Prediction of hemophilia A severity using a small-input machine-learning framework

Prediction of inhibitor development in previously untreated and minimally treated children with severe and moderately severe hemophilia A using a machine-learning network

BACKGROUND

Prediction of inhibitor development in patients with hemophilia A (HA) remains a challenge.

OBJECTIVES

To construct a predictive model for inhibitor development in HA using a network of clinical variables and biomarkers based on the individual similarity network.

METHODS

Previously untreated and minimally treated children with severe/moderately severe HA, participants of the HEMFIL Cohort Study, were followed up until reaching 75 exposure days (EDs) without inhibitor (INH-) or upon inhibitor development (INH+). Clinical data and biological samples were collected before the start of factor (F)VIII replacement (T0). A predictive model (HemfilNET) was built to compare the networks and potential global topological differences between INH- and INH+ at T0, considering the network robustness. For validation, the "leave-one-out" cross-validation technique was employed. Accuracy, precision, recall, and F1-score were used as evaluation metrics for the machine-learning model.

RESULTS

We included 95 children with HA (CHA), of whom 31 (33%) developed inhibitors. The algorithm, featuring 37 variables, identified distinct patterns of networks at T0 for INH+ and INH-. The accuracy of the model was 74.2% for CHA INH+ and 98.4% for INH-. By focusing the analysis on CHA with high-risk F8 mutations for inhibitor development, the accuracy in identifying CHA INH+ increased to 82.1%.

CONCLUSION

Our machine-learning algorithm demonstrated an overall accuracy of 90.5% for predicting inhibitor development in CHA, which further improved when restricting the analysis to CHA with a high-risk F8 genotype. However, our model requires validation in other cohorts. Yet, missing data for some variables hindered more precise predictions.

Application of Artificial Intelligence and Machine Learning Was Not Able to Reliably Predict Poor Outcomes in People With Hemophilia

Background Artificial intelligence (AI) and machine learning (ML) are currently used in the clinical field to improve the outcome predictions on disease diagnosis and prognosis. However, to date, few AI/ML applications have been reported in rare diseases, such as hemophilia. In this study, taking advantage of the ATHNdataset, an extensive repository of hemostasis and thrombosis data, we aimed to demonstrate the application of AI/ML approaches to build predictive models to identify persons with hemophilia (PwH) who are at risk of poor outcome and to inform providers in clinical decision-making towards helping patients prevent long-term complications. Materials and methods This project was carried out in two steps. First, the data were mined from ATHN 7, a subset study of the ATHNdataset, to determine markers that defined "poor outcome." Second, we applied multiple AI/ML approaches on the ATHNdataset to validate our findings and to develop predictive models to identify PwH at risk of poor outcomes. The classical regression-based predictive model was used as a reference to evaluate the performance of various AI/ML models. Results Our models included features similarly distributed to response variables of interest, resulting in a limited ability to distinguish poor outcomes. Low recall (<53%) resulted in no single model reliably predicting poor outcomes out of all actual positive cases. Our results suggest that, to build a more useful AI/ML model, we may need a larger dataset size along with additional features. Furthermore, our results showed that most of the AI/ML models outperformed the classical logistic regression model in both model accuracy and precision. Conclusions Our AI and ML model showed limited ability to predict poor outcomes in people with hemophilia.

Application of machine learning approaches for predicting hemophilia A severity

BACKGROUND

Hemophilia A (HA) is an X-linked congenital bleeding disorder, which leads to deficiency of clotting factor (F) VIII. It mostly affects males, and females are considered carriers. However, it is now recognized that variants of F8 in females can result in HA. Nonetheless, most females go undiagnosed and untreated for HA, and their bleeding complications are attributed to other causes. Predicting the severity of HA for female patients can provide valuable insights for treating the conditions associated with the disease, such as heavy bleeding.

OBJECTIVES

To predict the severity of HA based on F8 genotype using a machine learning (ML) approach.

METHODS

Using multiple datasets of variants in the F8 and disease severity from various repositories, we derived the sequence for the FVIII protein. Using the derived sequences, we used ML models to predict the severity of HA in female patients.

RESULTS

Utilizing different classification models, we highlight the validity of the datasets and our approach with predictive F1 scores of 0.88, 0.99, 0.93, 0.99, and 0.90 for all the validation sets.

CONCLUSION

Although with some limitations, ML-based approaches demonstrated the successful prediction of disease severity in female HA patients based on variants in the F8. This study confirms previous research findings that ML can help predict the severity of hemophilia. These results can be valuable for future studies in achieving better treatment and clinical outcomes for female patients with HA, which is an urgent unmet need.

Clinical Implications of Discrepancy between One-Stage Clotting and Chromogenic Factor IX Activity in Hemophilia B

BACKGROUND

 Discrepancy in factor IX activity (FIX:C) between one-stage assay (OSA) and chromogenic substrate assay (CSA) in patients with hemophilia B (PwHB) introduces challenges for clinical management.

AIM

 To study the differences in FIX:C using OSA and CSA in moderate and mild hemophilia B (HB), their impact on classification of severity, and correlation with genotype.

METHODS

 Single-center study including 21 genotyped and clinically characterized PwHB. FIX:C by OSA was measured using ActinFSL (Siemens) and CSA by Biophen (Hyphen). In addition, in vitro experiments with wild-type FIX were performed. Reproducibility of CSA was assessed between three European coagulation laboratories.

RESULTS

 FIX:C by CSA was consistently lower than by OSA, with 10/17 PwHB having a more severe hemophilia type by CSA. OSA displayed a more accurate description of the clinical bleeding severity, compared with CSA. A twofold difference between OSA:CSA FIX:C was present in 12/17 PwHB; all patients had genetic missense variants in the FIX serine protease domain. Discrepancy was also observed with diluted normal plasma, most significant for values below 0.10 IU/mL. Assessment of samples with low FIX:C showed excellent reproducibility of the CSA results between the laboratories.

CONCLUSION

 FIX:C was consistently higher by OSA compared with the CSA. Assessing FIX:C by CSA alone would have led to diagnosis of a more severe hemophilia type in a significant proportion of patients. Our study suggests using both OSA and CSA FIX:C together with genotyping to classify HB severity and provide essential information for clinical management.

Full-scale network analysis reveals properties of the FV protein structure organization

Blood coagulation is a vital process for humans and other species. Following an injury to a blood vessel, a cascade of molecular signals is transmitted, inhibiting and activating more than a dozen coagulation factors and resulting in the formation of a fibrin clot that ceases the bleeding. In this process, the Coagulation factor V (FV) is a master regulator, coordinating critical steps of this process. Mutations to this factor result in spontaneous bleeding episodes and prolonged hemorrhage after trauma or surgery. Although the role of FV is well characterized, it is unclear how single-point mutations affect its structure. In this study, to understand the effect of mutations, we created a detailed network map of this protein, where each node is a residue, and two residues are connected if they are in close proximity in the three-dimensional structure. Overall, we analyzed 63 point-mutations from patients and identified common patterns underlying FV deficient phenotypes. We used structural and evolutionary patterns as input to machine learning algorithms to anticipate the effects of mutations and anticipated FV-deficiency with fair accuracy. Together, our results demonstrate how clinical features, genetic data and in silico analysis are converging to enhance treatment and diagnosis of coagulation disorders.

A graph-based machine learning framework identifies critical properties of FVIII that lead to hemophilia A

Introduction: Blood coagulation is an essential process to cease bleeding in humans and other species. This mechanism is characterized by a molecular cascade of more than a dozen components activated after an injury to a blood vessel. In this process, the coagulation factor VIII (FVIII) is a master regulator, enhancing the activity of other components by thousands of times. In this sense, it is unsurprising that even single amino acid substitutions result in hemophilia A (HA)-a disease marked by uncontrolled bleeding and that leaves patients at permanent risk of hemorrhagic complications. Methods: Despite recent advances in the diagnosis and treatment of HA, the precise role of each residue of the FVIII protein remains unclear. In this study, we developed a graph-based machine learning framework that explores in detail the network formed by the residues of the FVIII protein, where each residue is a node, and two nodes are connected if they are in close proximity on the FVIII 3D structure. Results: Using this system, we identified the properties that lead to severe and mild forms of the disease. Finally, in an effort to advance the development of novel recombinant therapeutic FVIII proteins, we adapted our framework to predict the activity and expression of more than 300 in vitro alanine mutations, once more observing a close agreement between the in silico and the in vitro results. Discussion: Together, the results derived from this study demonstrate how graph-based classifiers can leverage the diagnostic and treatment of a rare disease.

F8/F9 variants in the population-based PedNet Registry cohort compared with locus-specific genetic databases of the European Association for Haemophilia and Allied Disorders and the Centers for Disease Control and Prevention Hemophilia A or Hemophilia B Mutation Project

Background

Hemophilia A and B are caused by variants in the factor (F) VIII or FIX gene. Selective reporting may influence the distribution of variants reported in genetic databases.

Objectives

To compare the spectrum of F8 and F9 variants in an international population-based pediatric cohort (PedNet Registry) with the spectrum found in the European Association for Haemophilia and Allied Disorders (EAHAD) and the Centers for Disease Control and Prevention Hemophilia A or Hemophilia B Mutation Project (CHAMP/CHBMP) databases.

Methods

All patients registered in the PedNet Registry on January 1, 2021 were included in this study. As comparators, data from patients with severe hemophilia included in the CHAMP/CHBMP registry (US center data) and EAHAD were used.

Results

Genetic information was available for 1941 patients. Intron 22 inversion was present in 52% of patients with severe hemophilia A; frameshift (36%), missense (28%), and nonsense (20%) were the most frequent variants in patients with severe hemophilia A who were inversion-negative. The most frequent variants in severe hemophilia B were missense (48%). In nonsevere disease, most variants were missense variants (moderate hemophilia A: 91%; mild hemophilia A: 95%, moderate and mild hemophilia B: 86% each). Comparison with the databases demonstrated a higher proportion of missense variants associated with severe hemophilia B in EAHAD (68%) than in PedNet (48%) and CHBMP (46%).

Conclusion

The PedNet population-based cohort provides an alternative to the established databases, which collect data by selective reporting, as it is a well-maintained database covering the full spectrum of pathogenic F8 and F9 variants, and indicates the number of patients affected by each particular variant.

Computational analyses reveal fundamental properties of the AT structure related to thrombosis

Summary

Blood coagulation is a vital process for humans and other species. Following an injury to a blood vessel, a cascade of molecular signals is transmitted, inhibiting and activating more than a dozen coagulation factors and resulting in the formation of a fibrin clot that ceases the bleeding. In this process, antithrombin (AT), encoded by the SERPINC1 gene is a key player regulating the clotting activity and ensuring that it stops at the right time. In this sense, mutations to this factor often result in thrombosis-the excessive coagulation that leads to the potentially fatal formation of blood clots that obstruct veins. Although this process is well known, it is still unclear why even single residue substitutions to AT lead to drastically different phenotypes. In this study, to understand the effect of mutations throughout the AT structure, we created a detailed network map of this protein, where each node is an amino acid, and two amino acids are connected if they are in close proximity in the three-dimensional structure. With this simple and intuitive representation and a machine-learning framework trained using genetic information from more than 130 patients, we found that different types of thrombosis have emerging patterns that are readily identifiable. Together, these results demonstrate how clinical features, genetic data and in silico analysis are converging to enhance the diagnosis and treatment of coagulation disorders.

Supplementary information

Supplementary data are available at Bioinformatics Advances online.

Adoption of Machine Learning in Pharmacometrics: An Overview of Recent Implementations and Their Considerations

Pharmacometrics is a multidisciplinary field utilizing mathematical models of physiology, pharmacology, and disease to describe and quantify the interactions between medication and patient. As these models become more and more advanced, the need for advanced data analysis tools grows. Recently, there has been much interest in the adoption of machine learning (ML) algorithms. These algorithms offer strong function approximation capabilities and might reduce the time spent on model development. However, ML tools are not yet an integral part of the pharmacometrics workflow. The goal of this work is to discuss how ML algorithms have been applied in four stages of the pharmacometrics pipeline: data preparation, hypothesis generation, predictive modelling, and model validation. We will also discuss considerations before the use of ML algorithms with respect to each topic. We conclude by summarizing applications that hold potential for adoption by pharmacometricians.

The current role of artificial intelligence in hemophilia

INTRODUCTION

The utilization of artificial intelligence (AI) in hemophilia is still in its early phases.

AREAS COVERED

In this paper a review of the available information on AI in hemophilia has been performed, to better understand the relationship between hemophilia and AI. Regarding the physical elements of AI (robotics), robotic-assisted total knee arthroplasty and laparoscopic prostatectomy have been successfully performed in hemophilic patients. Concerning the virtual elements of AI, machine learning (ML) in hemophilia has been used with encouraging results for the following: prediction of disease severity, recognition of factor V as an essential modifier of thrombin generation in mild to moderate hemophilia A, development hemophilia-focused user-centered app, gene therapy, estimation of the risk of myocardial infarction, and identification of CRISPR/Cas9 nuclease off-target for the treatment of hemophilia. AI is an emerging reality that can produce a paradigm shift in hemophilia.

EXPERT OPINION

Various AI systems can facilitate clinical care for professionals, improving the diagnosis and treatment of hemophilia. However, AI systems still have many limitations and raise operational and ethical issues. AI systems should be integrated prudently and reasonably within the practitioner's workflow.

A Machine Learning Framework Predicts the Clinical Severity of Hemophilia B Caused by Point-Mutations

Blood coagulation is a vital physiological mechanism to stop blood loss following an injury to a blood vessel. This process starts immediately upon damage to the endothelium lining a blood vessel, and results in the formation of a platelet plug that closes the site of injury. In this repair operation, an essential component is the coagulation factor IX (FIX), a serine protease encoded by the F9 gene and whose deficiency causes hemophilia B. If not treated by prophylaxis or gene therapy, patients with this condition are at risk of life-threatening bleeding episodes. In this sense, a deep understanding of the FIX protein and its activated form (FIXa) is essential to develop efficient therapeutics. In this study, we used well-studied structural analysis techniques to create a residue interaction network of the FIXa protein. Here, the nodes are the amino acids of FIXa, and two nodes are connected by an edge if the two residues are in close proximity in the FIXa 3D structure. This representation accurately captured fundamental properties of each amino acid of the FIXa structure, as we found by validating our findings against hundreds of clinical reports about the severity of HB. Finally, we established a machine learning framework named HemB-Class to predict the effect of mutations of all FIXa residues to all other amino acids and used it to disambiguate several conflicting medical reports. Together, these methods provide a comprehensive map of the FIXa protein architecture and establish a robust platform for the rational design of FIX therapeutics.

Protein residue network analysis reveals fundamental properties of the human coagulation factor VIII

Hemophilia A is an X-linked inherited blood coagulation disorder caused by the production and circulation of defective coagulation factor VIII protein. People living with this condition receive either prophylaxis or on-demand treatment, and approximately 30% of patients develop inhibitor antibodies, a serious complication that limits treatment options. Although previous studies performed targeted mutations to identify important residues of FVIII, a detailed understanding of the role of each amino acid and their neighboring residues is still lacking. Here, we addressed this issue by creating a residue interaction network (RIN) where the nodes are the FVIII residues, and two nodes are connected if their corresponding residues are in close proximity in the FVIII protein structure. We studied the characteristics of all residues in this network and found important properties related to disease severity, interaction to other proteins and structural stability. Importantly, we found that the RIN-derived properties were in close agreement with in vitro and clinical reports, corroborating the observation that the patterns derived from this detailed map of the FVIII protein architecture accurately capture the biological properties of FVIII.