The Impact of Implementing Artificial Intelligence Systems on Pressure Injury Indicators.

The project aims to analyze the impact of implementing Materials and methods: AI system on Hospital-acquired pressure injuries indicators. The AI Pressure injury prediction system was introduced in all adult inpatients at the medical center in central Taiwan in 2022. Hospital CCH and granted a waiver of informed consent (protocol code No: 200721). As the dataset, we retrospectively reviewed the EHR of 95,027 adult patients (aged ≥ 20 years) and without including the PI was determined before admission who admission nursing assessment from Aug 2018 to Jun 2020. After evaluating different subsets of features, we determined that selecting 21 features(for example: Braden Scale, BMI, ICD, Barthel Index Total Number…) provided the best balance between predictive power and computational efficiency. The AI model's accuracy was maintained between 83-85% and the Changes in pressure injuries incidence rates were tracked over three years. Data Collection the Pressure injuries were recorded daily through nursing information systems. The Business Intelligence (BI) tools were used to analyze trends in pressure injuries incidence. The incidence rate of pressure injuries improved from 0.33% in 2022 to 0.21% in 2024. The proposed feature aggregation was critical in predicting risk of PI because feature aggregation could reduce false prediction in positive cases and make the higher predicting ability. Hence, our method may assist clinical physicians and nurses provide valuable information and understanding the method decision-making process. The AI for identifying high-risk cases and providing care recommendations, along with BI automated analysis, appears to have positively impacted the incidence of hospital-acquired pressure injuries.