Every hospital manager aims to build harmonious, mutually beneficial, and steady-state departments. Therefore, it is important to explore a hospital department development assessment model based on objective hospital data. This study aims to use a novel machine learning algorithm to identify key evaluation indexes for hospital departments, offering insights for strategic planning and resource allocation in hospital management. Data related to the development of a hospital department over the past 3 years were extracted from various hospital information systems. The resulting data set was mined using neural machine algorithms to assess the possible role of hospital departments in the development of a hospital. A questionnaire was used to consult senior experts familiar with the hospital to assess the actual work in each hospital department and the impact of each department's development on overall hospital discipline. We used the results from this questionnaire to verify the accuracy of the departmental risk scores calculated by the machine learning algorithm. Deep machine learning was performed and modeled on the hospital system training data set. The model successfully leveraged the hospital's training data set to learn, predict, and evaluate the working and development of hospital departments. A comparison of the questionnaire results with the risk ranking set from the departments machine learning algorithm using the cosine similarity algorithm and Pearson correlation analysis showed a good match. This indicates that the department development assessment model and risk score based on the objective data of hospital systems are relatively accurate and objective. This study demonstrated that our machine learning algorithm provides an accurate and objective assessment model for hospital department development. The strong alignment of the model's risk assessments with expert opinions, validated through statistical analysis, highlights its reliability and potential to guide strategic hospital management decisions.
Read full abstract