Learning Domain Research Articles

Non-linear development in statistical learning of visual orthographic regularities

Statistical learning is a core ability for individuals in extracting and integrating regularities and patterns from linguistic input. Yet, the developmental trajectory of visual statistical learning has not been fully examined in the orthographic learning domain. Employing an artificial orthographic learning task, we manipulated three levels of positional consistency of radicals, i.e., high (100%), moderate (80%), and low (60%), embedded in pseudocharacters to investigate visual statistical learning across a wide age range between 4–12-year-olds and adults. The non-linear power-function models indicated that the rates of improvement in acquiring varying positional consistencies increased with age, particularly for high and moderate levels. Specifically, we observed a significant enhancement in statistical learning abilities between the ages of 4–5 years and 5–6 years, followed by a stabilization of performance after 8–9 years. Our findings support the age-dependent perspective that individuals’ visual statistical learning ability improves significantly in early childhood and then decelerates its improvement progressively until adulthood.

Integrating drone technology in STEM education: Curriculum, pedagogy and learning outcomes

Abstract This study explores the integration of drone technology into STEM education through the development and evaluation of the Drone Technology Enabled STEM Curriculum (DTESC). Grounded in “dronagogy”, a pedagogical framework that utilizes drones to enhance learning, DTESC effectively incorporates drone technology within science, mathematics, technology, and the humanities. The curriculum is structured to progress from multidisciplinary exploration to interdisciplinary integration, ultimately culminating in transdisciplinary applications. Sixteen Grade 9–10 students from a secondary school in Hong Kong participated in the study, engaging in activities such as designing drone-assisted water sampling systems and formulating entrepreneurial proposals for future applications. Employing mixed research methods, the study assessed the curriculum’s impact across cognitive, psychomotor, and affective learning domains. Results indicated high levels of student engagement and satisfaction, alongside significant improvements in psychomotor and affective skills. However, enhancements in cognitive learning outcomes, particularly in mathematics and programming, are necessary. This research offers valuable insights into the implementation of dronagogy to foster cross-disciplinary STEM education.

Co-Cultural Identities of Thai-Lao: A Guideline for Cultural Curriculum Development

This qualitative study explores the Thai-Lao co-cultural identities and develops guidelines for integrating these identities into cultural studies curricula. Data were collected through semi-structured interviews with 14 stakeholders in education and cultural preservation from both the Kingdom of Thailand and Lao PDR. Thematic analysis was employed to interpret the data, focusing on historical, literary, and belief systems that characterize Thai-Lao co-cultural identities. The findings highlight key aspects of these identities and propose a curriculum framework that addresses cognitive, psychomotor, and affective learning domains. This framework is designed to enhance students’ understanding of their cultural heritage and to foster intercultural competence. The study contributes to the fields of cultural studies and education by offering insights into the integration of co-cultural identities into educational practices, promoting a richer, more inclusive educational experience.

BreastHybridNet: A Hybrid Deep Learning Framework for Breast Cancer Diagnosis Using Mammogram Images

As a common malignancy in females, breast cancer represents one of the most serious threats to a female's life, which is also closely associated with the Sustainable Development Goal 3 (SDG 3) of the United Nations for keeping healthy lives and promoting the well-being of all people. Breast cancer accounts for the highest number of cancer mortality for females, and early diagnosis is key to reducing disease-specific mortality and mortality in general. Current methods struggle to accurately localize important regions, model sequential dependencies, or combine different features despite considerable improvements in artificial intelligence and deep learning domains. They prevent diagnostic frameworks from being reliable and scalable, especially in low-resourced healthcare settings. This study proposes a novel hybrid deep learning framework, BreastHybridNet, using mammogram images to tackle these mutual challenges. The proposed framework combines a pre-trained CNN backbone for feature extraction, a spatial attention mechanism to automatically highlight the image area, which contains signature patterns carrying diagnostic information, a BiLSTM layer to obtain sequential dependencies of diagnostic features, and a feature fusion strategy to process complementarily. Experimental results show that the accuracy of the proposed model is 98.30%, which outperforms the state-of-the-art methods LMHistNet, BreastMultiNet, and DOTNet 2.0 to a considerable extent quantitatively. BreastHybridNet works towards the feasibility of interpretability and scalability on existing systems while contributing to worldwide efforts to alleviate cancer-related mortality using cost-efficient diagnostic lenses. This study highlights the need for AI-enabled solutions to contribute to accessing reliable healthcare technologies for breast cancer screening.

Current Awareness of Removable Prosthodontics in Dentists at King Abdulaziz University

Background: Removable prosthodontics is a branch of prosthodontics that requires a high level of knowledge and great skills. This include the basic and advanced understanding of various principles and the practical skills that are crucial to deliver a good quality and satisfactory removable prostheses. To provide a successful prosthodontic treatment in partially dentate patients with removable partial denture (RPD), the awareness of basic RPD principles and design is mandatory.Therefore, academic education utilizes the good teaching strategies and precise assessment methods for the students to fulfil the required preclinical and clinical skills. In addition, to meet the planned level of competencies, the basic learning domains are required to be fulfilled. In removable prosthodontics, fulfilment of the required knowledge and skills are very crucial to deliver a high-quality prosthesis. In academic study, students are exposed to different learning experiences throughout their years of study to prepare them becoming competent graduates. Some students may face difficulties in understanding and applying the mechanobiological principles of removable partial denture. Therefore, the assessment of their knowledge and cognitive skills, as well as, the current teaching strategies is crucial to reflect upon those strategies and to plan for implementation of future teaching improvements.In this study, students and general dentists were assessed regarding their awareness and understanding of RPD principles and deign which is the initial step in performing this prosthodontic treatment option. Also, pre-lecture and post-lecture assessment was planned to reflect upon the teaching strategy in removable prosthodontics among undergraduate students. Materials and methods: A well-structured online questionnaire in removable partial denture design and principles was formulated to assess the knowledge and cognitive skills of undergraduate, postgraduate students and general dentists at KAU. Also, a pre-lecture and post-lecture assessments were carried out for fifth-year undergraduate students. Results: The proportion of correct answers were higher in female than male students among postgraduates and undergraduates (P<0.05). Also, Postgraduate students showed the highest level of proportion compared to the undergraduates (P<0.05). In addition, a significant higher proportion and scores of correct answers were reported in the after-lecture response among fifth-year students (P<0.05). Conclusion: The level of awareness of removable partial denture increases with the level of experience and lectures are considered as the corner stone in teaching of his branch of prosthodontics

A reliable and privacy-preserved federated learning framework for real-time smoking prediction in healthcare

The ever-evolving domain of machine learning has witnessed significant advancements with the advent of federated learning, a paradigm revered for its capacity to facilitate model training on decentralized data sources while upholding data confidentiality. This research introduces a federated learning-based framework designed to address gaps in existing smoking prediction models, which often compromise privacy and lack data generalizability. By utilizing a distributed approach, the framework ensures secure, privacy-preserved model training on decentralized devices, enabling the capture of diverse smoking behavior patterns. The proposed framework incorporates careful data preprocessing, rational model architecture selection, and optimal parameter tuning to predict smoking with high precision. The results demonstrate the efficacy of the model, achieving an accuracy rate of 97.65%, complemented by an F1-score of 97.41%, precision of 97.31%, and recall rate of 97.36%, significantly outperforming traditional approaches. This research also discusses the benefits of federated learning, including efficient time management, parallel processing, secure model updates, and enhanced data privacy, while addressing limitations such as computational overhead. These findings underscore the transformative potential of federated learning in healthcare, paving the way for future advancements in privacy-preserved predictive modeling.

Pharmacogenomic curriculum in Australian medical schools: a content analysis study.

To ascertain and describe pharmacogenomic concepts included in the intended curriculum of accredited Australian medical schools. Content analysis of curriculum learning objectives of Australian medical schools was conducted, focusing on keywords and phrases pertaining to pharmacogenomic education. Learning objectives related to pharmacogenomics were categorized using (1) undergraduate medical genomic competencies per the Association of Professors in Human and Medical Genetics (2) Bloom's Taxonomy for cognitive and knowledge dimensions and (3) knowledge translation (enabling science, translation science and clinical implementation). The curricula of 19 accredited medical schools in Australia were analyzed. Two-thirds (68%) contained genomic/pharmacogenomic education. Eight schools had content relating to undergraduate medical genomic competencies. Of those which had pharmacogenomic-related learning objectives, the majority (65%) were categorized in Bloom's Taxonomy's lower levels (Remember and Understand) and 15% were deemed to be at the level of 'Clinical Implementation.' The majority of Australian medical schools have incorporated pharmacogenomics in their current curriculum; however, learning objectives addressing application and clinical implementation are required. Doctors have a unique role to play in implementing pharmacogenomics into clinical practice. Comprehensiveness of course curricula across all learning domains would support uptake of pharmacogenomics into routine practice.

Level Up with ML Vulnerability Identification: Leveraging Domain Constraints in Feature Space for Robust Android Malware Detection

Machine Learning (ML) promises to enhance the efficacy of Android Malware Detection (AMD); however, ML models are vulnerable to realistic evasion attacks—crafting realizable Adversarial Examples (AEs) that satisfy Android malware domain constraints. To eliminate ML vulnerabilities, defenders aim to identify susceptible regions in the feature space where ML models are prone to deception. The primary approach to identifying vulnerable regions involves investigating realizable AEs, but generating these feasible apps poses a challenge. For instance, previous work has relied on generating either feature-space norm-bounded AEs or problem-space realizable AEs in adversarial hardening. The former is efficient but lacks full coverage of vulnerable regions while the latter can uncover these regions by satisfying domain constraints but is known to be time-consuming. To address these limitations, we propose an approach to facilitate the identification of vulnerable regions. Specifically, we introduce a new interpretation of Android domain constraints in the feature space, followed by a novel technique that learns them. Our empirical evaluations across various evasion attacks indicate effective detection of AEs using learned domain constraints, with an average of 89.6%. Furthermore, extensive experiments on different Android malware detectors demonstrate that utilizing our learned domain constraints in Adversarial Training (AT) outperforms other AT-based defenses that rely on norm-bounded AEs or state-of-the-art non-uniform perturbations. Finally, we show that retraining a malware detector with a wide variety of feature-space realizable AEs results in a 77.9% robustness improvement against realizable AEs generated by unknown problem-space transformations, with up to 70 × faster training than using problem-space realizable AEs.

A Multi-Source Domain Adaptation Method for Bearing Fault Diagnosis with Dynamically Similarity Guidance on Incomplete Data

In actual industrial scenarios, collecting a complete dataset with all fault categories under the same conditions is challenging, leading to a loss in fault category knowledge in single-source domains. Deep learning domain adaptation methods face difficulties in multi-source scenarios due to insufficient labeled data and significant distribution differences, hindering domain-specific knowledge transfer and reducing fault diagnosis efficiency. To address these issues, the Dynamic Similarity-guided Multi-source Domain Adaptation Network (DS-MDAN) is proposed. This method leverages incomplete data from multiple-source domains to address distribution disparities in deep domain adaptation. It enhances diagnostic performance in the target domain by transferring knowledge across diverse domains. DS-MDAN uses convolution kernels of different scales to extract multi-scale feature information and achieves feature fusion through upsampling and operations like addition and concatenation. Adversarial training with domain and fault classifiers optimizes feature extraction for widely applicable representations. The similarity between source and target domain data is calculated based on features extracted by a shared-weight network, dynamically adjusting the contribution of different source domain data to minimize distribution differences. Finally, matched source and target domain samples are mapped to the same feature space for fault diagnosis. Experimental validation on various bearing fault datasets shows that DS-MDAN improves performance in multiple fault diagnosis tasks, increasing accuracy and demonstrating good generalization capabilities.

Affective Computing for Learning in Education: A Systematic Review and Bibliometric Analysis

Affective computing is an emerging area of education research and has the potential to enhance educational outcomes. Despite the growing number of literature studies, there are still deficiencies and gaps in the domain of affective computing in education. In this study, we systematically review affective computing in the education domain. Methods: We queried four well-known research databases, namely the Web of Science Core Collection, IEEE Xplore, ACM Digital Library, and PubMed, using specific keywords for papers published between January 2010 and July 2023. Various relevant data items are extracted and classified based on a set of 15 extensive research questions. Following the PRISMA 2020 guidelines, a total of 175 studies were selected and reviewed in this work from among 3102 articles screened. The data show an increasing trend in publications within this domain. The most common research purpose involves designing emotion recognition/expression systems. Conventional textual questionnaires remain the most popular channels for affective measurement. Classrooms are identified as the primary research environments; the largest research sample group is university students. Learning domains are mainly associated with science, technology, engineering, and mathematics (STEM) courses. The bibliometric analysis reveals that most publications are affiliated with the USA. The studies are primarily published in journals, with the majority appearing in the Frontiers in Psychology journal. Research gaps, challenges, and potential directions for future research are explored. This review synthesizes current knowledge regarding the application of affective computing in the education sector. This knowledge is useful for future directions to help educational researchers, policymakers, and practitioners deploy affective computing technology to broaden educational practices.

Cardiac MR image reconstruction using cascaded hybrid dual domain deep learning framework.

Recovering diagnostic-quality cardiac MR images from highly under-sampled data is a current research focus, particularly in addressing cardiac and respiratory motion. Techniques such as Compressed Sensing (CS) and Parallel Imaging (pMRI) have been proposed to accelerate MRI data acquisition and improve image quality. However, these methods have limitations in high spatial-resolution applications, often resulting in blurring or residual artifacts. Recently, deep learning-based techniques have gained attention for their accuracy and efficiency in image reconstruction. Deep learning-based MR image reconstruction methods are divided into two categories: (a) single domain methods (image domain learning and k-space domain learning) and (b) cross/dual domain methods. Single domain methods, which typically use U-Net in either the image or k-space domain, fail to fully exploit the correlation between these domains. This paper introduces a dual-domain deep learning approach that incorporates multi-coil data consistency (MCDC) layers for reconstructing cardiac MR images from 1-D Variable Density (VD) random under-sampled data. The proposed hybrid dual-domain deep learning models integrate data from both the domains to improve image quality, reduce artifacts, and enhance overall robustness and accuracy of the reconstruction process. Experimental results demonstrate that the proposed methods outperform than conventional deep learning and CS techniques, as evidenced by higher Structural Similarity Index (SSIM), lower Root Mean Square Error (RMSE), and higher Peak Signal-to-Noise Ratio (PSNR).

Combating trade in illegal wood and forest products with machine learning.

Trade in wood and forest products spans the global supply chain. Illegal logging and associated trade in forest products present a persistent threat to vulnerable ecosystems and communities. Illegal timber trade has been linked to violations of tax and conservation laws, as well as broader transnational crimes. The United States is the largest importer globally of wood and forest products, such as pulp, paper, flooring, and furniture-importing $78 billion in 2021. Transaction-level data such as shipping container manifests and bills of lading provide a comprehensive data source that can be used to detect and disrupt trade that may be suspected of containing illegally harvested or traded forest products. Owing to the volume, velocity, and complexity of shipment data, an automated decision support system is required for the purposes of detecting suspicious forest product shipments. We present a proof of concept framework using machine learning and big data approaches-combining domain expertise with automation-to achieve this objective. We formulated the underlying machine learning problem as an anomaly detection problem and collected and collated forest sector-specific domain knowledge to filter and target shipments of interest. In this work, we provide the overview of our framework, with the details of domain knowledge extraction and machine learning models, and discuss initial results and analysis of flagged anomalous and potentially suspicious records to demonstrate the efficacy of this approach. The proof of concept work presented here provides the groundwork for an actionable and feasible approach to assisting enforcement agencies with the detection of suspicious shipments that may contain illegally harvested or traded wood.

Caregiver Psychosocial Factors & Stroke Survivor Cognitive Outcomes: A REGARDS-CARES Cohort Study.

Caring for an individual with cognitive impairment carries a physical, mental, and emotional toll. This manuscript examines the relationship between caregiver psychosocial measures and longitudinal cognitive outcomes of stroke survivors, as well as analyzing the psychosocial factors as moderators of stroke severity and cognition. This analysis was conducted on caregiver and stroke survivor dyads (n=157) that participated in the Caring for Adults Recovering from the Effects of Stroke (CARES) project, an ancillary study of the REasons for Geographic and Racial Differences in Stroke (REGARDS) national cohort study. Stroke severity at hospitalization discharge was included as the primary predictor of cognitive outcomes and caregiver psychosocial factors were included as additional predictors. Cognition was assessed biennially and measured the domains of learning, memory, and executive functioning. Individual mixed-effect models included each psychosocial factor and were covariate-adjusted for pre-stroke cognitive scores and demographic variables. Caregiver psychosocial factors included caregiver strain, depressive symptoms, life and leisure time satisfaction, and overall quality of life. Decreased caregiver strain (b=-0.230, 95% CI: -0.39 to -0.07; p=0.006) and increased leisure time satisfaction (b=0.045, 95% CI: 0.01 to 0.08; p=0.005) were both found to be significant predictors, alongside stroke severity (b=-0.137, 95% CI: -0.22 to -0.05; p=0.002), of better stroke survivor cognition overall. No variables were found to be moderating factors of the relationship between stroke severity and cognition. Understanding the caregiver psychosocial factors that predict stroke outcomes will help clinicians to identify stroke survivor and caregiver dyads at higher risk for worst longitudinal cognitive outcomes following stroke.

Pengaruh Pendekatan Pembelajaran SAVI (Somatic, Auditory, Visual, Intellectual) terhadap Hasil Belajar Matematika Peserta Didik Materi Bangun Ruang Kelas IV SDN Randegan

This study aimed to determine the effect of the SAVI learning approach (Somatic, Auditory, Visual, and Intellectual) on student's mathematics learning outcomes in three-dimensional materials for grade IV SDN Randegan. This study used quantitative methods with a one-group pretest-posttest design. The population of this study was all IV grade students of SDN Randegan in the school year of 2023/2024. The Sampling used in this study was a total sampling technique. The sample in this study consisted of 28 students. Data were collected using pretest and posttest questions, psychomotor assessment, and observation. The results of this study indicate that: 1.) the learning outcomes of students in the cognitive domain got a t-observed of -19.357 ≥ -1.701 and in the psychomotor domain obtained a t-observed of -17.860 ≥ -1.701 (negative signs only indicating the direction of hypothesis testing, not indicating the amount) which means there is a significant difference between pretest scores and posttest scores in each learning domain, 2). Because there is a significant difference between the pretest and posttest values in each learning domain, it can be concluded that there is an effect of the SAVI learning approach (Somatic, Auditory, Visual, and Intellectual) on student's mathematics learning outcomes in three-dimensional material for grade IV of SDN Randegan.

Churn Detection and User Classification via Machine Learning in the Food and Beverage Sector

In the modern business world, detecting and predicting customer behavior is one of the key factors for any operation to achieve their goals. Customer churn is one of these behaviors of interest, which makes churn detection and prediction a hot topic in the Machine Learning domain. The customer data that was studied is obtained from a global food and beverage company’s operations in Turkey: Their gift-based mobile application rewards customers who buy their products, and the user data of many sorts is stored within its database. In this study, the unlabeled customer data of a large scale was analyzed and classified via the combination of various supervised and unsupervised ML methods such as K-Means Clustering, Random Forest, Support Vector Machines, Logistic Regression, XGBoost. Then, a score-based churn detection & prediction algorithm is developed after picking the best performing models based on their performance metrics.

SchNet_IIA: Potential Energy Surface Fitting by Interatomic Interactions Attention Based on Transfer Learning Analysis.

Machine learning methods for fitting potential energy surfaces and molecular dynamics simulations are becoming increasingly popular due to their potentially high accuracy and savings in computational resources. However, existing application models often rely on basic architectures like artificial neural networks (ANNs) and multilayer perceptron (MLP), lagging behind cutting-edge technologies in the machine learning domain. Furthermore, the complexity of current machine learning frameworks leads to reduced interpretability and challenges for improvement. Herein, we developed a model analysis method based on the feature-representation-transfer approach to directly perform causal analysis on the model. The internal action characteristics of the SchNet framework were successfully analyzed by constructing different source tasks and we proposed interatomic interactions attention for the characterization of doped clusters. The accuracy was enhanced by 0.015 eV/atom compared to the original model. The ability to capture atomic environment characteristics was significantly improved. The activation function was smoothed resulting in a 23.47% increase in the convergence speed. Our SchNet_IIA model demonstrates superior performance in capturing interatomic interactions. Our present work is of distinctive value as it presents a novel transfer learning analysis method with the potential to evolve into a generalized model analysis approach, providing new perspectives and solutions for the field.

Leveraging Transfer Learning Domain Adaptation Model With Federated Learning to Revolutionise Healthcare

ABSTRACTThe application of artificial intelligence (AI) in healthcare has been witnessing an increasing interest. Particularly, federated learning (FL) has become favourable due to its potential for enhancing model quality whilst maintaining data privacy and security. However, the effectiveness of present FL methodologies could underperform under non‐IID conditions, characterised by divergent data distributions across clients. The globally constructed FL model may suffer potent issues by allowing the least‐performing models to equal participation. Thus, we propose a new accuracy‐based FL approach (FedAcc) which only takes into account the clients' validation accuracy to consider their participation during global aggregation, also called Smart Healthcare Amplified (SHA). However, with limited supervised data it is challenging to increase the model performance thus concept of transfer learning (TL) is used. TL enables the global model to integrate knowledge from precomputed systems, resulting in an efficient model. However, the complexity of the global system is amplified by these TL models, leading to challenges related to vanishing gradients, particularly when dealing with a substantial number of layers. To mitigate this, we present a Transfer Learning Domain Adaptation Model (TLDAM). TLDAM employs a two‐layered sequentially trained TL model, which contains approximately 50% fewer layers compared to traditional TL models. TLDAM is trained on multiple datasets such as MNIST and CIFAR10, to enhance its knowledge and make it domain‐adaptive. Moreover, experimental results conducted on the UCI‐HAR dataset reveal the supremacy of our proposed framework with an accuracy of 94.2990%, F‐score of 94.2820%, precision of 94.3058%, and recall of 94.2993% over traditional FL techniques and state‐of‐the‐art techniques.

Accelerating Awareness: Hidden Equity, Diversity, and Inclusion Learning Outcomes within an Immersive Classroom Activity

This work addresses the challenge of meaningfully introducing engineering students to equity, diversity, and inclusion (EDI) principles. Traditionally, tertiary education introduces these concepts through online modules and/or workshops, often ‘down-selling’ them compared to technical courses. This paper proposes a more impactful experiential approach targeting the affective learning domain. In collaboration with the Centre for Student Wellness, instructors designed a lesson integrating EDI principles into an interactive car simulation game. Students form teams and customize their cars based on survey responses, with rewards and penalties reflecting real-world inequalities. For instance, certain car colors or gender characteristics may result in penalties, mirroring observed racism and gender biases in STEM fields. This innovative approach aims to foster critical thinking about the role of personal privilege in decision-making, offering a dynamic learning experience beyond conventional methods.

Domain Adversarial Convolutional Neural Network Improves the Accuracy and Generalizability of Wearable Sleep Assessment Technology.

Wearable accelerometers are widely used as an ecologically valid and scalable solution for long-term at-home sleep monitoring in both clinical research and care. In this study, we applied a deep learning domain adversarial convolutional neural network (DACNN) model to this task and demonstrated that this new model outperformed existing sleep algorithms in classifying sleep-wake and estimating sleep outcomes based on wrist-worn accelerometry. This model generalized well to another dataset based on different wearable devices and activity counts, achieving an accuracy of 80.1% (sensitivity 84% and specificity 58%). Compared to commonly used sleep algorithms, this model resulted in the smallest error in wake after sleep onset (MAE of 48.7, Cole-Kripke of 86.2, Sadeh of 108.2, z-angle of 57.5) and sleep efficiency (MAE of 11.8, Cole-Kripke of 18.4, Sadeh of 23.3, z-angle of 9.3) outcomes. Despite being around for many years, accelerometer-alone devices continue to be useful due to their low cost, long battery life, and ease of use. Improving the accuracy and generalizability of sleep algorithms for accelerometer wrist devices is of utmost importance. We here demonstrated that domain adversarial convolutional neural networks can improve the overall accuracy, especially the specificity, of sleep-wake classification using wrist-worn accelerometer data, substantiating its use as a scalable and valid approach for sleep outcome assessment in real life.

Impact of Psychomotor Domain in Enhancing Science Learning in Khyber Pakhtunkhwa, Pakistan: A Review

Learning can be divided into three categories: emotional (attitudes), psychomotor (skills), and cognitive (knowledge). The Taxonomy of Learning Domains, developed in 1956 by a team of academics under the direction of Benjamin Bloom, provides the clearest explanation of this grouping. The learning domains were initially developed and described between 1956 and 1972. Psychomotor goals are specific to reflexes, interpretive motions, and body processes. It's interesting to note that although cognitive and affective taxonomies were described in 1956 and 1964, respectively, the psychomotor domain was not fully developed until the 1970s.