Abdullah Al-Barakati

Background: Discharge instructions are essential for patients after hospital care but are time-consuming to write. With the rise of large language models (LLMs), there is a strong potential to automate this process. This study explores the use of open-source LLMs for generating discharge instructions. Objective: We investigated whether a Mistral model can reliably generate patient-oriented discharge instructions. Two distinct instruction-tuning paradigms were compared, each using a different mechanism for embedding guidance during fine-tuning. Methods: In our experiment, we applied Mistral-NeMo-Instruct, an LLM, in combination with 2 distinct instruction strategies for fine-tuning. The first were detailed instructions tailored to the task of discharge instruction generation. The second was a basic instruction with minimal guidance and no task-specific detail. The independent variable in this study is the instruction strategy (detailed vs generic), while the dependent variables are the evaluation scores of the generated discharge instructions. The generated discharge instructions were evaluated against 3621 ground-truth references. We used Bilingual Evaluation Understudy (BLEU-1) to BLEU-4, Recall-Oriented Understudy for Gisting Evaluation (ROUGE-1, ROUGE-2, and Recall-Oriented Understudy for Gisting Evaluation—Longest Common Subsequence), SentenceTransformer similarity, and Bidirectional Encoder Representations From Transformers Score as evaluation metrics to assess the quality of the generated outputs in comparison to the corresponding ground-truth instructions for the same discharge summaries. Results: The detailed instruction model demonstrated superior performance across all automated evaluation metrics compared with the generic instruction model. Bidirectional Encoder Representations From Transformers Score increased from 78.92% to 87.05%, while structural alignment measured by Recall-Oriented Understudy for Gisting Evaluation—Longest Common Subsequence improved from 8.59% to 26.52%. N-gram precision (BLEU-4) increased from 0.81% to 21.24%, and the Metric for Evaluation of Translation With Explicit Ordering scores rose from 15.33% to 18.47%. Additional metrics showed consistent gains: ROUGE-1 improved from 16.59% to 42.72%, and ROUGE-2 increased from 1.97% to 45.84%. All improvements were statistically significant (P<.001), indicating that detailed, task-specific instruction design substantially enhances model performance. Conclusions: The use of detailed, task-specific instruction strategies significantly enhances the effectiveness of open-source LLMs in generating discharge instructions. These findings indicate that carefully designed instructions during fine-tuning substantially improve model performance.

Trying to understand the information transfer between neurons more clearly through synapses has been of great interest. In this work, a memristive autapse spiking (MAS) neuron is constructed by using discrete memristors as the autapse of neurons, combined with a discrete neuron mathematical model. Nonlinear theory plays an important role in the analysis of neural dynamics. Therefore, nonlinear numerical analysis methods, including time-domain waveforms trajectories in phase space and bifurcation behaviors, are used to analyze the basic dynamical behavior and complex multistability of MAS neuron. The complex dynamical behavior of the neuron mainly depends on the variation of the memristor parameters, and setting different parameters, the MAS neuron shows bi-stability and extreme multi-stability, which fully verifies the bionic behavior generated by the constructed MAS neuron. The implementation of MAS neuron is an important prerequisite for its application in engineering. Finally, a DSP experimental platform is built to successfully capture the attractor generated by the MAS neuron. This work provides new ideas for the construction and application of discrete neurons.

BACKGROUND Discharge instructions are essential for patients after hospital care but are time-consuming to write. With the rise of large language models (LLMs), there is a strong potential to automate this process. This study explores the use of open-source LLMs for generating discharge instructions. OBJECTIVE We investigated whether a Mistral model can reliably generate patient-oriented discharge instructions. Two distinct instruction-tuning paradigms were compared, each using a different mechanism for embedding guidance during fine-tuning. METHODS In our experiment, we applied Mistral-NeMo-Instruct, an LLM, in combination with 2 distinct instruction strategies for fine-tuning. The first were detailed instructions tailored to the task of discharge instruction generation. The second was a basic instruction with minimal guidance and no task-specific detail. The independent variable in this study is the instruction strategy (detailed vs generic), while the dependent variables are the evaluation scores of the generated discharge instructions. The generated discharge instructions were evaluated against 3621 ground-truth references. We used Bilingual Evaluation Understudy (BLEU-1) to BLEU-4, Recall-Oriented Understudy for Gisting Evaluation (ROUGE-1, ROUGE-2, and Recall-Oriented Understudy for Gisting Evaluation—Longest Common Subsequence), SentenceTransformer similarity, and Bidirectional Encoder Representations From Transformers Score as evaluation metrics to assess the quality of the generated outputs in comparison to the corresponding ground-truth instructions for the same discharge summaries. RESULTS The detailed instruction model demonstrated superior performance across all automated evaluation metrics compared with the generic instruction model. Bidirectional Encoder Representations From Transformers Score increased from 78.92% to 87.05%, while structural alignment measured by Recall-Oriented Understudy for Gisting Evaluation—Longest Common Subsequence improved from 8.59% to 26.52%. N-gram precision (BLEU-4) increased from 0.81% to 21.24%, and the Metric for Evaluation of Translation With Explicit Ordering scores rose from 15.33% to 18.47%. Additional metrics showed consistent gains: ROUGE-1 improved from 16.59% to 42.72%, and ROUGE-2 increased from 1.97% to 45.84%. All improvements were statistically significant (P<.001), indicating that detailed, task-specific instruction design substantially enhances model performance. CONCLUSIONS The use of detailed, task-specific instruction strategies significantly enhances the effectiveness of open-source LLMs in generating discharge instructions. These findings indicate that carefully designed instructions during fine-tuning substantially improve model performance.

Ensuring rapid and efficient evacuation in high-density environments, such as stadiums, is critical for public safety during fire emergencies. Traditional fire alarm systems rely on reactive detection mechanisms, often resulting in delayed response times, increased panic, and overcrowding. This study introduces an AI-driven predictive fire alarm and evacuation model that leverages machine learning algorithms and real-time environmental sensor data to anticipate fire hazards before ignition, improving emergency response efficiency. To detect early fire risk indicators, the system processes data from 62,630 sensor measurements across 15 ecological parameters, including temperature, humidity, total volatile organic compounds (TVOC), CO₂ levels, and particulate matter. A comparative analysis of six machine learning models—Logistic Regression, Support Vector Machines (SVM), Random Forest, and proposed EvacuNet—demonstrates that EvacuNet outperforms all other models, achieving an accuracy of 99.99%, precision of 1.00, recall of 1.00, and an AUC-ROC score close to 1.00. The predictive alarm system significantly reduces false alarm rates and enhances fire detection speed, allowing emergency responders to take preemptive action. Moreover, integrating AI-driven evacuation optimization minimizes bottlenecks and congestion, reduces evacuation times, and improves structured crowd movement. These findings underscore the necessity of intelligent fire detection systems in high-occupancy venues, demonstrating that AI-based predictive modeling can drastically improve fire response and evacuation efficiency. Future research should focus on integrating IoT-enabled emergency navigation, reinforcement learning algorithms, and real-time crowd management systems to further enhance predictive accuracy and minimize casualties. By adopting such advanced technologies, large-scale venues can significantly improve emergency preparedness, reduce evacuation delays, and enhance public safety.

In the field of civil engineering education, accurately evaluating the effectiveness of budget courses is crucial. However, traditional methods of evaluation tend to be cumbersome and subjective. In recent years, machine learning technology has demonstrated immense potential in educational evaluation. Nevertheless, in practical application, the machine learning-based evaluation model for civil engineering budget courses faces the predicament of inadequate evaluation accuracy. To solve this problem, the squirrel search algorithm technology was used to establish support vector machine parameters and create optimization algorithms. The performance of the proposed optimization algorithm was tested, and the results showed that the accuracy of the proposed algorithm was 0.927, which was better than similar prediction algorithms. Then, the empirical analysis of the proposed civil engineering budget course evaluation model showed that student satisfaction and student examination scores had increased to 92 and 94 points, respectively. The above results reveal that the proposed optimization algorithm and course evaluation model have good performance. Therefore, the implementation of the proposed curriculum evaluation method can significantly improve the learning efficiency of students and the teaching quality of civil engineering budgeting methods courses.

It is well known that the permanent magnet synchronous motor (PMSM) exhibits chaotic characteristics when its parameters fall within a certain range, which can lead to system instability. This article proposes an adaptive control strategy for achieving the fixed-time chaotic stabilization of PMSM, even in the presence of unknown parameters and perturbations. The developed controller is synthesized by combining a parametric adaptive mechanism with a fixed-time control technique. The stability analysis demonstrates that the system states under the developed controller can converge to small neighborhoods around the equilibrium point within a fixed time. Thanks to the adoption of the parametric adaptive mechanism, the developed controller is not only insensitive to unknown parameters but also robust against perturbations. Finally, simulated studies are conducted to verify and emphasize the effectiveness of the developed control strategy.

Healthcare waste management has been an extensively attractive topic recently since it is one of the key concerns regarding both environment and public health, predominantly in developing nations. The optimization of the treatment procedure for healthcare waste is indeed a complex “multi-criteria decision-making (MCDM)” problem that involves contradictory and interweaved critical criteria. To successfully handle this issue, this study extends the original method, named the “double normalization-based multi-aggregation (DNMA)” approach, with “interval-valued intuitionistic fuzzy sets (IVIFSs)” for decision-making problems taking criteria in terms of benefit or cost types. This method involves two target-based normalizations and three subordinate utility models. To estimate the criteria weights, we propose a new parametric divergence measure and discuss the feasibility of the developed divergence measure based on existing divergence measures for IVIFSs. Further, the developed framework is implemented to elucidate the “healthcare waste treatment (HCWT)” problem. The comparative and sensitivity analyses of the outcomes indicate that the proposed approach efficiently tackles the problem of HCWT selection. The outcomes show that steam sterilization (0.462) is the optimal one for HCWT. The prioritization options, obtained by presented approach, are dependable and suitable, which are steam sterilization ≻ microwave ≻ incineration ≻ landfilling.