Comparison between an artificial neural network and logistic regression in predicting acute graft-vs-host disease after unrelated donor hematopoietic stem cell transplantation in thalassemia patients

Nutritional intervention with TGF-beta enriched food for special medical purposes (TGF-FSMP) is associated with a reduction of malnutrition, acute GVHD, pneumonia and may improve overall survival in patients undergoing allogeneic hematopoietic stem transplantation

Malnutrition in allogeneic stem cell transplant (allo-SCT) is associated with poor outcomes. Supplementation with Foods for Special Medical Purposes may be a valid alternative to enteral nutrition or total parental nutrition to reduce malnutrition in allo-SCT. In this study, 133 patients consecutively allo-transplanted were assessed for nutritional status by Patient- Generated Subjective Global Assessment (PG-SGA) and supplemented with TGF-beta enriched Food for Special Medical Purposes (TGF-FSMP). PG-SGA, gold standard for nutritional assessment in oncologic patients, was assessed at admission and on day 0, +7, +14, +21, and + 28 from transplant and categorized as follows: A = good nutritional status; B = moderate malnutrition; C = severe malnutrition. TGF-FSMP (Modulen-IBD) is currently used in Inflammatory Bowel Diseases (IBD) as primary nutritional support and in this study the dose was calculated according to BMI and total daily energy expenditure (TDEE). The patients assuming ≥50% of the prescribed TGF-FSMP dose were classified in Group A; the patients who received < 50% were included in Group B per protocol. The primary endpoint of the study was the assessment of the malnourished patients in Group A and B at day+28 after transplantation, according to the criteria of PG-SGA C categorization. At day +28 after transplant: i) patients in Group A were significantly less severely malnourished than patients in the Group B (21/76,28% vs 42/53, 79% respectively, OR 2.86 - CI 1.94-4.23 -, p = 0.000); ii) the incidence of severe (MAGIC II-IV) aGVHD and of any grade gastrointestinal (GI) aGVHD was higher in Group B than in Group A, (43% vs 21% p = 0.003) and (34.5% vs 9.2% p = 0.001); iii) Pneumonia was more frequent in the malnourished patients of Group B than in well/moderate nourished patients of Group A (52.7% vs 27.6% p = 0.002). In group A parenteral nutrition was avoided more frequently than in group B (67.5% vs 33.3% p = 0.000) and a median hospital stay of 27 days in comparison to 32 was reported (p = 0.006). The estimated median overall survival (OS) of the population was 33 months in Group A and 25.1 months in group B (p = 0.03). By multivariate and ANN analysis, TGF-FSMP TR < 50% assumption was significantly correlated with malnutrition, severe and GI aGVHD, pneumonia and reduced OS.

Machine learning applications and challenges in graft-versus-host disease: a scoping review

PURPOSE OF REVIEW

This review delves into the potential of artificial intelligence (AI), particularly machine learning (ML), in enhancing graft-versus-host disease (GVHD) risk assessment, diagnosis, and personalized treatment.

RECENT FINDINGS

Recent studies have demonstrated the superiority of ML algorithms over traditional multivariate statistical models in donor selection for allogeneic hematopoietic stem cell transplantation. ML has recently enabled dynamic risk assessment by modeling time-series data, an upgrade from the static, "snapshot" assessment of patients that conventional statistical models and older ML algorithms offer. Regarding diagnosis, a deep learning model, a subset of ML, can accurately identify skin segments affected with chronic GVHD with satisfactory results. ML methods such as Q-learning and deep reinforcement learning have been utilized to develop adaptive treatment strategies (ATS) for the personalized prevention and treatment of acute and chronic GVHD.

SUMMARY

To capitalize on these promising advancements, there is a need for large-scale, multicenter collaborations to develop generalizable ML models. Furthermore, addressing pertinent issues such as the implementation of stringent ethical guidelines is crucial before the widespread introduction of AI into GVHD care.

Malnutrition Prevention after Allogeneic Hematopoietic Stem Cell Transplantation (alloHSCT): A Prospective Explorative Interventional Study with an Oral Polymeric Formulation Enriched with Transforming Growth Factor Beta 2 (TGF-β2)

Malnutrition is common after allogeneic Hematopoietic Stem Cell Transplantation (alloHSCT), and interventions directed to correct nutritional status are warranted to improve transplant outcomes. In this prospective study, an oral polymeric formulation enriched with TGF-β2 (TE-OPF) was explored to correct malnutrition according to Patient-Generated Subjective Global Assessment (PG-SGA). TE-OPF was proposed to 51 consecutive patients who received transplants at our institution for hematological malignancies, and sufficient dose intake was established per protocol as at least 50% of the prescribed dose of TE-OPF: group A received adequate nutritional support; group B, inadequate. The study met the primary outcomes in terms of safety (no adverse events reported during TE-OPF intake except for its disgusting taste) and malnutrition (PG-SGA C 28 days after transplant): severely malnourished patients (PG-SGA C) accounted for 13% in group A and 88.9% in group B (p = 0.000). At the end of the study, after a median follow-up of 416 days, the estimated median Overall Survival (OS) was 734 days for well or moderately nourished patients (PG-SGA A/B) in comparison to 424 for malnourished patients (p = 0.03). Inadequate TE-OPF intake was associated with an increase in acute gastrointestinal Graft Versus Host Disease (GVHD) cumulative incidence (38% vs. 0% p = 0.006). A higher incidence of pneumonia was reported in group B (p = 0.006). IGF-1 levels at 14 and 28 days after transplant were significantly higher in group A and were associated with a lower incidence of acute GVHD (aGVHD). Higher subsets of B, T, and NK cells were found in group A, and a higher number of CD16+ NK cells was associated with a lower incidence of acute GVHD (p = 0.005) and increased survival at the end of the study (p = 0.023). Artificial neural network analysis suggested that inadequate TE-OPF intake, pneumonia, and sepsis significantly affected malnutrition 28 days after alloHSCT and survival 365 days after alloHSCT (normalized importance 100%, 82%, and 68%, respectively). In this exploratory and preliminary study, the use of TE-OPF appeared to reduce the incidence of malnutrition after alloHSCT, but larger and controlled studies are required.

A systematic review of the predicted outcomes related to hematopoietic stem cell transplantation: focus on applied machine learning methods' performance

INTRODUCTION

: Hematopoietic stem cell transplantation (HSCT) is a critical therapeutic procedure in blood diseases, and the investigation of HSCT data can provide valuable information. Machine learning (ML) techniques are novel and useful data analysis tools that have been applied in many studies to predict HSCT survival and estimate the risk of transplantation.

AREAS COVERED

: A systematic review was performed with a search of PubMed, Science Direct, Embase, Scopus, and the European Society for Blood and Marrow Transplantation, the Center for International Blood and Marrow Transplant Research, and the American Society for Transplantation and Cellular Therapy publications for articles published by September 2020.

EXPERT OPINION

: After investigating the results, 24 papers that met eligibility criteria were included in this study. The applied ML algorithms with the highest performance were Random Survival Forests (AUC=0.72) for survival-related, Random Survival Forests and Logistic Regression (AUC=0.77) for mortality-related, Deep Learning (AUC=0.8) for relapse, L2-Regularized Logistic Regression (AUC=0.66) for Acute-Graft Versus Host Disease, Random Survival Forests (AUC=0.88) for sepsis, Elastic-Net Regression (AUC=0.89) for cognitive impairment, and Bayesian Network (AUC=0.997) for oral mucositis outcome. This review reveals the potential of ML techniques to predict HSCT outcomes and apply them to developing clinical decision support systems.

Machine Learning Applications in the Diagnosis of Benign and Malignant Hematological Diseases

The use of machine learning (ML) and deep learning (DL) methods in hematology includes diagnostic, prognostic, and therapeutic applications. This increase is due to the improved access to ML and DL tools and the expansion of medical data. The utilization of ML remains limited in clinical practice, with some disciplines further along in their adoption, such as radiology and histopathology. In this review, we discuss the current uses of ML in diagnosis in the field of hematology, including image-recognition, laboratory, and genomics-based diagnosis. Additionally, we provide an introduction to the fields of ML and DL, highlighting current trends, limitations, and possible areas of improvement.

A Review on the Role of Artificial Intelligence in Stem Cell Therapy: An Initiative for Modern Medicines

Stem Cells (SCs) show a wide range of applications in the treatment of numerous diseases, including neurodegenerative diseases, diabetes, cardiovascular diseases, cancer, etc. SC related research has gained popularity owing to the unique characteristics of self-renewal and differentiation. Artificial Intelligence (AI), an emerging field of computer science and engineering, has shown potential applications in different fields like robotics, agriculture, home automation, healthcare, banking, and transportation since its invention. This review aims to describe the various applications of AI in SC biology, including understanding the behavior of SCs, recognizing individual cell type before undergoing differentiation, characterization of SCs using mathematical models and prediction of mortality risk associated with SC transplantation. This review emphasizes the role of neural networks in SC biology and further elucidates the concepts of machine learning and deep learning and their applications in SC research.

Prediction of future gene expression profile by analyzing its past variation pattern

A number of initial Hematopoietic Stem Cells (HSC) are considered in a container that are able to divide into HSCs or differentiate into various types of descendant cells. In this paper, a method is designed to predict an approximate gene expression profile (GEP) for future descendant cells resulted from HSC division/differentiation. First, the GEP prediction problem is modeled into a multivariate time series prediction problem. A novel method called EHSCP (Extended Hematopoietic Stem Cell Prediction) is introduced which is an artificial neural machine to solve the problem. EHSCP accepts the initial sequence of measured GEPs as input and predicts GEPs of future descendant cells. This prediction can be performed for multiple stages of cell division/differentiation. EHSCP considers the GEP sequence as time series and computes correlation between input time series. Two novel artificial neural units called PLSTM (Parametric Long Short Term Memory) and MILSTM (Multi-Input LSTM) are designed. PLSTM makes EHSCP able to consider this correlation in output prediction. Since there exist thousands of time series in GEP prediction, a hierarchical encoder is proposed that computes this correlation using 101 MILSTMs. EHSCP is trained using 155 datasets and is evaluated on 39 test datasets. These evaluations show that EHSCP surpasses existing methods in terms of prediction accuracy and number of correctly-predicted division/differentiation stages. In these evaluations, number of correctly-predicted stages in EHSCP was 128 when as many as 8 initial stages were given.

Artificial Intelligence Approaches in Hematopoietic Cell Transplantation: A Review of the Current Status and Future Directions

The evidence-based literature on healthcare is currently expanding exponentially. The opportunities provided by the advancement in artificial intelligence (AI) tools such as machine learning are appealing in tackling many of the current healthcare challenges. Thus, AI integration is expanding in most fields of healthcare, including the field of hematology. This study aims to review the current applications of AI in the field of hematopoietic cell transplantation (HCT). A literature search was done involving the following databases: Ovid MEDLINE, including In-Process and other non-indexed citations, and Google Scholar. The abstracts of the following professional societies were also screened: American Society of Hematology, American Society for Blood and Marrow Transplantation, and European Society for Blood and Marrow Transplantation. The literature review showed that the integration of AI in the field of HCT has grown remarkably in the last decade and offers promising avenues in diagnosis and prognosis in HCT populations targeting both pre- and post-transplant challenges. Studies of AI integration in HCT have many limitations that include poorly tested algorithms, lack of generalizability, and limited use of different AI tools. Machine learning techniques in HCT are an intense area of research that needs much development and extensive support from hematology and HCT societies and organizations globally as we believe that this will be the future practice paradigm.

Predicting ventriculoperitoneal shunt infection in children with hydrocephalus using artificial neural network

OBJECTIVES

The relationships between shunt infection and predictive factors have not been previously investigated using Artificial Neural Network (ANN) model. The aim of this study was to develop an ANN model to predict shunt infection in a group of children with shunted hydrocephalus.

MATERIALS AND METHODS

Among more than 800 ventriculoperitoneal shunt procedures which had been performed between April 2000 and April 2011, 68 patients with shunt infection and 80 controls that fulfilled a set of meticulous inclusion/exclusion criteria were consecutively enrolled. Univariate analysis was performed for a long list of risk factors, and those with p value < 0.2 were used to create ANN and logistic regression (LR) models.

RESULTS

Five variables including birth weight, age at the first shunting, shunt revision, prematurity, and myelomeningocele were significantly associated with shunt infection via univariate analysis, and two other variables (intraventricular hemorrhage and coincided infections) had a p value of less than 0.2. Using these seven input variables, ANN and LR models predicted shunt infection with an accuracy of 83.1 % (AUC; 91.98 %, 95 % CI) and 55.7 % (AUC; 76.5, 95 % CI), respectively. The contribution of the factors in the predictive performance of ANN in descending order was history of shunt revision, low birth weight (under 2000 g), history of prematurity, the age at the first shunt procedure, history of intraventricular hemorrhage, history of myelomeningocele, and coinfection.

CONCLUSION

The findings show that artificial neural networks can predict shunt infection with a high level of accuracy in children with shunted hydrocephalus. Also, the contribution of different risk factors in the prediction of shunt infection can be determined using the trained network.

What Unrelated Hematopoietic Stem Cell Transplantation in Thalassemia Taught us about Transplant Immunogenetics

Although the past few decades have shown an improvement in the survival and complication-free survival rates in patients with beta-thalassemia major and gene therapy is already at an advanced stage of experimentation, hematopoietic stem cell transplantation (HSCT) continues to be the only effective and realistic approach to the cure of this chronic non-malignant disease. Historically, human leukocyte antigen (HLA)-matched siblings have been the preferred source of donor cells owing to superior outcomes compared with HSCT from other sources. Nowadays, the availability of an international network of voluntary stem cell donor registries and cord blood banks has significantly increased the odds of finding a suitable HLA matched donor. Stringent immunogenetic criteria for donor selection have made it possible to achieve overall survival (OS) and thalassemia-free survival (TFS) rates comparable to those of sibling transplants. However, acute and chronic graft-versus-host disease (GVHD) remains the most important complication in unrelated HSCT in thalassemia, leading to significant rates of morbidity and mortality for a chronic non-malignant disease. A careful immunogenetic assessment of donors and recipients makes it possible to individualize appropriate strategies for its prevention and management. This review provides an overview of recent insights about immunogenetic factors involved in GVHD, which seem to have a potential role in the outcome of transplantation for thalassemia.

Role of Pharmacogenetics in Hematopoietic Stem Cell Transplantation Outcome in Children

Hematopoietic stem cell transplantation (HSCT) is an established therapeutic procedure for several congenital and acquired disorders, both malignant and nonmalignant. Despite the great improvements in HSCT clinical practices over the last few decades, complications, such as graft vs. host disease (GVHD) and sinusoidal obstructive syndrome (SOS), are still largely unpredictable and remain the major causes of morbidity and mortality. Both donor and patient genetic background might influence the success of bone marrow transplantation and could at least partially explain the inter-individual variability in HSCT outcome. This review summarizes some of the recent studies on candidate gene polymorphisms in HSCT, with particular reference to pediatric cohorts. The interest is especially focused on pharmacogenetic variants affecting myeloablative and immunosuppressive drugs, although genetic traits involved in SOS susceptibility and transplant-related mortality are also reviewed.

Absence of activating killer immunoglobulin-like receptor genes combined with hepatitis C viral genotype is predictive of hepatocellular carcinoma

Killer immunoglobulin-like receptors and their human leukocyte antigen class I ligands have a critical role in natural killer cell response to viral pathogens and tumors. To investigate whether killer immunoglobulin-like receptor genes could influence the chronic course of hepatitis C virus infection and/or progression to hepatocellular carcinoma we retrospectively analyzed a cohort of 228 patients transplanted for hepatitis C virus-induced cirrhotic end stage liver disease, combined or not with hepatocellular carcinoma. We found that patients completely lacking activating killer immunoglobulin-like receptor genes had a high risk of developing hepatocellular carcinoma. Hepatitis C viral genotype and viral load are other risk factors that can influence the course of chronic hepatitis C virus infection. In our study, the risk conferred by hepatitis C viral genotypes was enhanced in patients lacking activating killer immunoglobulin-like receptors. These results point to an important role for activating killer immunoglobulin-like receptors in the control of hepatitis C virus infection and progression to hepatocellular carcinoma. In clinical practice, assessment of killer immunoglobulin-like receptor and hepatitis C viral genotype combinations should allow for more accurate monitoring of patients with chronic hepatitis C virus infection.

Interactions between killer immunoglobulin-like receptors and their human leucocyte antigen Class I ligands influence the outcome of unrelated haematopoietic stem cell transplantation for thalassaemia: a novel predictive algorithm

In a study conducted on 114 patients undergoing unrelated donor haematopoietic stem cell transplantation (HSCT) for thalassaemia, we observed that the lack of activating killer immunoglobulin-like receptors (KIRs) on donor natural killer (NK) cells significantly increased the risk of graft-versus-host disease (GvHD) [hazard risk (HR) 4.2, 95% confidence interval (CI) 1.7-10.1, P = 0.002] and transplantation-related mortality (HR 4.7, 95% CI 1.6-14.2, P = 0.01). The risk of GvHD furthermore increased when recipients heterozygous for HLA-C KIR ligand groups (C1/C2) were transplanted from donors completely lacking activating KIRs (HR 6.1, 95% CI 1.9-19.2, P = 0.002). We also found that the risk of rejection was highest when the recipient was homozygous for the C2 HLA-KIR ligand group and the donor carried two or more activating KIRs (HR 6.8, 95% CI 1.9-24.4, P = 0.005). By interpolating the number of donor activating KIRs with recipient HLA-C KIR ligands, we created an algorithm capable of stratifying patients according to the immunogenetic risk of complications following unrelated HSCT. In clinical practice, this predictive tool could serve as an important supplement to clinical judgement and decision-making.

Recognition of morphometric vertebral fractures by artificial neural networks: analysis from GISMO Lombardia Database

BACKGROUND

It is known that bone mineral density (BMD) predicts the fracture's risk only partially and the severity and number of vertebral fractures are predictive of subsequent osteoporotic fractures (OF). Spinal deformity index (SDI) integrates the severity and number of morphometric vertebral fractures. Nowadays, there is interest in developing algorithms that use traditional statistics for predicting OF. Some studies suggest their poor sensitivity. Artificial Neural Networks (ANNs) could represent an alternative. So far, no study investigated ANNs ability in predicting OF and SDI. The aim of the present study is to compare ANNs and Logistic Regression (LR) in recognising, on the basis of osteoporotic risk-factors and other clinical information, patients with SDI≥1 and SDI≥5 from those with SDI = 0.

METHODOLOGY

We compared ANNs prognostic performance with that of LR in identifying SDI≥1/SDI≥5 in 372 women with postmenopausal-osteoporosis (SDI≥1, n = 176; SDI = 0, n = 196; SDI≥5, n = 51), using 45 variables (44 clinical parameters plus BMD). ANNs were allowed to choose relevant input data automatically (TWIST-system-Semeion). Among 45 variables, 17 and 25 were selected by TWIST-system-Semeion, in SDI≥1 vs SDI = 0 (first) and SDI≥5 vs SDI = 0 (second) analysis. In the first analysis sensitivity of LR and ANNs was 35.8% and 72.5%, specificity 76.5% and 78.5% and accuracy 56.2% and 75.5%, respectively. In the second analysis, sensitivity of LR and ANNs was 37.3% and 74.8%, specificity 90.3% and 87.8%, and accuracy 63.8% and 81.3%, respectively.

CONCLUSIONS

ANNs showed a better performance in identifying both SDI≥1 and SDI≥5, with a higher sensitivity, suggesting its promising role in the development of algorithm for predicting OF.