SEARCH · PubMed Health

Results for “Predictive Learning Models”

Explore indexed PubMed citations for clinical trials, systematic reviews and public health research. Read source abstracts and follow each citation to its original PubMed record.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 721 records · Page 40Linked to original sources

PubMed2004

Predictors of substance use among homeless youth in San Diego.

This study examined the frequency of substance use among 14- to 24-year-old homeless youth (N=113) recruited from two community drop-in centers and explored the relationship between substance use and hypothesized psychosocial predictors. Audio-computer-assisted self-interviewing (A-CASI) was used for assessment. Including alcohol and tobacco, the mean number of different drugs used was 3.55 for lifetime and 2.34 for the last 3 months. A three-block hierarchical multiple regression was conducted to determine potential predictors of overall drug use (the sum of all different drugs used) during the last 3 months. Block 1 included demographic variables, Block 2 included a parental monitoring variable, and Block 3 included peer and environmental variables derived from learning theories. Parental monitoring (-) and peer variables (+) predicted overall 3-month drug use. The final model explained 36% of the variance in overall drug use. Results suggest that homeless adolescent drug use exists at high levels and is related to parental monitoring and peer modeling of other risk behaviors. These results may inform future prevention strategies for homeless youth and other high-risk populations.

Adolescent↗

PubMed2002

Hippocampal and neocortical contributions to memory: advances in the complementary learning systems framework.

The complementary learning systems framework provides a simple set of principles, derived from converging biological, psychological and computational constraints, for understanding the differential contributions of the neocortex and hippocampus to learning and memory. The central principles are that the neocortex has a low learning rate and uses overlapping distributed representations to extract the general statistical structure of the environment, whereas the hippocampus learns rapidly using separated representations to encode the details of specific events while minimizing interference. In recent years, we have instantiated these principles in working computational models, and have used these models to address human and animal learning and memory findings, across a wide range of domains and paradigms. Here, we review a few representative applications of our models, focusing on two domains: recognition memory and animal learning in the fear-conditioning paradigm. In both domains, the models have generated novel predictions that have been tested and confirmed.

Journal Article↗

PubMedDate not supplied

Distinguishing between natural products and synthetic molecules by descriptor Shannon entropy analysis and binary QSAR calculations.

Molecular descriptors were identified by Shannon entropy analysis that correctly distinguished, in binary QSAR calculations, between naturally occurring molecules and synthetic compounds. The Shannon entropy concept was first used in digital communication theory and has only very recently been applied to descriptor analysis. Binary QSAR methodology was originally developed to correlate structural features and properties of compounds with a binary formulation of biological activity (i.e., active or inactive) and has here been adapted to correlate molecular features with chemical source (i.e., natural or synthetic). We have identified a number of molecular descriptors with significantly different Shannon entropy and/or "entropic separation" in natural and synthetic compound databases. Different combinations of such descriptors and variably distributed structural keys were applied to learning sets consisting of natural and synthetic molecules and used to derive predictive binary QSAR models. These models were then applied to predict the source of compounds in different test sets consisting of randomly collected natural and synthetic molecules, or, alternatively, sets of natural and synthetic molecules with specific biological activities. On average, greater than 80% prediction accuracy was achieved with our best models. For the test case consisting of molecules with specific activities, greater than 90% accuracy was achieved. From our analysis, some chemical features were identified that systematically differ in many naturally occurring versus synthetic molecules.

Algorithms↗

PubMed1985

Observing response acquisition: preference for unpredictable appetitive rewards obtained under conditions predicted by DMOD.

In five E-maze experiments, rats were given a choice between receiving reward and nonreward in a situation where stimuli were correlated with reward outcome (predictable situation) versus one where the stimuli were uncorrelated with reward outcome (unpredictable situation). Preference for the unpredictable situation occurred under the following conditions: (a) small (one 37-mg pellet), immediate rewards; (b) small, delayed (15 s) rewards, if the cues correlated with reward outcome were absent during the delay interval; (c) large (15 pellets), immediate rewards if a difficult discrimination was required; and (d) if the stimulus predicting nonreward was present at the choice point. Preference for the predictable situation was strongest if reinforcement was delayed and large or the stimulus predicting reward was present at the choice point. A weaker preference for the predictable situation occurred if reinforcement was immediate and large and a simple discrimination was required or if reinforcement was large and delayed and the cues that correlated with reward outcome were absent during the delay interval. The results support the predictions of DMOD (Daly modification of the Rescorla-Wagner model), a mathematical model of appetitive learning (Daly & Daly, 1982).

Animals↗

PubMed2005

Why pictures look right when viewed from the wrong place.

A picture viewed from its center of projection generates the same retinal image as the original scene, so the viewer perceives the scene correctly. When a picture is viewed from other locations, the retinal image specifies a different scene, but we normally do not notice the changes. We investigated the mechanism underlying this perceptual invariance by studying the perceived shapes of pictured objects viewed from various locations. We also manipulated information about the orientation of the picture surface. When binocular information for surface orientation was available, perceived shape was nearly invariant across a wide range of viewing angles. By varying the projection angle and the position of a stimulus in the picture, we found that invariance is achieved through an estimate of local surface orientation, not from geometric information in the picture. We present a model that explains invariance and other phenomena (such as perceived distortions in wide-angle pictures).

Attention↗

PubMed2006

Causal reasoning in rats.

Empirical research with nonhuman primates appears to support the view that causal reasoning is a key cognitive faculty that divides humans from animals. The claim is that animals approximate causal learning using associative processes. The present results cast doubt on that conclusion. Rats made causal inferences in a basic task that taps into core features of causal reasoning without requiring complex physical knowledge. They derived predictions of the outcomes of interventions after passive observational learning of different kinds of causal models. These competencies cannot be explained by current associative theories but are consistent with causal Bayes net theories.

Animals↗

PubMed2026

Comprehensive in silico genomics analysis of global trends and host-specific emergence of aminoglycoside resistance in Staphylococcus aureus: a One-Health perspective.

BACKGROUND: Aminoglycosides remain clinically valuable against Staphylococcus aureus. Aminoglycoside resistance in S. aureus represents a critical One Health concern and is primarily driven by aminoglycoside-modifying enzymes (AMEs), which are frequently plasmid-encoded. Although regional studies have provided valuable insights, the global epidemiology of aminoglycoside resistance determinants remains poorly characterized because comprehensive data integrating human, animal, and environmental reservoirs are still lacking. This study addresses this gap by analyzing over 110,000 S. aureus genomes (2000-2025) to map the global resistome, quantify temporal and host-specific trends, and assess the association between genetic determinants and phenotypic resistance. METHODS: We performed a retrospective One Health meta-analysis of 110,309 S. aureus genomes collected between 2000 and 2025 from 128 countries. Genomes were quality-filtered and aminoglycoside resistance determinants were identified using NCBI AMRFinderPlus (v4.0.23). Multilocus sequence typing and host-source harmonization (Human, Animal, Environment, Unknown) enabled clonal and reservoir stratification. Temporal trends in gene prevalence and resistance burden were modeled with robust regression. Geographic and host-associated structuring of key genes was assessed via χ2 and enrichment tests. Machine-learning models (elastic-net, random forests, XGBoost) were benchmarked for minimum inhibitory concentration (MIC) prediction via nested cross-validation, with performance evaluated by mean absolute error, RMSE, and SHAP-based feature importance. All analyses were conducted in R and Python using publicly available, de-identified genomic data. RESULTS: Aminoglycoside resistance-associated genes were dominated by modifying enzyme determinants, with ant(6)-Ia, ant(9)-Ia, aph(3')-IIIa, sat4, aadD1, and aac(6')-Ie/aph(2'')-Ia occurring in 14-22% of isolates worldwide. Temporal analysis revealed significant declines in several major determinants, most notably ant(9)-Ia (-2.22 percentage points per year, p < 0.001), whereas apmA exhibited a non-significant decreasing trend in animal isolates. Host structuring was marked: human clinical isolates concentrated common determinants, while animal and environmental isolates harbored rare alleles (apmA, spw, str, spd). Geographic mapping confirmed near-universal distribution of common genes but focal restriction of rare ones. Publicly available phenotypic data indicated strong activity of amikacin, whereas gentamicin showed a distinct resistant subpopulation that closely corresponded with AME gene carriage. Genotype-phenotype analyses demonstrated strong concordance, with gene-rich complements predicting resistant MIC strata and absence of determinants predicting susceptibility. Analysis across different gene classes revealed frequent co-occurrence of aminoglycoside resistance genes with determinants from other classes, such as mecA, blaZ, and MLS_B, embedding them within multidrug-resistant (MDR) genomic contexts. CONCLUSION: Over 25 years, the prevalence of aminoglycoside resistance-associated genes in S. aureus has declined for several common determinants, while rare veterinary-linked alleles are emerging in animal isolates. Strong genotype-phenotype concordance supports genomic prediction for gentamicin and amikacin, where MIC data are available, although phenotypic confirmation remains essential. The frequent co-occurrence of aminoglycoside resistance genes with other antimicrobial resistance determinants indicates their integration within co-occurrence patterns of MDR genes, defined here as clusters of co-occurring resistance genes often carried on shared mobile genetic elements. These patterns highlight the need for integrated One Health surveillance combining clinical, veterinary, and environmental monitoring with plasmid-context resolution to anticipate emerging threats.

Aminoglycosides↗

PubMed2005

Confirmatory factor analysis of the Teacher Efficacy Scale for prospective teachers.

BACKGROUND: Research on teacher self-efficacy has revealed substantive problems concerning the validity of instruments used to measure teacher self-efficacy beliefs. Although claims about the influence of teachers' self-efficacy beliefs on student achievement, success with curriculum innovation, and so on, may be true statements, one cannot make those claims on the basis of that body of evidence if the instruments are not valid measures of teachers' self-efficacy beliefs. AIMS: The purpose of this investigation is to employ the use of modern confirmatory factor-analytic techniques to investigate the validity of the hypothesized dimensions of the Teacher Efficacy Scale (Gibson & Dembo, 1984; Woolfolk & Hoy, 1990). SAMPLE: Participants for this investigation were 387 prospective teachers recruited from a university located in the south-western region of the UA. Participants for Study 2 were 131 prospective elementary teachers recruited from the same university as in Study 1. RESULTS: A confirmatory factor analysis (CFA) procedure was used to evaluate the goodness-of-fit for two theoretical models of the TES items. The proposed two- and three-factor models of teacher self-efficacy for prospective teachers were rejected. A re-specified three-factor model of the TES was then derived from theoretical and empirical considerations. The re-specified model hypothesized three dimensions: self-efficacy beliefs, outcome expectations, and external locus-of-causality. In Study 2, the re-specified three-factor measurement model was evaluated in a new sample. Results of the CFA procedure indicated satisfactory fit of the re-specified model to the data; however, the results were not consistent with predictions derived from social learning theory. CONCLUSIONS: The results of this study call into question the use of the TES and the interpretation of a large body of literature purporting to study the relationship of teachers' self-efficacy beliefs to important educational outcomes.

Adult↗

PubMed1996

Pharmaceutical granulation and tablet formulation using neural networks.

Current-day pharmaceutical formulation may be trial and error in nature due to the absence of a clear relationship between the formulation characteristics (output variables) and the material and process variables (input variables). Neural networks are networks of adaptable nodes, which through a process of learning from task examples, store experiential knowledge and make it available for prediction. Prediction of a model granulation and tablet system characteristics from the knowledge of material and process variables utilizing neural networks is the basis of this presentation. The formulation design contained the following variables: granulation equipment, diluent, method of binder addition, and the binder concentration. The material, process, granulation evaluation, and tablet evaluation data of the formulations were used as the data set for training and testing of the neural network models. A comparison of the neural network prediction performance with that of regression models was also done. Both the granulation model and the tablet model converged fairly rapidly in the training step. In the testing step, the predictions for all granulation model variables (geometric mean particle size, flow value, bulk density, and tap density) were satisfactory. In the tablet model, the predictions for disintegration and thickness were also satisfactory. The predictions for hardness and friability were less than satisfactory. Two situations where the neural network may not perform adequately are discussed. The neural network prediction is better or comparable for all the predicted variables in this study compared to regression methods. The results clearly show the applicability of neural networks to formulation modeling.

Caffeine↗

PubMed1995

QSAR of 3-methylfentanyl derivatives studied with neural networks method.

AIM: To use neural networks, which simulate the functions of living nervous systems, in QSAR studies; METHODS: Using the back-propagation neural networks program devised by us, combining with partial least squares (PLS) method, we studied the relationships of quantum chemical indices and analgesic activities of 25 3-methylfentanyl derivatives; RESULTS: Through learning process, a good QSAR model was established, and the activities of these compounds were predicted; the correlation between the activities and quantum chemical indices: the net charge of the atom N1, the net charge of the atom O16, the torsional angle of atoms C10-C9-N8-C4, the interatomic distance between atom C7 and the center of phenyl plane C9-14 (PhA), is quite well-matched. Based on these results, an interactive pattern between 3-methylfentanyl derivatives and opioid receptors was suggested; CONCLUSION: Not only are the results of neural networks superior to those of PLS method but they also provide accurate predictions of the activity of the compounds and also combine the PLS method with neural networks.

Analgesics↗

PubMed2003

Retrospective and prospective persistent activity induced by Hebbian learning in a recurrent cortical network.

Recordings from cells in the associative cortex of monkeys performing visual working memory tasks link persistent neuronal activity, long-term memory and associative memory. In particular, delayed pair-associate tasks have revealed neuronal correlates of long-term memory of associations between stimuli. Here, a recurrent cortical network model with Hebbian plastic synapses is subjected to the pair-associate protocol. In a first stage, learning leads to the appearance of delay activity, representing individual images ('retrospective' activity). As learning proceeds, the same learning mechanism uses retrospective delay activity together with choice stimulus activity to potentiate synapses connecting neural populations representing associated images. As a result, the neural population corresponding to the pair-associate of the image presented is activated prior to its visual stimulation ('prospective' activity). The probability of appearance of prospective activity is governed by the strength of the inter-population connections, which in turn depends on the frequency of pairings during training. The time course of the transitions from retrospective to prospective activity during the delay period is found to depend on the fraction of slow, N-methyl-d-aspartate-like receptors at excitatory synapses. For fast recurrent excitation, transitions are abrupt; slow recurrent excitation renders transitions gradual. Both scenarios lead to a gradual rise of delay activity when averaged over many trials, because of the stochastic nature of the transitions. The model reproduces most of the neuro-physiological data obtained during such tasks, makes experimentally testable predictions and demonstrates how persistent activity (working memory) brings about the learning of long-term associations.

Animals↗

PubMed2005

Different neural correlates of reward expectation and reward expectation error in the putamen and caudate nucleus during stimulus-action-reward association learning.

To select appropriate behaviors leading to rewards, the brain needs to learn associations among sensory stimuli, selected behaviors, and rewards. Recent imaging and neural-recording studies have revealed that the dorsal striatum plays an important role in learning such stimulus-action-reward associations. However, the putamen and caudate nucleus are embedded in distinct cortico-striatal loop circuits, predominantly connected to motor-related cerebral cortical areas and frontal association areas, respectively. This difference in their cortical connections suggests that the putamen and caudate nucleus are engaged in different functional aspects of stimulus-action-reward association learning. To determine whether this is the case, we conducted an event-related and computational model-based functional MRI (fMRI) study with a stochastic decision-making task in which a stimulus-action-reward association must be learned. A simple reinforcement learning model not only reproduced the subject's action selections reasonably well but also allowed us to quantitatively estimate each subject's temporal profiles of stimulus-action-reward association and reward-prediction error during learning trials. These two internal representations were used in the fMRI correlation analysis. The results revealed that neural correlates of the stimulus-action-reward association reside in the putamen, whereas a correlation with reward-prediction error was found largely in the caudate nucleus and ventral striatum. These nonuniform spatiotemporal distributions of neural correlates within the dorsal striatum were maintained consistently at various levels of task difficulty, suggesting a functional difference in the dorsal striatum between the putamen and caudate nucleus during stimulus-action-reward association learning.

Adult↗

PubMed1999

Predicting physical activity intentions using a goal perspectives approach: a study of Hungarian youth.

Utilising a goal perspectives framework, a study predicting physical activity intentions in 12 to 16-year-old Hungarian adolescents was conducted with two samples. Theoretical predictions established a model that was tested through path analysis. Beliefs thought to underpin goal orientations were hypothesised to predict ego orientation (general and gift beliefs) and task orientation (learning and incremental beliefs). Task orientation was hypothesised to predict intentions directly, while ego orientation was hypothesised to predict intentions indirectly through perceived competence. Results from the first sample (n=301) suggested that the model could be improved by adding paths between general beliefs and perceived competence and between task orientation and perceived competence. This modified model was shown to fit data from a second sample (n=422) very well. Multi-group analysis confirmed a good fit and so the two samples were combined. The model fitted the data well for the total sample (n=723). Overall, results showed that 20.8% of the variance in intentions was explained by the model, and that sport ability beliefs were moderately associated with task orientation but only weakly associated with ego orientation. The motivational importance of a task orientation was confirmed with its direct prediction of intentions.

Achievement↗

PubMed2004

Do current connectionist learning models account for reading development in different languages?

Learning to read a relatively irregular orthography, such as English, is harder and takes longer than learning to read a relatively regular orthography, such as German. At the end of grade 1, the difference in reading performance on a simple set of words and nonwords is quite dramatic. Whereas children using regular orthographies are already close to ceiling, English children read only about 40% of the words and nonwords correctly. It takes almost 4 years for English children to come close to the reading level of their German peers. In the present study, we investigated to what extent recent connectionist learning models are capable of simulating this cross-language learning rate effect as measured by nonword decoding accuracy. We implemented German and English versions of two major connectionist reading models, Plaut et al.'s (Plaut, D. C., McClelland, J. L., Seidenberg, M. S., & Patterson, K. (1996). Understanding normal and impaired word reading: computational principles in quasi-regular domains. Psychological Review, 103, 56-115) parallel distributed model and Zorzi et al.'s (Zorzi, M., Houghton, G., & Butterworth, B. (1998a). Two routes or one in reading aloud? A connectionist dual-process model. Journal of Experimental Psychology: Human Perception and Performance, 24, 1131-1161); two-layer associative network. While both models predicted an overall advantage for the more regular orthography (i.e. German over English), they failed to predict that the difference between children learning to read regular versus irregular orthographies is larger earlier on. Further investigations showed that the two-layer network could be brought to simulate the cross-language learning rate effect when cross-language differences in teaching methods (phonics versus whole-word approach) were taken into account. The present work thus shows that in order to adequately capture the pattern of reading acquisition displayed by children, current connectionist models must not only be sensitive to the statistical structure of spelling-to-sound relations but also to the way reading is taught in different countries.

Child↗

PubMed1998

Identification of structural features and associated mechanisms of action for carcinogens in rats.

A set of chemicals tested for carcinogenicity in rats that have been analyzed in the Carcinogenic Potency Database (CPDB) was subjected to CASE/MULTICASE (a computer-automated structure evaluation system) structure-activity relationship (SAR) analyses. This SAR system identifies structural features of chemicals in a learning set that are associated with a predefined activity and produces an SAR model based on these characteristics. The rat CPDB used in this study consisted of 745 chemicals, 383 of which are carcinogens, 14 marginally active carcinogens (i.e., chemicals that require a relatively high dose to induce carcinogenesis) and 348 are non-carcinogens. In an internal prediction analysis where CASE/MULTICASE 'predicted' the activity of chemicals in the learning set, the system was able to achieve a concordance between experimental and predicted results of 95%. This indicates that the program is able to adequately assess the chemicals in the database. In a 10-fold cross-validation study where 10 disjoint sets of 10% of the chemicals were removed from the database and the remaining 90% of the chemicals were used as a learning set, CASE/MULTICASE was able to achieve a concordance between experimental and predicted results of 64%. Using a modified validation process designed to investigate the predictivity of a more focused SAR model, the system was able to achieve a concordance of 71% between experimental and predicted results. Among the major biophores identified by CASE/MULTICASE as associated with cancer causation in rats, several are derived from electrophilic or potentially electrophilic compounds (e.g., aromatic amines, nitrogen mustards, isocyanates, epoxides). Other biophores however are derived from chemicals seemingly devoid of actual or potential DNA-reactivity and as such may represent structural features of non-genotoxic carcinogens.

Animals↗

PubMed2026

Biallelic loss of RB1 in hepatocellular carcinoma as synthetic lethal target for artificial intelligence-guided therapy.

The retinoblastoma (RB1) gene is a critical tumor suppressor that regulates cell cycle progression and genomic stability. Although RB1 alterations have been reported in hepatocellular carcinoma (HCC), the biological and clinical consequences of biallelic RB1 inactivation (RB1-Bi) remain poorly defined. We performed a comprehensive allele-specific genomic analysis of HCC patients from the TCGA-LIHC (n = 355) and in-house AMC (n = 206) cohorts, collectively comprising the AMC-TCGA discovery cohort. In this combined cohort, RB1-Bi was identified in 14.6% of tumors, was enriched in poorly differentiated HCCs and was independently associated with significantly reduced overall survival (adjusted hazard ratio 3.32, 95% CI 1.93-5.72, p < 0.001). Additionally, a deep learning-based histopathology model using hematoxylin and eosin-stained slides (i.e., FR-MIL model) accurately predicted RB1-Bi status (F1 score 84.39% [95% CI, ±0.02]), making it readily identifiable in routine clinical practice. The prevalence and prognostic impact of RB1-Bi, as well as FR-MIL model performance, were consistent across independent validation cohorts, including advanced-stage tumors and external institutions. High-throughput drug screening in isogenic HCC models revealed that RB1-Bi HCC cells were particularly sensitive to inhibitors targeting mitotic regulators (e.g., AURKA, PLK1, KSP) and DNA damage response pathways (e.g., PARP inhibitors). Synthetic lethal interactions between RB1-Bi and these compounds were demonstrated in vitro and in vivo, and combination treatment with mitotic and PARP inhibitors had synergistic effects with acceptable tolerability. We conclude that RB1-Bi represents a clinically actionable biomarker that identifies a high-risk HCC subtype with specific therapeutic vulnerabilities, offering new opportunities for precision medicine.

Humans↗

PubMed2005

Learned helplessness: validity and reliability of depressive-like states in mice.

The learned helplessness paradigm is a depression model in which animals are exposed to unpredictable and uncontrollable stress, e.g. electroshocks, and subsequently develop coping deficits for aversive but escapable situations (J.B. Overmier, M.E. Seligman, Effects of inescapable shock upon subsequent escape and avoidance responding, J. Comp. Physiol. Psychol. 63 (1967) 28-33 ). It represents a model with good similarity to the symptoms of depression, construct, and predictive validity in rats. Despite an increased need to investigate emotional, in particular depression-like behaviors in transgenic mice, so far only a few studies have been published using the learned helplessness paradigm. One reason may be the fact that-in contrast to rats (B. Vollmayr, F.A. Henn, Learned helplessness in the rat: improvements in validity and reliability, Brain Res. Brain Res. Protoc. 8 (2001) 1-7)--there is no generally accepted learned helplessness protocol available for mice. This prompted us to develop a reliable helplessness procedure in C57BL/6N mice, to exclude possible artifacts, and to establish a protocol, which yields a consistent fraction of helpless mice following the shock exposure. Furthermore, we validated this protocol pharmacologically using the tricyclic antidepressant imipramine. Here, we present a mouse model with good face and predictive validity that can be used for transgenic, behavioral, and pharmacological studies.

Animals↗

PubMed2003

Performance of a new clinical grading system for chronic graft-versus-host disease: a multicenter study.

We recently reported 3 risk factors (RFs) at diagnosis of chronic graft-versus-host disease (cGVHD) that were significantly associated with increased nonrelapse mortality. These included extensive skin involvement (ESI), thrombocytopenia (TP), and progressive type of onset (PTO). The hazard ratio (HR) for mortality of the patients with prognostic score (PS) between 0 and 2 (intermediate-risk; 1 RF) compared to those with PS 0 (favorable-risk; 0 RF) was 3.7 (95% CI, 1.4, 9.3); the HR for patients with PS equal to or more than 2 (high-risk; > 1 RF) compared with intermediate-risk group was 6.9 (3.8, 12.4). A rare presentation of TP and PTO without ESI yielded a PS of 1.8 (intermediate-risk). This paper reports the performance of the prognostic model and the individual RFs using data from an additional 1105 patients from University of Nebraska (n = 60), International Bone Marrow Transplantation Registry (n = 708), Fred Hutchinson Cancer Research Center (n = 188), and University of Minnesota (n = 149). The extent of skin involvement was quantified in 3 cohorts using the available data collected in different formats before the analysis. Although the HR for mortality of the patients in the intermediate-risk group versus those in the favorable-risk group ranged from 2.3 to 8.9 across the centers, it was between 1.6 to 6.9 for patients in the high-risk group versus those in the intermediate-risk group. Although TP itself was uniformly associated with increased risk of mortality across all test samples, ESI and PTO showed statistically significant associations with mortality in 1 and 2 cohorts, respectively. In conclusion, the model was predictive of cGVHD-specific survival, but the mortality hazard associated with ESI was lower in each of these test samples compared with the learning sample. Although the new clinical grading based on the model is promising because of its utility across multiple independent data sets, prospective validation is needed.

Adolescent↗