Development of machine-learning models using pharmacy inquiry database for predicting dose-related inquiries in a tertiary teaching hospital.

BACKGROUND: Drug-related problems (DRPs) are a significant concern in healthcare. Pharmacists play a vital role in detecting and resolving DRPs to improve patient safety. A pharmacy inquiry program was established in a tertiary teaching hospital to document inquiries about physicians' orders, aimed at preventing potential DRPs or providing medication information during order reviews. OBJECTIVE: We aimed to develop machine-learning models using a pharmacy inquiry database to predict dose-related inquiries based on prescriptions and patient information. METHODS: This retrospective study analyzed 20,393 pharmacy inquiries collected between January 2018 and February 2023. Data included prescription information (drug ingredient, dose, unit, and frequency), patient characteristics (age, sex, weight, and department), and renal function. The inquiries were categorized into two classes: dose-related inquiries (e.g., wrong dose and inappropriate regimen) and non-dose-related inquiries (e.g., inappropriate drug form and administration route). Six machine-learning models were developed: logistic regression, support vector classifier, decision tree, random forest, extreme gradient boosting, and categorical boosting. To evaluate the performance of the models, the area under the receiver operating characteristic curve and the accuracy were compared. RESULTS: The CatBoost model achieved the highest performance (sensitivity: 0.92; accuracy: 0.79). The SHapley Additive exPlanations values highlighted the importance of features in the model predictions, drug ingredients, units, and renal function, in that order. Notably, lower renal function positively contributed to the prediction of dose-related inquiries. Additionally, the subsequent feature importance among drug ingredients showed that drugs such as acetylsalicylic acid, famotidine, metformin, and spironolactone strongly influenced the prediction of dose-related inquiries. CONCLUSION: Machine-learning models that use pharmacy inquiry data can effectively predict dose-related inquiries. Further external validation and refinement of the models are required for broader applications in healthcare settings. These findings provide valuable guidance for healthcare professionals and highlight the potential of machine learning in pharmacists' decision-making.

CHDC: Common Hospital Data Connector for exchanging medical information.

With the advance of technology, smart healthcare is emerging, and the amount of medical data is exploding. Currently, these large-scale medical data is scattered and stored in a variety of institutions. The data format is also different. These make medical data exchange difficult. Although there are many attempts to integrate EMR systems into a standardized structure, a high rate of EMR adoption in Korea using a standardized model such as HL7 RIM (Reference Information Model), attempts to replace systems that are already in use have largely failed due to a variety of causes, including time, cost and implementation manpower, and so on. In this paper, we propose a Common Hospital Data Connector model that has a way to integrate medical data without replacing the systems. This model defines what data should be extracted from the system with a variety of interfaces. These data profiles are based on statutory document and are able to be accepted in other medical institution. To ensure interoperability, the document is based on the international standard of Fast Healthcare Interoperability Resources and can be created in the form of a statutory document that can be printed. We applied this model to the hospital and verified that it can be used in actual services.