Articles
Hadian, S.A.,
Rezayatmand, R.,
Ketabi, S.,
Pourghaderi, A.R.,
Shaarbafchizadeh, N. Publication Date: 2026
Bmc Health Services Research (14726963)26(1)
Background: In the context of healthcare systems, evaluating hospital performance usign appropriate quantitative models is essential. This paper introduces a novel approach for assessing hospital performance in uncertain environments. Methods: In this study, we applied an additive star network data envelopment analysis (star-NDEA) model to evaluate the interdepartmental efficiency of 26 hospitals whithin Isfahan University of Medical Sciences (MUI) using visual analytis tools. The analysis, based on data from 2019 to 2022, utilized 44 indicators (12 inputs, 17 outputs, 14 links), which were carefully selected as model variables. The relevance and suitability of these indicators were assessed through a systematic two-stage process, Results: This study indicated that the additive star-NDEA model identifies the specific sources of inefficiency and more accurately captures the inefficiencies of hospitals, reflecting the internal parameters of departmental-level performance evaluation. Across all the studied hospitals, the logistics department consistently had the highest efficiency score (approximately 0.9) among the eight departments evaluated over the four years. The highest efficiency scores for specific departments were recorded as follows: surgical inpatient wards in 2019 (0.64), imaging department in 2020 (0.56), clinic department in 2021 (0.59) and surgical inpatient wards again in 2022 (0.63). In contrast, the departments with the lowest efficiency scores were, the internal inpatient ward in 2019 (0.43), the clinic department in 2020 (0.33), and, the imaging department in both 2021 and 2022 (around 0.40). Overall, the efficiency scores derived from additive star-NDEA model were lower than those obtained using the classical additive DEA model. Conclusion: The model presented in this study is well-suited for assessing the efficiency of hospitals and their internal departments. These findings highlight the importance of hospitals adopting advanced quantitative models to improve to enhance performance, optimize resource utilization, plan targeted interventions, improve service delivery and ultimately increase overall technical efficiency. © The Author(s) 2025.
Yousefi, A.,
Ketabi, S.,
Moreno, A.C.,
Abedi, I.,
Kim, Y.,
Gholami, S. Publication Date: 2025
Medical Physics (24734209)52(9)
Background: Radiotherapy treatment planning is a time-consuming task that requires expert and skilled manpower, particularly for weight adjustment. Valuable attempts have been made to automate the treatment planning process as well as decrease computation time in recent years. Artificial intelligence tools and a knowledge-based planning (KBP) approach have played considerable roles in this regard. However, this area also requires more precise and smart approaches. Purpose: The current study aims to advance KBP in two practical and impactful areas. First, it presents a novel approach to automate IMRT treatment planning using a mathematical optimization framework. Second, it proposes two innovative downsizing techniques designed to enhance computational efficiency and significantly reduce solving time, while evaluating their performance in terms of both treatment plan quality and time savings in an integrated manner. Methods: Two mathematical models were applied: QuadLin for treatment plan optimization and its revised model for automatically adjusting the weights of the QuadLin objective function. The study emphasizes improving computational efficiency and reducing solving time by introducing an innovative algorithm, called SVSIDB, which clusters voxels based on the dominant beamlet concept. Additionally, the hybrid ultra-heuristic ABC-K-Means technique was developed for voxel clustering. All models and techniques have been run on the data of 30 patients with head and neck cancer from a recently published real dataset, Open-KBP. Problems were solved in the CVX framework, with commercial solver Mosek, as well as programming in MATLAB. The results have been evaluated by both plan quality approach, satisfied clinical criteria, and computational efficiency, solving time reduction. Results: The weights of the QuadLin objective function were automatically adjusted using the mathematical framework. Although the Auto-Imputed weights differed significantly from the manually assigned ones, the resulting plans showed no substantial differences in terms of plan quality. Automatic treatment plans improved satisfied clinical criteria by an average of over 21%, 15%, and at least 13% compared to the predicted dose, the reference plan, and previous research, respectively. Additionally, SVSIDB presented a systematic voxel clustering method that reduces solving time by approximately 50% compared to full-data models, while maintaining treatment plan quality. SVSIDB achieved an 81.3% clinical criteria satisfaction index, which was 10% higher than that of ABC-K-Means. In terms of time-saving performance, ABC-K-Means matched the efficiency of SVSIDB. Conclusions: This research makes two remarkable contributions: (1) the development of an automatic KBP framework and (2) introducing a novel, efficient downsizing technique. The Auto-Imputed weights preserved the quality of treatment plans despite substantial differences from manually adjusted weights. SVSIDB demonstrated an average quality index improvement of 12% compared to previous studies, including those by Fountain et al. and Babier et al. Notably, the SVSIDB-QuadLin pipeline not only reduced solving time but also improved plan quality, outperforming models based on full data and representing a substantial advancement over prior research. © 2025 American Association of Physicists in Medicine.
