In collaborative learning situations, monitoring is needed to maintain common progress toward shared goals. The present study aimed to analyze group-level monitoring events, as well as groups’ reactions to these events, to identify instances of adaptive regulation and maladaptive behavior. Three dimensions of monitoring events were qualitatively coded from video data: the monitoring target, valence, and phase, which provided insight into identifying critical moments during the collaborative process when regulation is needed. By looking at what kind of monitoring the groups engaged in, and how the groups progressed after the need for regulation arose, different types of adaptive regulation and maladaptive behavior were distinguished. In addition, group-level physiological state transitions in the heart rate were explored to see whether changes in regulation (adaptive regulation and maladaptive behavior) were reflected in the state transitions. Nine groups of three students each participated in a collaborative exam for an advanced high school physics course, during which video and heart rate data were collected. The results showed that on-track sequences were the most common, followed by adaptive sequences. The temporality of these sequences was examined, and four categories of group progress are described with case examples. A correlation analysis showed that physiological state transitions were positively correlated with on-track sequences. The opportunities and limitations of using three dimensions of monitoring and heart-rate based physiological state transitions to study adaptive regulation are discussed.
Computational requirements for data processing at different stages of the radiology value chain are increasing. Cone beam computed tomography (CBCT) is a diagnostic imaging technique used in dental and extremity imaging, involving a highly demanding image reconstruction task. In turn, artificial intelligence (AI) assisted diagnostics are becoming increasingly popular, thus increasing the use of computation resources. Furthermore, the need for fully independent imaging units outside radiology departments and with remotely performed diagnostics emphasize the need for wireless connectivity between the imaging unit and hospital infrastructure. In this feasibility study, we propose an approach based on a distributed edge-cloud computing platform, consisting of small-scale local edge nodes, edge servers with traditional cloud resources to perform data processing tasks in radiology. We are interested in the use of local computing resources with Graphics Processing Units (GPUs), in our case Jetson Xavier NX, for hosting the algorithms for two use-cases, namely image reconstruction in cone beam computed tomography and AI-assisted cancer detection from mammographic images. Particularly, we wanted to determine the technical requirements for local edge computing platform for these two tasks and whether CBCT image reconstruction and breast cancer detection tasks are possible in a diagnostically acceptable time frame. We validated the use-cases and the proposed edge computing platform in two stages. First, the algorithms were validated use-case-wise by comparing the computing performance of the edge nodes against a reference setup (regular workstation). Second, we performed qualitative evaluation on the edge computing platform by running the algorithms as nanoservices. Our results, obtained through real-life prototyping, indicate that it is possible and technically feasible to run both reconstruction and AI-assisted image analysis functions in a diagnostically acceptable computing time. Furthermore, based on the qualitative evaluation, we confirmed that the local edge computing capacity can be scaled up and down during runtime by adding or removing edge devices without the need for manual reconfigurations. We also found all previously implemented software components to be transferable as such. Overall, the results are promising and help in developing future applications, e.g., in mobile imaging scenarios, where such a platform is beneficial.
In the era of digital healthcare, electronic health records generate vast amounts of data, much of which is unstructured, and therefore, not in a usable format for conventional machine learning and artificial intelligence applications. This study investigates how to extract meaningful insights from unstructured radiology reports written in Finnish, a minority language, using machine learning techniques for text analysis. With this approach, unstructured information could be transformed into a structured format. The results of this research show that relevant information can be effectively extracted from Finnish medical reports using classification algorithms with default parameter values. For the detection of breast tumour mentions from medical texts, classifiers achieved high accuracy, almost 90%. Detection of metastasis mentions, however, proved more challenging, with the best-performing models Support Vector Machine (SVM) and logistic regression achieving an F1-score of 81%. The lower performance in metastasis detection is likely due to the more complex problem, ambiguous labeling, and the smaller dataset size. The results of classical classifiers were also compared with FinBERT, a domain-adapted Finnish BERT model. However, classical classifiers outperformed FinBERT. This highlights the challenge of medical language processing when working with minority languages. Moreover, it was noted that parameter tuning based on translated English reports did not significantly improve the detection rates, likely due to linguistic differences between the datasets. This larger translated dataset used for tuning comes from a different clinical domain and employs noticeably simpler, less nuanced language than the Finnish breast cancer reports, which are written by native Finnish-speaking medical experts. This underscores the need for localised datasets and models, particularly for minority languages with unique grammatical structures.
Accurate fatigue prediction using wearable sensor data is important for a range of digital health applications, including preventive health management. In real-world settings, however, fatigue datasets are typically small, heterogeneous, and severely imbalanced, which limits the reliability of conventional machine learning and deep learning approaches, especially for detecting less common but more pronounced fatigue conditions. This paper presents a hybrid generative and ensemble learning framework for robust fatigue-level prediction from wearable data. The proposed approach integrates a Conditional Variational Autoencoder (CVAE) with a Light Gradient Boosting Machine (LightGBM) classifier. The CVAE performs class-conditional generative oversampling to synthesize physiologically plausible samples for underrepresented fatigue levels, thereby mitigating data imbalance while preserving fatigue-specific characteristics. The resulting balanced dataset is then used to train a LightGBM classifier, which is well suited for heterogeneous, tabular wearable-sensor features commonly encountered in digital health systems. The framework is evaluated on two benchmark fatigue datasets derived from wearable sensors with different imbalance characteristics using stratified 10-fold cross-validation. Experimental results demonstrate consistent improvements over six baseline models, achieving Macro F1-scores of 0.947 and 0.965, respectively, while substantially reducing misclassification between adjacent fatigue levels. These findings suggest that combining class-aware generative modeling with ensemble-based discriminative learning is an effective strategy for fatigue prediction in imbalanced wearable-sensor datasets.