Amir Kamel Rahimi: transforming clinical decision-making by leveraging AI and Machine Learning Models
Published 24 May 2024
AI researcher | Data analyst | PhD candidate
Linkedin: Amir Kamel Rahimi
The emergence of Artificial Intelligence (AI) in healthcare has promised to improve clinical outcomes. With the rapid surge in AI advancements recently, particularly with the introduction of Large Language Models (LLMs) like Open AI GPT-4 publicly released in November 2022, the transformative potential of these technological advancements could reshape the landscape of healthcare systems by creating proactive models of care.
There are currently efforts underway to capitalise on the power of the data collected in Electronic Medical Records (EMRs) by creating real-time analytics to improve decision-making in hospitals. Our previous work explored the potential applications of clinical AI for patients with diabetes, a chronic disease that affects millions of lives each year. The lack of real-world implementation of clinical AI analytics in hospitals for chronic diseases such as diabetes is found a significant issue that can greatly impact our understanding of how these tools should be effectively integrated and utilised.
Ensuring the effective performance of AI analytics is crucial in both their development and integration phases to avoid negative clinical outcomes. A thorough understanding of existing disease definition standards is essential for evaluating the efficacy of AI analytics. Moreover, the incorporation of explainable AI techniques is pivotal in uncovering how AI makes decisions in clinical settings, increasing end-user trust and facilitating better assessment of AI efficacy based on the existing clinical knowledge. In our previous work, we demonstrated that various definitions used to define acute kidney injury can significantly impact the efficacy of AI models. Furthermore, we employed Explainable AI to gain deeper insights into how AI makes decisions for predicting acute kidney injury.
Despite the increasing interest in the application of AI innovations in hospitals, the knowledge of healthcare organisations for implementing such AI analytics is still immature. Successful AI implementation in hospitals requires a shift in conventional resource management to support a new AI implementation and maintenance strategy. It is crucial to form a diverse team of experts, enhance the existing processes, strive for better data quality and privacy and strengthen the technological infrastructure to effectively adopt AI technologies in hospital settings.