Using machine learning or deep learning models in a hospital setting to detect inappropriate prescriptions: a...

Area of Science:

• Pharmacy
• Artificial Intelligence
• Health Informatics

Background:

• Artificial intelligence (AI) is increasingly relevant to hospital pharmacy due to vast health data availability.
• AI models hold the potential to significantly impact current pharmacy practices and medication management.

Purpose of the Study:

• To systematically review the current state of machine learning and deep learning models for detecting inappropriate hospital medication orders.

Main Methods:

• A systematic review following PRISMA guidelines was conducted.
• Searched MEDLINE and Embase databases up to May 2023 for studies on AI models for hospital pharmacists.
• Assessed risk of bias using the Prediction model Risk Of Bias ASsessment Tool (PROBAST).

Main Results:

• 13 articles were selected; 12 had a high risk of bias.
• Most studies (11) were published between 2020-2023, primarily in North America and Asia.
• AI models, mainly supervised learning, analyzed medication orders to detect inappropriate prescriptions, including antibiotic resistance and dosage errors.

Conclusions:

• Few original studies currently report AI tools for hospital clinical pharmacy.
• Existing research, though preliminary, demonstrates the potential value of AI in clinical hospital pharmacy.
• Further research is needed to validate and implement AI tools effectively in medication order review.