Functional Networks Research Articles

Leveraging descriptor learning and functional map‐based shape matching for automated anatomical Landmarking in mouse mandibles

AbstractGeometric morphometrics is used in the biological sciences to quantify morphological traits. However, the need for manual landmark placement hampers scalability, which is both time‐consuming, labor‐intensive, and open to human error. The selected landmarks embody a specific hypothesis regarding the critical geometry relevant to the biological question. Any adjustment to this hypothesis necessitates acquiring a new set of landmarks or revising them significantly, which can be impractical for large datasets. There is a pressing need for more efficient and flexible methods for landmark placement that can adapt to different hypotheses without requiring extensive human effort. This study investigates the precision and accuracy of landmarks derived from functional correspondences obtained through the functional map framework of geometry processing. We utilize a deep functional map network to learn shape descriptors, which enable us to achieve functional map‐based and point‐to‐point correspondences between specimens in our dataset. Our methodology involves automating the landmarking process by interrogating these maps to identify corresponding landmarks, using manually placed landmarks from the entire dataset as a reference. We apply our method to a dataset of rodent mandibles and compare its performance to MALPACA's, a standard tool for automatic landmark placement. Our model demonstrates a speed improvement compared to MALPACA while maintaining a competitive level of accuracy. Although MALPACA typically shows the lowest RMSE, our models perform comparably well, particularly with smaller training datasets, indicating strong generalizability. Visual assessments confirm the precision of our automated landmark placements, with deviations consistently falling within an acceptable range for MALPACA estimates. Our results underscore the potential of unsupervised learning models in anatomical landmark placement, presenting a practical and efficient alternative to traditional methods. Our approach saves significant time and effort and provides the flexibility to adapt to different hypotheses about critical geometrical features without the need for manual re‐acquisition of landmarks. This advancement can significantly enhance the scalability and applicability of geometric morphometrics, making it more feasible for large datasets and diverse biological studies.

Neurobiological fingerprints of negative symptoms in schizophrenia identified by connectome‐based modeling

AimAs a central component of schizophrenia psychopathology, negative symptoms result in detrimental effects on long‐term functional prognosis. However, the neurobiological mechanism underlying negative symptoms remains poorly understood, which limits the development of novel treatment interventions. This study aimed to identify the specific neural fingerprints of negative symptoms in schizophrenia.MethodsBased on resting‐state functional connectivity data obtained in a large sample (n = 132) of first‐episode drug‐naïve schizophrenia patients (DN‐FES), connectome‐based predictive modeling (CPM) with cross‐validation was applied to identify functional networks that predict the severity of negative symptoms. The generalizability of identified networks was then validated in an independent sample of n = 40 DN‐FES.ResultsA connectivity pattern significantly driving the prediction of negative symptoms (ρ = 0.28, MSE = 81.04, P = 0.012) was identified within and between networks implicated in motivation (medial frontal, subcortical, sensorimotor), cognition (default mode, frontoparietal, medial frontal) and error processing (medial frontal and cerebellum). The identified networks also predicted negative symptoms in the independent validation sample (ρ = 0.37, P = 0.018). Importantly, the predictive model was symptom‐specific and robust considering the potential effects of demographic characteristics and validation strategies.ConclusionsOur study discovers and validates a comprehensive network model as the unique neural substrates of negative symptoms in schizophrenia, which provides a novel and comprehensive perspective to the development of target treatment strategies for negative symptoms.

Lateralization of Neural Speech Discrimination at Birth Is a Predictor for Later Language Development.

Newborns are able to neurally discriminate between speech and nonspeech right after birth. To date it remains unknown whether this early speech discrimination and the underlying neural language network is associated with later language development. Preterm-born children are an interesting cohort to investigate this relationship, as previous studies have shown that preterm-born neonates exhibit alterations of speech processing and have a greater risk of later language deficits. This investigation also holds clinical importance, as differences in neonatal speech discrimination and its functional networks may serve as predictors of later language outcomes. We therefore investigated neural speech discrimination using functional near-infrared spectroscopy in 92 preterm- and term-born neonates and its predictive value for language development in 45 of them. Three to five years later, preterm-born and term-born children did not significantly differ in language comprehension, sentence production, the use of morphological rules, or phonological short-term memory. In addition, the gestational age at birth was not a significant predictor of language development. Neural speech discrimination, in contrast, was strongly correlated with later phonological short-term memory. However, not the extent of speech discrimination, but rather its lateralization, was a predictor of language development. Children with less right hemisphere involvement-and therefore more left-lateralized speech discrimination at birth-showed better development of phonological short-term memory three to five years later. These findings suggest that the ability of fetuses to form memory traces is reflected by neonatal abilities to neurally discriminate speech, which in turn is a predictor for later phonological short-term memory.

IPRISM: Intelligent Predicting Response to Cancer Immunotherapy through Systematic Modeling

Immunotherapy has revolutionized cancer treatment, but predicting patient response remains challenging. Herein, we present iPRISM (Intelligent Predicting Response to cancer Immunotherapy through Systematic Modeling), which is a novel network‐based model that integrates multiomics data to predict immunotherapy outcomes. In this approach, iPRISM incorporates gene expression, biological functional network, tumor microenvironment characteristics, immune‐related pathways, and clinical data to provide a comprehensive view of factors influencing immunotherapy efficacy. Using stepwise logistic regression, we identified key predictive features and validated iPRISM across multiple cohorts including melanoma, bladder cancer, non‐small cell lung cancer, and stomach adenocarcinoma. We also find that iPRISM outperforms the existing methods, achieving high predictive accuracy and demonstrating significant prognostic value for overall and progression‐free survival. By identifying key genetic and immunological factors, this model provides a new insight for more personalized treatment strategies and combination therapies to overcome resistance mechanisms. iPRISM can be accessed at CRAN: https://CRAN.R‐project.org/package=iPRISM.

The gift of mobility: Transfer networks of local decision‐makers and interlocal collaboration

AbstractExploring the role of key individuals in institutional collective action (ICA) and interlocal collaboration contributes to understanding the micro‐mechanisms of the policy process. The career mobility of local decision‐makers through transfers among jurisdictions is a common phenomenon. This phenomenon gives rise to transfer networks of local decision‐makers. Despite the prevalence of these transfer networks, they have received relatively little attention in studies on interlocal collaboration. It is unclear how local decision‐makers' transfer networks and different transfer network characteristics affect interlocal collaboration. This article explains the role that transfer networks and transfer network characteristics play in interlocal collaboration by embedding decision‐makers' social networks with transfer networks based on the ICA framework. This study used a dyadic panel dataset of Chinese inter‐provincial watershed environmental collaboration from 2007 to 2019 for empirical research. The results show that existence of transfer networks of local decision‐makers increases the likelihood of interlocal collaboration. The social network ties constructed by decision‐makers in their career paths are responsible for the functioning of transfer networks. The likelihood of interlocal collaboration is higher when transfer networks exhibit successiveness, diagonal promotion characteristics, and collegiality.

Gendered dynamics in Moroccan transnational migrant networks in Tenerife: environmental concerns, engagement, and remittances

The concept of transnational knowledge networks has entered the field of study of environmental mobility. Yet, how these networks play a role in adaptation to environmental change remains understudied. Power dynamics and familial ties impact the functioning of transnational networks and intricately shape responses to environmental challenges, with gender serving as a salient factor. This research aims to theorize and empirically study how gender matters for how men and women engage with environmental concerns within transnational spaces of Moroccan migrants living in Tenerife, the largest of the Canary Islands. This study employs visual mapping techniques to uncover gendered patterns and modalities of communication within transnational networks. Findings reveal significant gender disparities in how environmental challenges are discussed and addressed within Moroccan transnational networks. While women prioritize family welfare and employ cultural practices to tackle environmental issues, men focus more on economic implications. Women’s perspectives and voices are marginalized, reflecting societal gender norms and biases that prioritize men’s voices and expertise in economic matters. This ignorance hinders their participation and access to support within transnational networks, perpetuating gender disparities in engagement and decision-making.

Establishing Artificial Grassland on Extremely Degraded Alpine Meadow Changes the Soil Fungal Community and Function in the Qilian Mountain Area

ABSTRACTThe establishment of artificial grasslands is a recognized method for restoring severely degraded alpine meadows, yet its impact on soil fungi remains poorly understood. To investigate the impact of artificial grassland establishment on soil fungal communities, we selected a typical 4‐year‐old artificial grassland as the research subject, with non‐degraded and extremely degraded alpine meadows serving as controls. Soil fungal communities and functional groups were analyzed using Illumina sequencing of internal transcribed spacer (ITS) genes, and functional groups were predicted using the FUNGuild database. The results indicated that artificial grassland establishment had a significant impact on vegetation properties, while not affecting soil nutrient contents. Additionally, established artificial grassland significantly increased fungal diversity but did not alter the abundance of dominant phyla except Ascomycota. The ecological network analysis revealed that artificial grassland establishment (0.0910) markedly decreased fungal robustness compared to non‐degraded (0.2040) and extremely degraded (0.1790), suggesting unfavorable soil conditions for fungal interactions. Moreover, vegetation properties (Shannon and Richness diversity indices) were the primary factors influencing soil fungal functional groups, communities, and ecological networks. Our findings indicate that the artificial grassland establishment rapidly restores vegetation but does not support the restoration of fungal communities.

Clinical genetic services in the Emilia-Romagna region, Italy: current activity and open issues: a mixed-method study.

In 2002, in the Emilia-Romagna region of Italy, a comprehensive strategic plan was developed with the aim of improving the integration and efficiency of the genetic services. Two decades later, this report aims to explore the current functioning of the regional network, with special focus on clinical genetics in the evolving scenarios. To this aim, we analyzed the activity data of the medical genetics services in the region, to identify and possibly improve currently open issues. This is a mixed-method study, analyzing quantitatively and qualitatively the activities of seven medical genetics services in Emilia-Romagna region. Quantitative analysis considered the number of consultations and the composition of the staff in the year 2021. Qualitative analysis examined a focus group of directors of the services through reflexive thematic analysis. A total of 14,925 counseling sessions have been delivered by the medical genetics services, staffed with 22.4 full-time equivalent clinical geneticists. A physician performed an average of 14.5 consultations per week and approximately 1166h of patient care per year. The clinical geneticists/inhabitants ratio was 0.54 per 100,000 inhabitants, and it is estimated that one every 278 inhabitants, on average, underwent a genetic counseling session in 2021. Qualitative analysis highlighted issues concerning patients' access to service, general organization and staff composition. In order to meet the growing demand for genetic counseling services, expansion of the workforce and adjustment of current practice models are required to increase the access to genetic services and the application of test results to clinical management.

Zero-echo time imaging achieves whole brain activity mapping without ventral signal loss in mice.

Functional MRI (fMRI) is an important tool for investigating functional networks. However, the widely used fMRI with T2*-weighted imaging in rodents has the problem of signal lack in the lateral ventral area of forebrain including the amygdala, which is essential for not only emotion but also noxious pain. Here, we scouted the zero-echo time (ZTE) sequence, which is robust to magnetic susceptibility and motion-derived artifacts, to image activation in the whole brain including the amygdala following the noxious stimulation to the hind paw. ZTE exhibited higher spatial and temporal signal-to-noise ratios than conventional fMRI sequences. Electrical sensory stimulation of the hind paw evoked ZTE signal increase in the primary somatosensory cortex. Formalin injection into the hind paw evoked early and latent change of ZTE signals throughout the whole brain including the subregions of amygdala. Furthermore, resting-state fMRI using ZTE demonstrated the functional connectivity, including that of the amygdala. These results indicate the feasibility of ZTE for whole brain fMRI including the amygdala and we first show acute and latent activity in different subnuclei of the amygdala complex after nociceptive stimulation.

A Vehicle Adaptive Cruise Control Method Based on Deep Reinforcement Learning Algorithm

Adaptive cruise control (ACC) is an upgrade to the traditional cruise control system in vehicles. This paper presents an adaptive cruise control method based on the Deep Deterministic Policy Gradient (DDPG) algorithm. The Actor network computes actions based on the current state, adding noise to enhance exploration. The Critic network computes the Q-value of the current state-action pair. The Actor target network and Critic target network compute the Q-value of the next state-action pair. The gradient descent method is used to minimize the loss function of the Critic network, which includes the error between the real Q-value and the target Q-value. By integrating an Actor-Critic network, the system demonstrates improved adaptability and efficiency over traditional ACC methods, as shown through simulated experiments

Multiple constraint network classification reveals functional brain networks distinguishing 0-back and 2-back task.

Working memory is associated with general intelligence and is crucial for performing complex cognitive tasks. Neuroimaging investigations have recognized that working memory is supported by a distribution of activity in regions across the entire brain. Identification of these regions has come primarily from general linear model analyses of statistical parametric maps to reveal brain regions whose activation is linearly related to working memory task conditions. This approach can fail to detect nonlinear task differences or differences reflected in distributed patterns of activity. In this study, we take advantage of the increased sensitivity of multivariate pattern analysis in a multiple-constraint deep learning classifier to analyze patterns of whole-brain blood oxygen level dependent (BOLD) activity in children performing two different conditions of the emotional n-back task. Regional (supervoxel) whole-brain activation patterns from functional imaging runs of 20 children were used to train a set of neural network classifiers to identify task category (0-back vs. 2-back) and activation co-occurrence probability, which encoded functional connectivity. These simultaneous constraints promote the discovery of coherent networks that contribute towards task performance in each memory load condition. Permutation analyses discovered the global activation patterns and interregional coactivations that distinguish memory load. Examination of model weights identified the brain regions most predictive of memory load and the functional networks integrating these regions. Community detection analyses identified functional networks integrating task-predictive regions and found distinct patterns of network activation for each task type. Comparisons to functional network literature suggest more focused attentional network activation during the 2-back task. (PsycInfo Database Record (c) 2025 APA, all rights reserved).

The Italian Manosphere: Composition, Structure, and Functions of a Digital Network

The Italian Manosphere: Composition, Structure, and Functions of a Digital Network

Altered Static and Dynamic Functional Network Connectivity and Combined Machine Learning in Stroke.

Stroke is a condition characterized by damage to the cerebral vasculature from various causes, resulting in focal or widespread brain tissue damage. Prior neuroimaging research has demonstrated that individuals with stroke present structural and functional brain abnormalities, evident through disruptions in motor, cognitive, and other vital functions. Nevertheless, there is a lack of studies on alterations in static and dynamic functional network connectivity in the brains of stroke patients. Fifty stroke patients and 50 healthy controls (HCs) underwent resting-state functional magnetic resonance imaging (rs-fMRI) scanning. Initially, the independent component analysis (ICA) method was utilized to extract the resting-state network (RSN). Subsequently, the disparities in static functional network connectivity both within and between networks among the two groups were computed and juxtaposed. Following this, five consistent and robust dynamic functional network connectivity (dFNC) states were derived by integrating the sliding time window method with k-means cluster analysis, and the distinctions in dFNC between the groups across different states, along with the intergroup variations in three dynamic temporal metrics, were assessed. Finally, a support vector machine (SVM) approach was employed to discriminate stroke patients from HCs using FC and FNC as classification features. Comparing the stroke group to the healthy control (HC) group, the stroke group exhibited reduced intra-network functional connectivity (FC) in the right superior temporal gyrus of the ventral attention network (VAN), the left calcarine of the visual network (VN), and the left precuneus of the default mode network (DMN). Regarding static functional network connectivity (FNC), we identified increased connectivity between the executive control network (ECN) and dorsal attention network (DAN), salience network (SN) and DMN, SN-ECN, and VN-ECN, along with decreased connectivity between DAN-DAN, ECN-SN, SN-SN, and DAN-VN between the two groups. Noteworthy differences in dynamic FNC (dFNC) were observed between the groups in states 3 to 5. Moreover, stroke patients demonstrated a significantly higher proportion of time and longer mean dwell time in state 4, alongside a decreased proportion of time in state 5 compared to HC. Finally, utilizing FC and FNC as features, stroke patients could be distinguished from HC with an accuracy exceeding 70% and an area under the curve ranging from 0.8284 to 0.9364. In conclusion, our study reveals static and dynamic changes in large-scale brain networks in stroke patients, potentially linked to abnormalities in visual, cognitive, and motor functions. This investigation offers valuable insights into the neural mechanisms underpinning the functional deficits observed in stroke, thereby aiding in the diagnosis and development of targeted therapeutic interventions for affected individuals.

Abnormal dynamic functional networks during pain-free periods: resting-state co-activation pattern analysis in primary dysmenorrhea: Abnormal dynamic functional networks in primary dysmenorrhea.

Chronic pain alters the configuration of brain functional networks. Primary dysmenorrhea (PDM) is a form of chronic visceral pain, which has been identified spatial alterations in brain functional networks using static functional connectivity analysis methods. However, the dynamics alterations of brain functional networks during pain-free periovulation phase remain unclear. Using the co-activation pattern (CAP) method, we investigated the dynamic network characteristics of brain functional networks and their relationship with pain-related emotions in a sample of 59 women with PDM and 57 demographically matched healthy controls (HCs) during the pain-free periovulation phase. We observed that patients with PDM showed significant alterations in brain dynamics compared to HCs in the slow-4 (0.027-0.073 Hz) frequency band during the pain-free periovulation phase. Additionally, the fraction of time for CAP state 2 was positively correlated with the Pain Catastrophizing Scale-helplessness score, while the persistence time for CAP state 1 was positively correlated with the McGill Pain Questionnaire score. Our results provide new insights, suggesting that the atypical brain functional network dynamics may serve as a potential biological marker of patients with PDM during the pain-free periovulation phase.

Engaging dystonia networks with subthalamic stimulation

Deep brain stimulation is an efficacious treatment for dystonia. While the internal pallidum serves as the primary target, recently, stimulation of the subthalamic nucleus (STN) has been investigated. However, optimal targeting within this structure and its surroundings have not been studied in depth. Indeed, historical targets that have been used for surgical treatment of dystonia are directly adjacent to the STN. Further, multiple types of dystonia exist, and outcomes are variable, suggesting that not all types would profit maximally from the same target. Therefore, a thorough investigation of neural substrates underlying stimulation effects on dystonia signs and symptoms is warranted. Here, we analyze a multicenter cohort of isolated dystonia patients with subthalamic implantations (N = 58) and relate their stimulation sites to improvements of appendicular and cervical symptoms as well as blepharospasm. Stimulation of the ventral oral posterior nucleus of thalamus and surrounding regions were associated with improvements in cervical dystonia, while stimulation of the dorsolateral STN was associated with improvements in limb dystonia and blepharospasm. This dissociation was matched by structural connectivity analysis, where the cerebellothalamic, corticospinal, and pallidosubthalamic tracts were associated with improvements of cervical dystonia, while hyperdirect and subthalamopallidal pathways with alleviation of limb dystonia and blepharospasm. On the level of functional networks, improvements of limb dystonia were associated with connectivity to the corresponding somatotopic regions in the primary motor cortex, while alleviation of cervical dystonia to the cingulo-opercular network. These findings shed light on the pathophysiology of dystonia and may guide DBS targeting and programming in the future.

Molecular Evolution of Paralogous Cold Shock Proteins in E. coli: A Study of Asymmetric Divergence and Protein Functional Networks.

Nine homologous Cold Shock Proteins (Csps) have been recognized in the E.coli Cold Shock Domain gene family. These Csps function as RNA chaperones. This study aims to establish the evolutionary relationships among these genes by identifying and classifying their paralogous counterparts. It focuses on the physicochemical, structural, and functional analysis of the genes to explore the phylogeny of the Csp gene family. Computational tools were employed for protein molecular modeling, conformational analysis, functional studies, and duplication-divergence assessments. The research also examined amino acid conservation, protein mutations, domain-motif patterns, and evolutionary residue communities to better understand residual interactions, evolutionary coupling, and co-evolution. H33, M5, W11 and F53 residues were highly conserved within the protein family. It was further seen that residues M5, G17, G58, G61, P62, A64, V67 were intolerant to any kind of mutation whereas G3, D40, G41, Y42, S44, T54, T68, S69 were most tolerable towards substitutions. The study of residue communities displayed that the strongest residue coupling was observed in N13, F18, S27, F31, and W11. It was observed that all the gene pairs except CspF/CspH had new motifs generated over time. It was ascertained that all the gene pairs underwent asymmetric expression divergence after duplication. The Ka/ Ks ratio also revealed that all residues undertook neutral and purifying selection pressure. New functions were seen to develop in gene pairs evident from generation of new motifs. The discovery of new motifs and functions in Csps highlights their adaptive versatility, crucial for E. coli's resilience to environmental stressors and valuable for understanding bacterial stress response mechanisms. These findings will pave the way for future investigations into Csp evolution, with potential applications in microbial ecology and antimicrobial strategy development.

Alterations in cerebral perfusion and corresponding brain functional networks in systemic lupus erythematosus with cognitive impairment

Cognitive impairment (CI) frequently occurs in patients with systemic lupus erythematosus (SLE) and may result from neuroinflammation processes and neurovascular changes in the brain. The cerebral hemodynamics underlying SLE with CI (SLE-CI) remain unclear. 97 patients with SLE and 51 heathy controls (HCs) matched for age and gender underwent MRI. The CI status of patients was measured using the MoCA, and we classify those with a score of 28 or above as the SLE cognitive normal group (SLE-NC). 3D T1-weighted, ASL and resting-state functional (rs-fMRI) sequences were obtained. Seed-based functional connectivity (FC) was calculated using the cerebral blood flow (CBF) results. Compared with SLE-NC, patients with SLE-CI had higher CBF in the left hippocampus, thalamus, and cerebellum crus II and lower CBF in the left frontal lobe. Secondary analyses revealed that compared with patients with SLE-NC, patients with SLE-CI had increased FC of the left insula gyrus when the left cerebellum crus II was set as the seed region and decreased FC in the homolateral para-hippocampus when the left hippocampus was set as the seed region. These structural, functional, and network changes may serve as potential biomarkers for cognitive impairment in SLE-CI patients.

Large-scale network mechanisms underlying postoperative cognitive improvement after spine surgery.

The outcome of major surgery is determined not only by the success of the procedure itself but also by its neurocognitive effects. We previously reported improved cognition following spine surgery (Müller et al. 2023 Spine), but the mechanisms underlying these changes remain unknown. We analyzed resting-state (rs) functional magnetic resonance images of 79 patients (mean/SD age: 71/7years) acquired at baseline in this previously published trial. For 26 patients, data was additionally available at 3-months follow-up visits. To delineate large-scale connectivity, we calculated functional connectivity (FC) within and between three core neural networks, the central executive network (CEN), the salience network (SAL), and the default mode network (DMN). FC between CEN and SAL predicted cognitive improvement (beta=0.36, 95%-CI 0.28 to 0.45, P=0.033). Average FC between all nodes of the CEN showed changes towards an increase after surgery (beta=0.057, 95%-CI-0.01 to 0.123, P=0.086). Further seed-based FC analyses revealed that this increase was most pronounced in the functional coupling between left dorsolateral prefrontal and right posterior parietal cortex (beta=0.10, T(24)=2.73, Punc=0.012, PFDR=0.035). The increase of CEN-FC correlated with individual enhancements of executive scores (beta=0.34, 95%-CI 0.32 to 0.36, P=0.034). Integration of activity between the CEN and SAL networks predicted postoperative cognitive improvements, suggesting that less segregated large-scale functional networks may facilitate beneficial cognitive changes following surgery. Postoperative increases in functional coupling may serve as a biomarker for individual improvements in executive functions. These results indicate that surgery should not be routinely deferred in elderly patients due to concerns about postoperative neurocognitive complications. Moreover, our findings highlight potential targets for non-invasive brain stimulation interventions aimed at preventing neurocognitive complications.

Nauka podziemna, odczytana (i napisana) na nowo (Friedrich Cain, Wissen im Untergrund. Praxis und Politik klandestiner Forschung im besetzten Polen (1939–1945), Mohr Siebeck, Tübingen 2021 (Historische Wissensforschung, XIV), ss. 534)

Underground science, re-read (and rewritten) Friedrich Caina goes beyond the existing reconstructive and factual descriptions of underground science. He focuses on three, broadly defined areas: social sciences (mainly sociology), medicine, as well as physics, astronomy and technical sciences. They provide a starting point for related issues and case studies. Taking into account both macro and micro levels, the author focuses on individual issues, phenomena and characters, reconstructing contexts, motivations, limitations and frameworks for action, risks and benefits, daily practices and the functioning of social networks, etc. He defines and contextualises ‘underground science’ in an innovative way, pointing to its psychological, material or political determinants. Scholarly work made it possible, on the one hand, to regain a sense of subjectivity and agency destroyed by the occupiers, and on the other involved compromises, entering spaces that were associated with collaboration, such as Rudolf Weigl’s anti-typhus institute in Lviv or work at the Institut für Deutsche Ostarbeit in Kraków. At the same time Cain sees underground science as an important form of resistance.

Item Classification by Difficulty Using Functional Principal Component Clustering and Neural Networks.

Maintaining consistent item difficulty across test forms is crucial for accurately and fairly classifying examinees into pass or fail categories. This article presents a practical procedure for classifying items based on difficulty levels using functional data analysis (FDA). Methodologically, we clustered item characteristic curves (ICCs) into difficulty groups by analyzing their functional principal components (FPCs) and then employed a neural network to predict difficulty for ICCs. Given the degree of similarity between many ICCs, categorizing items by difficulty can be challenging. The strength of this method lies in its ability to provide an empirical and consistent process for item classification, as opposed to relying solely on visual inspection. The findings reveal that most discrepancies between visual classification and FDA results differed by only one adjacent difficulty level. Approximately 67% of these discrepancies involved items in the medium to hard range being categorized into higher difficulty levels by FDA, while the remaining third involved very easy to easy items being classified into lower levels. The neural network, trained on these data, achieved an accuracy of 79.6%, with misclassifications also differing by only one adjacent difficulty level compared to FDA clustering. The method demonstrates an efficient and practical procedure for classifying test items, especially beneficial in testing programs where smaller volumes of examinees tested at various times throughout the year.