful to improve precision medicine of solid organ transplantation.

Big Data in cardiac surgery: real world and perspectives

Big Data, and the derived analysis techniques, such as artificial intelligence and machine learning, have been considered a revolution in the modern practice of medicine. Big Data comes from multiple sources, encompassing electronic health records, clinical studies, imaging data, registries, administrative databases, patient-reported outcomes and OMICS profiles. The main objective of such analyses is to unveil hidden associations and patterns. In cardiac surgery, the main targets for the use of Big Data are the construction of predictive models to recognize patterns or associations better representing the individual risk or prognosis compared to classical surgical risk scores. The results of these studies contributed to kindle the interest for personalized medicine and contributed to recognize the limitations of randomized controlled trials in representing the real world. However, the main sources of evidence for guidelines and recommendations remain RCTs and meta-analysis. The extent of the revolution of Big Data and new analytical models in cardiac surgery is yet to be determined.

Applications of machine learning in computational nanotechnology

Machine learning (ML) has gained extensive attention in recent years due to its powerful data analysis capabilities. It has been successfully applied to many fields and helped the researchers to achieve several major theoretical and applied breakthroughs. Some of the notable applications in the field of computational nanotechnology are ML potentials, property prediction, and material discovery. This review summarizes the state-of-the-art research progress in these three fields. ML potentials bridge the efficiency versus accuracy gap between density functional calculations and classical molecular dynamics. For property predictions, ML provides a robust method that eliminates the need for repetitive calculations for different simulation setups. Material design and drug discovery assisted by ML greatly reduce the capital and time investment by orders of magnitude. In this perspective, several common ML potentials and ML models are first introduced. Using these state-of-the-art models, developments in property predictions and material discovery are overviewed. Finally, this paper was concluded with an outlook on future directions of data-driven research activities in computational nanotechnology.

Radiogenomics and Artificial Intelligence Approaches Applied to Cardiac Computed Tomography Angiography and Cardiac Magnetic Resonance for Precision Medicine in Coronary Heart Disease: A Systematic Review

The risk of coronary heart disease (CHD) clinical manifestations and patient management is estimated according to risk scores accounting multifactorial risk factors, thus failing to cover the individual cardiovascular risk. Technological improvements in the field of medical imaging, in particular, in cardiac computed tomography angiography and cardiac magnetic resonance protocols, laid the development of radiogenomics. Radiogenomics aims to integrate a huge number of imaging features and molecular profiles to identify optimal radiomic/biomarker signatures. In addition, supervised and unsupervised artificial intelligence algorithms have the potential to combine different layers of data (imaging parameters and features, clinical variables and biomarkers) and elaborate complex and specific CHD risk models allowing more accurate diagnosis and reliable prognosis prediction. Literature from the past 5 years was systematically collected from PubMed and Scopus databases, and 60 studies were selected. We speculated the applicability of radiogenomics and artificial intelligence through the application of machine learning algorithms to identify CHD and characterize atherosclerotic lesions and myocardial abnormalities. Radiomic features extracted by cardiac computed tomography angiography and cardiac magnetic resonance showed good diagnostic accuracy for the identification of coronary plaques and myocardium structure; on the other hand, few studies exploited radiogenomics integration, thus suggesting further research efforts in this field. Cardiac computed tomography angiography resulted the most used noninvasive imaging modality for artificial intelligence applications. Several studies provided high performance for CHD diagnosis, classification, and prognostic assessment even though several efforts are still needed to validate and standardize algorithms for CHD patient routine according to good medical practice.

Optimizing indices of AF susceptibility and burden to evaluate AF severity, risk and outcomes

Atrial fibrillation (AF) has heterogeneous patterns of presentation concerning symptoms, duration of episodes, AF burden, and the tendency to progress towards the terminal step of permanent AF. AF is associated with a risk of stroke/thromboembolism traditionally considered dependent on patient-level risk factors rather than AF type, AF burden, or other characterizations. However, the time spent in AF appears related to an incremental risk of stroke, as suggested by the higher risk of stroke in patients with clinical AF vs. subclinical episodes and in patients with non-paroxysmal AF vs. paroxysmal AF. In patients with device-detected atrial tachyarrhythmias, AF burden is a dynamic process with potential transitions from a lower to a higher maximum daily arrhythmia burden, thus justifying monitoring its temporal evolution. In clinical terms, the appearance of the first episode of AF, the characterization of the arrhythmia in a specific AF type, the progression of AF, and the response to rhythm control therapies, as well as the clinical outcomes, are all conditioned by underlying heart disease, risk factors, and comorbidities. Improved understanding is needed on how to monitor and modulate the effect of factors that condition AF susceptibility and modulate AF-associated outcomes. The increasing use of wearables and apps in practice and clinical research may be useful to predict and quantify AF burden and assess AF susceptibility at the individual patient level. This may help us reveal why AF stops and starts again, or why AF episodes, or burden, cluster. Additionally, whether the distribution of burden is associated with variations in the propensity to thrombosis or other clinical adverse events. Combining the improved methods for data analysis, clinical and translational science could be the basis for the early identification of the subset of patients at risk of progressing to a longer duration/higher burden of AF and the associated adverse outcomes.