Hsien-Wei Ting’s research while affiliated with Taipei City Hospital and other places

Introduction Since the late 1990s, misuse of prescription sedative-hypnotics like benzodiazepines (BZDs) and non-benzodiazepines (non-BZDs) has surged, becoming a significant public health issue. To address this, a community health program was evaluated for its effectiveness in raising awareness, providing preventive education, and implementing interventions. The study assessed a comprehensive health promotion initiative involving healthcare providers, trained community members, and volunteers to encourage responsible use of sedative-hypnotics among diverse populations. Methods The study employed Community Alignment Theory to emphasize the significance of cohesive collaboration among community members, organizations, and institutions, aiming to achieve sustainable and impactful outcomes. Eight experienced institutions across various regions, proficient in implementing individual and population health education programs, were selected with support from academic universities and non-governmental organizations (NGOs). These entities actively participated in developing and executing comprehensive programs. To assess awareness regarding the accurate use of sedative-hypnotics, five-item questionnaires were developed. These questionnaires exhibited a high content validity index of over .85, determined by eight experts, and demonstrated strong internal consistency reliability of .89. Results Eight institutions across diverse regions collaborated with 161 stakeholders and enrolled 875 participants, comprising 447 males (51.1%) with an average age of 58.3 ± 17.6 years. The current prevalence of insomnia, occurring at least three days a week, stood at 15.2% (n=133), with over half of the participants (n=466, 53.3%) having sought medical assistance for insomnia. Following the execution of 646 educational programs conducted in 275 locations and 13,348 participants attended, there was a significant improvement in awareness regarding insomnia and the appropriate utilization of sedative-hypnotics (p<.001). This enhancement encompassed various aspects such as daytime naps, sleep hygiene, prescription adherence, and awareness of hypnotics' adverse effects. Conclusion The utilization of the community alignment theory serves as a fundamental framework for steering educational initiatives within communities, demonstrating advantageous impacts on promoting the precise utilization of sedative-hypnotics. Support (if any) The Food and Drug Administration, Ministry of Health and Welfare, Taiwan, supported the study (Project number MOHW112-FDA-D-114-000641).

Background Large-scale burn disasters can produce casualties that threaten medical care systems. This study proposes a new approach for developing hospital readiness and preparedness plan for these challenging beyond-surge-capacity events. Methods The Formosa Fun Coast Dust Explosion (FFCDE) was studied. Data collection consisted of in-depth interviews with clinicians from four initial receiving hospitals and their relevant hospital records. A detailed timeline of patient flow and emergency department (ED) workload changes of individual hospitals were examined to build the EDs' overload patterns. Data analysis of the multiple hospitals' responses involved chronological process-tracing analysis, synthesis, and comparison analysis in developing an integrated adaptations framework. Results A four-level ED overload pattern was constructed. It provided a synthesis of specifics on patient load changes and the process by which hospitals' surge capacity was overwhelmed over time. Correspondingly, an integrated 19 adaptations framework presenting holistic interrelations between adaptations was developed. Hospitals can utilize the overload patterns and overload metrics to design new scenarios with diverse demands for surge capacity. The framework can serve as an auxiliary tool for directive planning and cross-check to address the insufficiencies of preparedness plans. Conclusions The study examined a wide-range spectrum of emergency care responses to the FFCDE. It indicated that solely depending on policies or guidelines for preparedness plans did not contribute real readiness to MCIs. Hospitals can use the study's findings and proposal to rethink preparedness planning for the future beyond surge capacity events.

Background: Issuing of correct prescriptions is a foundation of patient safety. Medication errors represent one of the most important problems in health care, with 'look-alike and sound-alike' (LASA) being the lead error. Existing solutions to prevent LASA still have their limitations. Deep learning techniques have revolutionized identification classifiers in many fields. In search of better image-based solutions for blister package identification problem, this study using a baseline deep learning drug identification (DLDI) aims to understand how identification confusion of look-alike images by human occurs through the cognitive counterpart of deep learning solutions and thereof to suggest further solutions to approach them. Methods: We collected images of 250 types of blister-packaged drug from the Out-Patient Department (OPD) of a medical center for identification. The deep learning framework of You Only Look Once (YOLO) was adopted for implementation of the proposed deep learning. The commonly-used F1 score, defined by precision and recall for large numbers of identification tests, was used as the performance criterion. This study trained and compared the proposed models based on images of either the front-side or back-side of blister-packaged drugs. Results: Our results showed that the total training time for the front-side model and back-side model was 5 h 34 min and 7 h 42 min, respectively. The F1 score of the back-side model (95.99%) was better than that of the front-side model (93.72%). Conclusions: In conclusion, this study constructed a deep learning-based model for blister-packaged drug identification, with an accuracy greater than 90%. This model outperformed identification using conventional computer vision solutions, and could assist pharmacists in identifying drugs while preventing medication errors caused by look-alike blister packages. By integration into existing prescription systems in hospitals, the results of this study indicated that using this model, drugs dispensed could be verified in order to achieve automated prescription and dispensing.

Background: Issuing of correct prescriptions is a foundation of patient safety. Medication errors represent one of the most important problems in health care, with ‘look-alike and sound-alike’ (LASA) being the lead error. Existing solutions to prevent LASA still have their limitations. Deep learning techniques have revolutionized identification classifiers in many fields. In search of better image-based solutions for blister package identification problem, this study using a baseline deep learning drug identification (DLDI) aims to understand how identification confusion of look-alike images by human occurs through the cognitive counterpart of deep learning solutions and thereof to suggest further solutions to approach them. Methods: We collected images of 250 types of blister-packaged drug from the Out-Patient Department (OPD) of a medical center for identification. The deep learning framework of You Only Look Once (YOLO) was adopted for implementation of the proposed deep learning. The commonly-used F1 score, defined by precision and recall for large numbers of identification tests, was used as the performance criterion. This study trained and compared the proposed models based on images of either the front-side or back-side of blister-packaged drugs. Results: Our results showed that the total training time for the front-side model and back-side model was 5 hours 34 minutes and 7 hours 42 minutes, respectively. The F1 score of the back-side model (95.99%) was better than that of the front-side model (93.72%). Conclusions: In conclusion, this study constructed a deep learning-based model for blister-packaged drug identification, with an accuracy greater than 90%. This model outperformed identification using conventional computer vision solutions, and could assist pharmacists in identifying drugs while preventing medication errors caused by look-alike blister packages. By integration into existing prescription systems in hospitals, the results of this study indicated that using this model, drugs dispensed could be verified in order to achieve automated prescription and dispensing.

Background: Issuing of correct prescriptions is a foundation of patient safety. Medication errors represent one of the most important problems in health care, with ‘look-alike and sound-alike’ (LASA) being the lead error. Existing solutions to prevent LASA still have their limitations. Deep learning techniques have revolutionized identification classifiers in many fields. Aiming to identify blister packages, this study employed a visual-based identification solution, called the deep learning drug identification (DLDI) model, which reduced medication identification errors caused by look-alike drugs. Methods: We collected images of 250 types of blister-packaged drug from the Out-Patient Department (OPD) of a medical center for identification. The deep learning framework of You Only Look Once (YOLO) was adopted for implementation of the proposed deep learning. The commonly-used F1 score, defined by precision and recall for large numbers of identification tests, was used as the performance criterion. This study trained and compared the proposed models based on images of either the front-side or back-side of blister-packaged drugs. Results: Our results showed that the total training time for the front-side model and back-side model was 5 hours 34 minutes and 7 hours 42 minutes, respectively. The F1 score of the back-side model (95.99%) was better than that of the front-side model (93.72%). Conclusions: In conclusion, this study constructed a deep learning-based model for blister-packaged drug identification, with an accuracy greater than 90%. This model outperformed identification using conventional computer vision solutions, and could assist pharmacists in identifying drugs while preventing medication errors caused by look-alike blister packages. By integration into existing prescription systems in hospitals, the results of this study indicated that using this model, drugs dispensed could be verified in order to achieve automated prescription and dispensing.

Background: Issuing of correct prescriptions is a foundation of patient safety. Medication errors represent one of the most important problems in health care, with ‘look-alike and sound-alike’ (LASA) being the lead error. Existing solutions to prevent LASA still have their limitations. Deep learning techniques have revolutionized identification classifiers in many fields. In search of better image-based solutions for blister package identification problem, this study using a baseline deep learning drug identification (DLDI) aims to understand how identification confusion of look-alike images by human occurs through the cognitive counterpart of deep learning solutions and thereof to suggest further solutions to approach them. Methods: We collected images of 250 types of blister-packaged drug from the Out-Patient Department (OPD) of a medical center for identification. The deep learning framework of You Only Look Once (YOLO) was adopted for implementation of the proposed deep learning. The commonly-used F1 score, defined by precision and recall for large numbers of identification tests, was used as the performance criterion. This study trained and compared the proposed models based on images of either the front-side or back-side of blister-packaged drugs. Results: Our results showed that the total training time for the front-side model and back-side model was 5 hours 34 minutes and 7 hours 42 minutes, respectively. The F1 score of the back-side model (95.99%) was better than that of the front-side model (93.72%). Conclusions: In conclusion, this study constructed a deep learning-based model for blister-packaged drug identification, with an accuracy greater than 90%. This model outperformed identification using conventional computer vision solutions, and could assist pharmacists in identifying drugs while preventing medication errors caused by look-alike blister packages. By integration into existing prescription systems in hospitals, the results of this study indicated that using this model, drugs dispensed could be verified in order to achieve automated prescription and dispensing. Key words: deep learning; drug identification; look-alike and sound-alike (lasa); medication error; patient safety.

Objective The study provides a comprehensive insight into how an initial receiving hospital without adequate capacity adapted to coping with a mass casualty incident after the Formosa Fun Coast Dust Explosion (FFCDE). Methods Data collection was via in-depth interviews with 11 key participants. This was combined with information from medical records of FFCDE patients and admission logs from the emergency department (ED) to build a detailed timeline of patients flow and ED workload changes. Process tracing analysis focused on how the ED and other units adapted to coping with the difficulties created by the patient surge. Results The hospital treated 30 victims with 36.3% average total body surface area burn for over 5 hours alongside 35 non-FFCDE patients. Overwhelming demand resulted in the saturation of ED space and intensive care unit beds, exhaustion of critical materials, and near-saturation of clinicians. The hospital reconfigured human and physical resources differently from conventional drills. Graphical timelines illustrate anticipatory or reactive adaptations. The hospital’s ability to adapt was based on anticipation during uncertainty and coordination across roles and units to keep pace with varying demands. Conclusion Adapting to beyond-surge capacity incident is essential to effective disaster response. Building organizational support for effective adaptation is critical for disaster planning.