Predicting Acute Graft-Versus-Host Disease Using Machine Learning and Longitudinal Vital Sign Data From Electronic Health Records

Leveraging machine learning for predicting acute graft-versus-host disease grades in allogeneic hematopoietic cell transplantation for T-cell prolymphocytic leukaemia

Orphan diseases, exemplified by T-cell prolymphocytic leukemia, present inherent challenges due to limited data availability and complexities in effective care. This study delves into harnessing the potential of machine learning to enhance care strategies for orphan diseases, specifically focusing on allogeneic hematopoietic cell transplantation (allo-HCT) in T-cell prolymphocytic leukemia. The investigation evaluates how varying numbers of variables impact model performance, considering the rarity of the disease. Utilizing data from the Center for International Blood and Marrow Transplant Research, the study scrutinizes outcomes following allo-HCT for T-cell prolymphocytic leukemia. Diverse machine learning models were developed to forecast acute graft-versus-host disease (aGvHD) occurrence and its distinct grades post-allo-HCT. Assessment of model performance relied on balanced accuracy, F1 score, and ROC AUC metrics. The findings highlight the Linear Discriminant Analysis (LDA) classifier achieving the highest testing balanced accuracy of 0.58 in predicting aGvHD. However, challenges arose in its performance during multi-class classification tasks. While affirming the potential of machine learning in enhancing care for orphan diseases, the study underscores the impact of limited data and disease rarity on model performance.

Prediction of hospital-acquired influenza using machine learning algorithms: a comparative study

BACKGROUND

Hospital-acquired influenza (HAI) is under-recognized despite its high morbidity and poor health outcomes. The early detection of HAI is crucial for curbing its transmission in hospital settings.

AIM

This study aimed to investigate factors related to HAI, develop predictive models, and subsequently compare them to identify the best performing machine learning algorithm for predicting the occurrence of HAI.

METHODS

This retrospective observational study was conducted in 2022 and included 111 HAI and 73,748 non-HAI patients from the 2011-2012 and 2019-2020 influenza seasons. General characteristics, comorbidities, vital signs, laboratory and chest X-ray results, and room information within the electronic medical record were analysed. Logistic Regression (LR), Random Forest (RF), Extreme Gradient Boosting (XGB), and Artificial Neural Network (ANN) techniques were used to construct the predictive models. Employing randomized allocation, 80% of the dataset constituted the training set, and the remaining 20% comprised the test set. The performance of the developed models was assessed using metrics such as the area under the receiver operating characteristic curve (AUC), the count of false negatives (FN), and the determination of feature importance.

RESULTS

Patients with HAI demonstrated notable differences in general characteristics, comorbidities, vital signs, laboratory findings, chest X-ray result, and room status compared to non-HAI patients. Among the developed models, the RF model demonstrated the best performance taking into account both the AUC (83.3%) and the occurrence of FN (four). The most influential factors for prediction were staying in double rooms, followed by vital signs and laboratory results.

CONCLUSION

This study revealed the characteristics of patients with HAI and emphasized the role of ventilation in reducing influenza incidence. These findings can aid hospitals in devising infection prevention strategies, and the application of machine learning-based predictive models especially RF can enable early intervention to mitigate the spread of influenza in healthcare settings.

Using Targeted Transcriptome and Machine Learning of Pre- and Post-Transplant Bone Marrow Samples to Predict Acute Graft-versus-Host Disease and Overall Survival after Allogeneic Stem Cell Transplantation

Acute graft-versus-host disease (aGvHD) remains a major cause of morbidity and mortality after allogeneic hematopoietic stem cell transplantation (HSCT). We performed RNA analysis of 1408 candidate genes in bone marrow samples obtained from 167 patients undergoing HSCT. RNA expression data were used in a machine learning algorithm to predict the presence or absence of aGvHD using either random forest or extreme gradient boosting algorithms. Patients were randomly divided into training (2/3 of patients) and validation (1/3 of patients) sets. Using post-HSCT RNA data, the machine learning algorithm selected 92 genes for predicting aGvHD that appear to play a role in PI3/AKT, MAPK, and FOXO signaling, as well as microRNA. The algorithm selected 20 genes for predicting survival included genes involved in MAPK and chemokine signaling. Using pre-HSCT RNA data, the machine learning algorithm selected 400 genes and 700 genes predicting aGvHD and overall survival, but candidate signaling pathways could not be specified in this analysis. These data show that NGS analyses of RNA expression using machine learning algorithms may be useful biomarkers of aGvHD and overall survival for patients undergoing HSCT, allowing for the identification of major signaling pathways associated with HSCT outcomes and helping to dissect the complex steps involved in the development of aGvHD. The analysis of pre-HSCT bone marrow samples may lead to pre-HSCT interventions including choice of remission induction regimens and modifications in patient health before HSCT.

Longitudinal clinical data improve survival prediction after hematopoietic cell transplantation using machine learning

ABSTRACT

Serial prognostic evaluation after allogeneic hematopoietic cell transplantation (allo-HCT) might help identify patients at high risk of lethal organ dysfunction. Current prediction algorithms based on models that do not incorporate changes to patients' clinical condition after allo-HCT have limited predictive ability. We developed and validated a robust risk-prediction algorithm to predict short- and long-term survival after allo-HCT in pediatric patients that includes baseline biological variables and changes in the patients' clinical status after allo-HCT. The model was developed using clinical data from children and young adults treated at a single academic quaternary-care referral center. The model was created using a randomly split training data set (70% of the cohort), internally validated (remaining 30% of the cohort) and then externally validated on patient data from another tertiary-care referral center. Repeated clinical measurements performed from 30 days before allo-HCT to 30 days afterwards were extracted from the electronic medical record and incorporated into the model to predict survival at 100 days, 1 year, and 2 years after allo-HCT. Naïve-Bayes machine learning models incorporating longitudinal data were significantly better than models constructed from baseline variables alone at predicting whether patients would be alive or deceased at the given time points. This proof-of-concept study demonstrates that unlike traditional prognostic tools that use fixed variables for risk assessment, incorporating dynamic variability using clinical and laboratory data improves the prediction of mortality in patients undergoing allo-HCT.

A dynamic time-to-event model for prediction of acute graft-versus-host disease in patients after allogeneic hematopoietic stem cell transplantation

BACKGROUND

Acute graft-versus-host disease (aGvHD) is a major cause of death for patients following allogeneic hematopoietic stem cell transplantation (HSCT). Effective management of moderate to severe aGvHD remains challenging despite recent advances in HSCT, emphasizing the importance of prophylaxis and risk factor identification.

METHODS

In this study, we analyzed data from 1479 adults who underwent HSCT between 2005 and 2017 to investigate the effects of aGvHD prophylaxis and time-dependent risk factors on the development of grades II-IV aGvHD within 100 days post-HSCT.

RESULTS

Using a dynamic longitudinal time-to-event model, we observed a non-monotonic baseline hazard overtime with a low hazard during the first few days and a maximum hazard at day 17, described by Bateman function with a mean transit time of approximately 11 days. Multivariable analysis revealed significant time-dependent effects of white blood cell counts and cyclosporine A exposure as well as static effects of female donors for male recipients, patients with matched related donors, conditioning regimen consisting of fludarabine plus total body irradiation, and patient age in recipients of grafts from related donors on the risk to develop grades II-IV aGvHD. Additionally, we found that higher cumulative hazard on day 7 after allo-HSCT are associated with an increased incidence of grades II-IV aGvHD within 100 days indicating that an individual assessment of the cumulative hazard on day 7 could potentially serve as valuable predictor for later grades II-IV aGvHD development. Using the final model, stochastic simulations were performed to explore covariate effects on the cumulative incidence over time and to estimate risk ratios.

CONCLUSION

Overall, the presented model showed good descriptive and predictive performance and provides valuable insights into the interplay of multiple static and time-dependent risk factors for the prediction of aGvHD.

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.

Assessment of risk factors for acute graft-versus-host disease post-hematopoietic stem cell transplantation: a retrospective study based on a proportional odds model using a nonlinear mixed-effects model

Background

Acute graft-versus-host disease (aGVHD) is a major complication following hematopoietic stem cell transplantation (HSCT).

Objective

This study aimed to explore the risk factors for the incidence of aGVHD in patients post-HSCT.

Design

This was a retrospective study.

Methods

A total of 407 patients were enrolled. The patients' data were recorded from the medical records. The exposure of cyclosporine was estimated based on a population pharmacokinetics model. The occurrence of aGVHD was clinically graded and staged in severity from grades I to IV. A proportional odds model that estimated the cumulative probabilities of aGVHD was used to analyze the data using a nonlinear mixed-effects model. Then, the model parameters and plausibility were evaluated by bootstrap and visual predictive checks.

Results

The typical probabilities were 18.9% and 17.9% for grade II and grades III-IV, respectively. The incidence of grade II and grade III-IV aGVHD for human leukocyte antigen (HLA) haplo sibling donor patients was higher than that for HLA-matched donor patients. The incidence of grade II and grade III-IV aGVHD decreased with increasing early cyclosporine trough concentration; however, cyclosporine exposure was not associated with the incidence of aGVHD.

Conclusion

HLA matching and early cyclosporine trough concentration were important factors for the occurrence of aGVHD.

Machine Learning for the Prediction of Survival Post-Allogeneic Hematopoietic Cell Transplantation: A Single-Center Experience

INTRODUCTION

Prediction of outcomes following allogeneic hematopoietic cell transplantation (HCT) remains a major challenge. Machine learning (ML) is a computational procedure that may facilitate the generation of HCT prediction models. We sought to investigate the prognostic potential of multiple ML algorithms when applied to a large single-center allogeneic HCT database.

METHODS

Our registry included 2,697 patients that underwent allogeneic HCT from January 1976 to December 2017. 45 pretransplant baseline variables were included in the predictive assessment of each ML algorithm on overall survival (OS) as determined by area under the curve (AUC). Pretransplant variables used in the EBMT ML study (Shouval et al., 2015) were used as a benchmark for comparison.

RESULTS

On the entire dataset, the random forest (RF) algorithm performed best (AUC 0.71 ± 0.04) compared to the second-best model, logistic regression (LR) (AUC = 0.69 ± 0.04) (p < 0.001). Both algorithms demonstrated improved AUC scores using all 45 variables compared to the limited variables examined by the EBMT study. Survival at 100 days post-HCT using RF on the full dataset discriminated patients into different prognostic groups with different 2-year OS (p < 0.0001). We then examined the ML methods that allow for significant individual variable identification, including LR and RF, and identified matched related donors (HR = 0.49, p < 0.0001), increasing TBI dose (HR = 1.60, p = 0.006), increasing recipient age (HR = 1.92, p < 0.0001), higher baseline Hb (HR = 0.59, p = 0.0002), and increased baseline FEV1 (HR = 0.73, p = 0.02), among others.

CONCLUSION

The application of multiple ML techniques on single-center allogeneic HCT databases warrants further investigation and may provide a useful tool to identify variables with prognostic potential.

Elaborating the potential of Artificial Intelligence in automated CAR-T cell manufacturing

This paper discusses the challenges of producing CAR-T cells for cancer treatment and the potential for Artificial Intelligence (AI) for its improvement. CAR-T cell therapy was approved in 2018 as the first Advanced Therapy Medicinal Product (ATMP) for treating acute leukemia and lymphoma. ATMPs are cell- and gene-based therapies that show great promise for treating various cancers and hereditary diseases. While some new ATMPs have been approved, ongoing clinical trials are expected to lead to the approval of many more. However, the production of CAR-T cells presents a significant challenge due to the high costs associated with the manufacturing process, making the therapy very expensive (approx. $400,000). Furthermore, autologous CAR-T therapy is limited to a make-to-order approach, which makes scaling economical production difficult. First attempts are being made to automate this multi-step manufacturing process, which will not only directly reduce the high manufacturing costs but will also enable comprehensive data collection. AI technologies have the ability to analyze this data and convert it into knowledge and insights. In order to exploit these opportunities, this paper analyses the data potential in the automated CAR-T production process and creates a mapping to the capabilities of AI applications. The paper explores the possible use of AI in analyzing the data generated during the automated process and its capabilities to further improve the efficiency and cost-effectiveness of CAR-T cell production.

A convolutional neural network-based model that predicts acute graft-versus-host disease after allogeneic hematopoietic stem cell transplantation

BACKGROUND

Forecasting acute graft-versus-host disease (aGVHD) after allogeneic hematopoietic stem cell transplantation (HSCT) is highly challenging with conventional statistical techniques due to complex parameters and their interactions. The primary object of this study was to establish a convolutional neural network (CNN)-based prediction model for aGVHD.

METHOD

We analyzed adult patients who underwent allogeneic HSCT between 2008 and 2018, using the Japanese nationwide registry database. The CNN algorithm, equipped with a natural language processing technique and an interpretable explanation algorithm, was applied to develop and validate prediction models.

RESULTS

Here, we evaluate 18,763 patients between 16 and 80 years of age (median, 50 years). In total, grade II-IV and grade III-IV aGVHD is observed among 42.0% and 15.6%. The CNN-based model eventually allows us to calculate a prediction score of aGVHD for an individual case, which is validated to distinguish the high-risk group of aGVHD in the test cohort: cumulative incidence of grade III-IV aGVHD at Day 100 after HSCT is 28.8% for patients assigned to a high-risk group by the CNN model, compared to 8.4% among low-risk patients (hazard ratio, 4.02; 95% confidence interval, 2.70-5.97; p < 0.01), suggesting high generalizability. Furthermore, our CNN-based model succeeds in visualizing the learning process. Moreover, contributions of pre-transplant parameters other than HLA information to the risk of aGVHD are determined.

CONCLUSIONS

Our results suggest that CNN-based prediction provides a faithful prediction model for aGVHD, and can serve as a valuable tool for decision-making in clinical practice.

A model based on machine learning for the prediction of cyclosporin A trough concentration in Chinese allo-HSCT patients

BACKGROUND

Cyclosporin A is a calcineurin inhibitor which has a narrow therapeutic window and high interindividual variability. Various population pharmacokinetic models have been reported; however, professional software and technical personnel were needed and the variables of the models were limited. Therefore, the aim of this study was to establish a model based on machine learning to predict CsA trough concentrations in Chinese allo-HSCT patients.

METHODS

A total of 7874 cases of CsA therapeutic drug monitoring data from 2069 allo-HSCT patients were retrospectively included. Sequential forward selection was used to select variable subsets, and eight different algorithms were applied to establish the prediction model.

RESULTS

XGBoost exhibited the highest prediction ability. Except for the variables that were identified by previous studies, some rarely reported variables were found, such as norethindrone, WBC, PAB, and hCRP. The prediction accuracy within ±30% of the actual trough concentration was above 0.80, and the predictive ability of the models was demonstrated to be effective in external validation.

CONCLUSION

In this study, models based on machine learning technology were established to predict CsA levels 3-4 days in advance during the early inpatient phase after HSCT. A new perspective for CsA clinical application is provided.

Pre-Transplant Prediction of Acute Graft-versus-Host Disease Using the Gut Microbiome

Gut microbiota is thought to influence host responses to allogeneic hematopoietic stem cell transplantation (aHSCT). Recent evidence points to this post-transplant for acute graft-versus-host disease (aGvHD). We asked whether any such association might be found pre-transplant and conducted a metagenome-wide association study (MWAS) to explore. Microbial abundance profiles were estimated using ensembles of Kaiju, Kraken2, and DeepMicrobes calls followed by dimensionality reduction. The area under the curve (AUC) was used to evaluate classification of the samples (aGvHD vs. none) using an elastic net to test the relevance of metagenomic data. Clinical data included the underlying disease (leukemia vs. other hematological malignancies), recipient age, and sex. Among 172 aHSCT patients of whom 42 developed aGVHD post transplantation, a total of 181 pre-transplant tool samples were analyzed. The top performing model predicting risk of aGVHD included a reduced species profile (AUC = 0.672). Beta diversity (37% in Jaccard's Nestedness by mean fold change, p < 0.05) was lower in those developing aGvHD. Ten bacterial species including Prevotella and Eggerthella genera were consistently found to associate with aGvHD in indicator species analysis, as well as relief and impurity-based algorithms. The findings support the hypothesis on potential associations between gut microbiota and aGvHD based on a data-driven approach to MWAS. This highlights the need and relevance of routine stool collection for the discovery of novel biomarkers.

Leukemia can be Effectively Early Predicted in Routine Physical Examination with the Assistance of Machine Learning Models

Objectives. The diagnosis of leukemia relies very much on the results of bone marrow examinations, which is never generally performed in routine physical examination. In many rural areas even community hospitals and primary care clinics, the lack of hematological specialist and facility does not allow a definite diagnosis of leukemia. Thus, there will be a significant benefit if machine learning (ML) models could help early predict leukemia using preliminary blood test data in a routine physical examination in community hospitals to save time before a definite diagnosis. Methods. We collected the routine physical examination data of 1230 newly diagnosed leukemia patients and 1300 healthy people. We trained and tested 3 machine learning (ML) models including linear support vector machine (LSVM), random forest (RF), and XGboost models. We not only examined the accordance between model results and statistical analysis of the input data but also examined the consistency of model accuracy scores and relative importance order of model factors with regard to different input data sets and different model arguments to check the applicability of both the models and the input data. Results. Generally, the RF and XGboost models give more identical, consistent, and robust relative importance order of factors that is also accordant with the statistical analysis, while the LSVM gives much different and nonsense orders for different inputs. Results of the RF and XGboost models show that (1) generally, the models achieve accuracy scores above 0.9, indicating effective identification of leukemia, and (2) the top three factors that contribute most to the identification of leukemia include red blood cell (RBC), hematocrit (HCT), and white blood cell (WBC), while the other factors contribute relatively less. Conclusions. This study shows a feasible case example for early identification of leukemia using routine physical examination data with the assistance of ML models, which can be conveniently, cheaply, and widely applied in community hospitals or primary care clinics to save time before definite diagnosis; however, more studies are still needed to validate the applicability of more ML models to a larger variety of input data sets.

Time-dependent prediction of mortality and cytomegalovirus reactivation after allogeneic hematopoietic cell transplantation using machine learning

Allogeneic hematopoietic cell transplantation (HCT) effectively treats high-risk hematologic diseases but can entail HCT-specific complications, which may be minimized by appropriate patient management, supported by accurate, individual risk estimation. However, almost all HCT risk scores are limited to a single risk assessment before HCT without incorporation of additional data. We developed machine learning models that integrate both baseline patient data and time-dependent laboratory measurements to individually predict mortality and cytomegalovirus (CMV) reactivation after HCT at multiple time points per patient. These gradient boosting machine models provide well-calibrated, time-dependent risk predictions and achieved areas under the receiver-operating characteristic of 0.92 and 0.83 and areas under the precision-recall curve of 0.58 and 0.62 for prediction of mortality and CMV reactivation, respectively, in a 21-day time window. Both models were successfully validated in a prospective, non-interventional study and performed on par with expert hematologists in a pilot comparison.

Dynamic forecasting of severe acute graft-versus-host disease after transplantation

Forecasting of severe acute graft-versus-host disease (aGVHD) after transplantation is a challenging 'large p, small n' problem that suffers from nonuniform data sampling. We propose a dynamic probabilistic algorithm, daGOAT, that accommodates sampling heterogeneity, integrates multidimensional clinical data and continuously updates the daily risk score for severe aGVHD onset within a two-week moving window. In the studied cohorts, the cross-validated area under the receiver operator characteristic curve (AUROC) of daGOAT rose steadily after transplantation and peaked at ≥0.78 in both the adult and pediatric cohorts, outperforming the two-biomarker MAGIC score, three-biomarker Ann Arbor score, peri-transplantation features-based models and XGBoost. Simulation experiments indicated that the daGOAT algorithm is well suited for short time-series scenarios where the underlying process for event generation is smooth, multidimensional and where there are frequent and irregular data missing. daGOAT's broader utility was demonstrated by performance testing on a remotely different task, that is, prediction of imminent human postural change based on smartphone inertial sensor time-series data.

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.

Use of Automated Thematic Annotations for Small Data Sets in a Psychotherapeutic Context: Systematic Review of Machine Learning Algorithms

BACKGROUND

A growing body of literature has detailed the use of qualitative analyses to measure the therapeutic processes and intrinsic effectiveness of psychotherapies, which yield small databases. Nonetheless, these approaches have several limitations and machine learning algorithms are needed.

OBJECTIVE

The objective of this study is to conduct a systematic review of the use of machine learning for automated text classification for small data sets in the fields of psychiatry, psychology, and social sciences. This review will identify available algorithms and assess if automated classification of textual entities is comparable to the classification done by human evaluators.

METHODS

A systematic search was performed in the electronic databases of Medline, Web of Science, PsycNet (PsycINFO), and Google Scholar from their inception dates to 2021. The fields of psychiatry, psychology, and social sciences were selected as they include a vast array of textual entities in the domain of mental health that can be reviewed. Additional records identified through cross-referencing were used to find other studies.

RESULTS

This literature search identified 5442 articles that were eligible for our study after the removal of duplicates. Following abstract screening, 114 full articles were assessed in their entirety, of which 107 were excluded. The remaining 7 studies were analyzed. Classification algorithms such as naive Bayes, decision tree, and support vector machine classifiers were identified. Support vector machine is the most used algorithm and best performing as per the identified articles. Prediction classification scores for the identified algorithms ranged from 53%-91% for the classification of textual entities in 4-7 categories. In addition, 3 of the 7 studies reported an interjudge agreement statistic; these were consistent with agreement statistics for text classification done by human evaluators.

CONCLUSIONS

A systematic review of available machine learning algorithms for automated text classification for small data sets in several fields (psychiatry, psychology, and social sciences) was conducted. We compared automated classification with classification done by human evaluators. Our results show that it is possible to automatically classify textual entities of a transcript based solely on small databases. Future studies are nevertheless needed to assess whether such algorithms can be implemented in the context of psychotherapies.

Next-Generation Implementation of Chimeric Antigen Receptor T-Cell Therapy Using Digital Health

Chimeric antigen receptor T-cell (CAR-T) therapy is a paradigm-shifting immunotherapy modality in oncology; however, unique toxicities such as cytokine release syndrome (CRS) and immune effector cell-associated neurotoxicity syndrome limit its ability to be implemented more widely in the outpatient setting or at smaller-volume centers. Three operational challenges with CAR-T therapy include the following: (1) the logistics of toxicity monitoring, ie, with frequent vital sign checks and neurologic assessments; (2) the specialized knowledge required for toxicity management, particularly with regard to CRS and immune effector cell-associated neurotoxicity syndrome; and (3) the need for high-quality symptomatic and supportive care during this intensive period. In this review, we explore potential niches for digital innovations that can improve the implementation of CAR-T therapy in each of these domains. These tools include patient-facing technologies and provider-facing platforms: for example, wearable devices and mobile health apps to screen for fevers and encephalopathy, electronic patient-reported outcome assessments-based workflows to assist with symptom management, machine learning algorithms to predict emerging CRS in real time, clinical decision support systems to assist with toxicity management, and digital coaching to help maintain wellness. Televisits, which have grown in prominence since the novel coronavirus pandemic, will continue to play a key role in the monitoring and management of CAR-T-related toxicities as well. Limitations of these strategies include the need to ensure care equity and stakeholder buy-in, both operationally and financially. Nevertheless, once developed and validated, the next-generation implementation of CAR-T therapy using these digital tools may improve both its safety and accessibility.

An Intelligent Clinical Decision Support System for Predicting Acute Graft-versus-host Disease (aGvHD) following Allogeneic Hematopoietic Stem Cell Transplantation

Background:

Acute graft-versus-host disease (aGvHD) is a complex and often multisystem disease that causes morbidity and mortality in 35% of patients receiving allogeneic hematopoietic stem cell transplantation (AHSCT).

Objective:

This study aimed to implement a Clinical Decision Support System (CDSS) for predicting aGvHD following AHSCT on the transplantation day.

Material and Methods:

In this developmental study, the data of 182 patients with 31 attributes, which referred to Taleghani Hospital Tehran, Iran during 2009–2017, were analyzed by machine learning (ML) algorithms which included XGBClassifier, HistGradientBoostingClassifier, AdaBoostClassifier, and RandomForestClassifier. The criteria measurement used to evaluate these algorithms included accuracy, sensitivity, and specificity. Using the machine learning developed model, a CDSS was implemented. The performance of the CDSS was evaluated by Cohen’s Kappa coefficient.

Results:

Of the 31 included variables, albumin, uric acid, C-reactive protein, donor age, platelet, lactate Dehydrogenase, and Hemoglobin were identified as the most important predictors. The two algorithms XGBClassifier and HistGradientBoostingClassifier with an average accuracy of 90.70%, sensitivity of 92.5%, and specificity of 89.13% were selected as the most appropriate ML models for predicting aGvHD. The agreement between CDSS prediction and patient outcome was 92%.

Conclusion:

ML methods can reliably predict the likelihood of aGvHD at the time of transplantation. These methods can help us to limit the number of risk factors to those that have significant effects on the outcome. However, their performance is heavily dependent on selecting the appropriate methods and algorithms. The next generations of CDSS may use more and more machine learning approaches.

Predicting postoperative opioid use with machine learning and insurance claims in opioid-naïve patients

BACKGROUND

The clinical impact of postoperative opioid use requires accurate prediction strategies to identify at-risk patients. We utilize preoperative claims data to predict postoperative opioid refill and new persistent use in opioid-naïve patients.

METHODS

A retrospective study was conducted on 112,898 opioid-naïve adult postoperative patients from Optum's de-identified Clinformatics® Data Mart database. Potential predictors included sociodemographic data, comorbidities, and prescriptions within one year prior to surgery.

RESULTS

Compared to linear models, non-linear models led to modest improvements in predicting refills - area under the receiver operating characteristics curve (AUROC) 0.68 vs. 0.67 (p < 0.05) - and performed identically in predicting new persistent use - AUROC = 0.66. Undergoing major surgery, opioid prescriptions within 30 days prior to surgery, and abdominal pain were useful in predicting refills; back/joint/head pain were the most important features in predicting new persistent use.

CONCLUSIONS

Preoperative patient attributes from insurance claims could potentially be useful in guiding prescription practices for opioid-naïve patients.

A Systematic Review of Machine Learning Techniques in Hematopoietic Stem Cell Transplantation (HSCT)

Machine learning techniques are widely used nowadays in the healthcare domain for the diagnosis, prognosis, and treatment of diseases. These techniques have applications in the field of hematopoietic cell transplantation (HCT), which is a potentially curative therapy for hematological malignancies. Herein, a systematic review of the application of machine learning (ML) techniques in the HCT setting was conducted. We examined the type of data streams included, specific ML techniques used, and type of clinical outcomes measured. A systematic review of English articles using PubMed, Scopus, Web of Science, and IEEE Xplore databases was performed. Search terms included "hematopoietic cell transplantation (HCT)," "autologous HCT," "allogeneic HCT," "machine learning," and "artificial intelligence." Only full-text studies reported between January 2015 and July 2020 were included. Data were extracted by two authors using predefined data fields. Following PRISMA guidelines, a total of 242 studies were identified, of which 27 studies met the inclusion criteria. These studies were sub-categorized into three broad topics and the type of ML techniques used included ensemble learning (63%), regression (44%), Bayesian learning (30%), and support vector machine (30%). The majority of studies examined models to predict HCT outcomes (e.g., survival, relapse, graft-versus-host disease). Clinical and genetic data were the most commonly used predictors in the modeling process. Overall, this review provided a systematic review of ML techniques applied in the context of HCT. The evidence is not sufficiently robust to determine the optimal ML technique to use in the HCT setting and/or what minimal data variables are required.

Machine learning and artificial intelligence in haematology

Digitalization of the medical record and integration of genomic methods into clinical practice have resulted in an unprecedented wealth of data. Machine learning is a subdomain of artificial intelligence that attempts to computationally extract meaningful insights from complex data structures. Applications of machine learning in haematological scenarios are steadily increasing. However, basic concepts are often unfamiliar to clinicians and investigators. The purpose of this review is to provide readers with tools to interpret and critically appraise machine learning literature. We begin with the elucidation of standard terminology and then review examples in haematology. Guidelines for designing and evaluating machine-learning studies are provided. Finally, we discuss limitations of the machine-learning approach.