Producción científica y docente
Producción científica reciente
SPE-ECG: An Ontology-Guided Knowledge Representation Framework for Enhancing LLM-Based Code Generation
Large Language Models (LLMs) can translate natural-language instructions into code, but when the input is unstructured clinical protocol text, the generated code may omit doses, monitoring intervals, or safety thresholds that the text states. Such omissions matter because they can propagate into software that supports clinical monitoring. We propose SPE-ECG (Structured Prompt Engineering for Enhanced Code Generation), a transparent, rule-based pipeline that converts unstructured domain text into a JSON prompt using text cleaning, part-of-speech tagging, bigram matching, and hand-built domain dictionaries, without fine-tuning any model. In a case study of 43 rheumatoid arthritis DMARD monitoring protocols, we compared raw and structured prompts on three LLMs (GPT-3.5, LLaMA 2, and Gemma), scoring the generated Python code using version 3.15.1 with BLEU against the raw-prompt code of the other two models. Mean BLEU rose for all three models (relative gains of 55.1%, 32.5%, and 19.9%); the paired differences were significant for GPT-3.5 and LLaMA 2 but not, after correction for multiple comparisons, for Gemma. These results suggest that lightweight input structuring can change how closely different models’ outputs agree, and that it could serve as a low-cost, auditable preparation step for protocol-to-code tools built on free or open models. BLEU does not measure correctness, no ablation or state-of-the-art baseline was run, and clinical validation is required before any use.
Initial validity evidence for a game-based digital assessment of phonological awareness in early readers
This study provides initial validity evidence, based on internal structure, together with reliability and grade-level sensitivity data for JILEC-1 (Interactive Language and Literacy Games for Children – Module 1), a seven-game digital phonological awareness battery for Brazilian Portuguese. Participants were 632 children (mean age 6.77 years) from the last year of kindergarten through Grade 2 in three Brazilian schools. Performance was indexed by the Game Progress Index (GPI), a weighted composite of accuracy, speed, and efficiency; results proved invariant to weighting decisions (rank-order agreement across five schemes ρ ≥ .99). Internal consistency was high (α = ω = .87). Exploratory factor analysis (n = 316) supported a dominant factor; confirmatory analysis in an independent subsample, using a calibration-confirmation design, supported a unidimensional model with two theoretically specified residual covariances (confirmation set: CFI = .999; RMSEA = .019). Comparison with correlated-factor and bifactor alternatives characterized the battery as essentially unidimensional (ωH = .84; ECV = .85). Full scalar measurement invariance held across sex; across grades, only configural invariance held, attributable to Grade 2 ceiling effects. GPI scores increased across grades (η2 = .49), robust to school clustering. Convergent, predictive, test-retest, and clinical-sample evidence is required before screening applications.
Fine-Grained Human Activity Recognition in Domestic Settings Using Ultra-Wideband Radar: A Comparative Study of Deep Learning Models
Human activity recognition with ultra-wideband (UWB) radar offers contactless, privacy-preserving monitoring of daily living, but distinguishing fine-grained activities performed in the same posture remains difficult. We recorded nine daily activities from 30 participants with a single PulsON 410 radar (3.1–5.3 GHz); seven are performed seated in the same chair and differ only by subtle hand and arm motions. Each 15 s recording is a 120×1440 channel–impulse–response matrix, preprocessed by background subtraction, two-dimensional wavelet denoising and standardization. Six deep learning models (Transformer, CNN-LSTM, 1D-CNN, LSTM, GRU and MLP) and a soft-voting ensemble were compared with participant-independent splits (22 training, 3 validation and 5 test participants) and five-fold cross-validation. The Transformer reached the highest cross-validation accuracy (97.74±0.30% ). On the held-out participants, the Transformer (97.16%), 1D-CNN (96.93%) and GRU (96.45%) performed similarly, ahead of the CNN-LSTM, LSTM and MLP (87.47%), and a three-model ensemble reached 98.35%. Two additional participant splits and a retrained re-run confirmed that the MLP is clearly the weakest model, but the order of the other models changed by several points between runs; in the re-run, an ensemble of all six models did not exceed the Transformer (97.74%), and a range–Doppler baseline reached only 50.63%. Eating versus drinking tea and mobile phone use versus reading or sitting idle were the most confused pairs. Given the small test cohort and the controlled setting, the results are indicative rather than definitive.
Hybrid deep learning and CNN-Isotonic Regression framework for accurate global mean sea level rise prediction using multi-source atmospheric and oceanic data
Keeping track of Global Mean Sea Level (GMSL) over long periods is important for assessing coastal risk. However, deep learning models may produce predictions that exhibit short-term increases and decreases, potentially conflicting with expected long-term physical behavior. In this study, we propose a combined Convolutional Neural Network and Isotonic Regression framework, termed CNN-ISO, for modeling and forecasting annual GMSL changes over the period 1880–2024. The framework uses 11 predictors representing atmospheric, oceanic, and cryospheric conditions. To provide longer temporal sequences, the original 144 annual observations are linearly interpolated into 1728 monthly observations. Model evaluation follows a strict chronological split, with training data from 1880–1979, validation data from 1980–1999, and test data from 2000–2024. All preprocessing transformations and the isotonic mapping are fitted exclusively on the training set to prevent information leakage from later periods. On the unseen test period, CNN-ISO achieves R 2 = 0.997 with a 95% confidence interval of [ 0.995,0.999 ] , together with an MAE of 0.56 mm and an RMSE of 0.72 mm. For comparison, linear regression achieves R 2 = 0.982 . Although the predictive improvement over linear regression is relatively modest, CNN-ISO provides an additional constraint by producing monotonic out-of-sample projections, which is consistent with the assumed long-term behavior of GMSL. Under statistical trend extrapolation to 2050, CNN-ISO estimates a GMSL increase of 184.10 mm relative to the baseline. However, estimates vary substantially across models, with linear regression and LSTM producing 116.60 mm and 299.19 mm, respectively. These results should therefore be interpreted as trend-based statistical extrapolations rather than comprehensive climate projections, as the framework does not explicitly model ice-sheet dynamics, changes in ocean circulation, or future emission pathways. Nevertheless, the proposed approach provides a transparent framework for examining model uncertainty and may support early-stage coastal adaptation planning.
Cost-sensitive random forest with adaptive threshold optimization for imbalanced financial fraud detection
Background: Financial transaction fraud detection is a challenging binary classification problem characterized by severe class imbalance, rare fraudulent cases, and asymmetric misclassification costs. In real-world banking systems, failing to detect fraudulent transactions can lead to substantial financial losses, whereas false alarms mainly increase operational review costs. Therefore, fraud detection models should not be optimized only for overall accuracy, as accuracy-oriented classifiers may favor the majority non-fraud class and fail to identify minority fraud cases effectively. To address this issue, this study proposes a cost-sensitive ML framework for financial transaction fraud detection under class imbalance and asymmetric financial risk.Methods: In this study, a cost-sensitive RF framework with adaptive probability threshold optimization is proposed for binary fraud classification. Misclassification costs are incorporated during model training through class-weighted learning, allowing the classifier to assign greater importance to minority fraud cases. In addition, instead of using the default probability threshold of 0.5, the decision threshold is optimized using validation data to minimize the expected misclassification cost and improve minority-class detection. The proposed framework was evaluated on two publicly available financial fraud datasets with different transaction characteristics: the Credit Card Fraud Detection dataset and the Financial Fraud Prediction dataset. Its performance was compared with conventional ML baselines using fraud-oriented evaluation metrics, including precision, recall, F1-score, false-negative rate, false-positive rate, and expected misclassification cost.Results: The proposed framework achieved strong and consistent performance across both datasets. On the Credit Card Fraud Detection dataset, it achieved a precision of 94.4%, recall of 85.7%, and F1-score of 89.8%. On the Financial Fraud Prediction dataset, it achieved a precision of 93.1%, a recall of 83.8%, and an F1-score of 88.2%. Compared with conventional fixed-threshold ML models, the adaptive threshold strategy reduced false-negative predictions while maintaining high precision. These results indicate that the proposed cost-sensitive framework provides a more balanced decision strategy for fraud detection under highly imbalanced transaction distributions.Conclusion: The findings demonstrate that combining class-weighted learning with adaptive threshold optimization can improve fraud detection performance under asymmetric misclassification costs. The proposed framework offers a practical and interpretable decision-support approach for banking fraud detection systems, particularly in settings where reducing undetected fraudulent transactions is more important than maximizing overall accuracy alone.