The Center for Clinical Artificial Intelligence (CCAI) focuses on developing, implementing, and evaluating high-performance clinical decision support (CDS) tools that are powered by artificial intelligence (AI), including machine learning. AI has the potential to support, enable, and improve medical decision-making to make it faster, more accurate, and more economical. The next generation of CDS will be powered by AI-enabled predicting, monitoring, and alerting.

Using principles from biomedical informatics, the Center creates data infrastructure and repositories to aid in the development of AI algorithms and the implementation of AI-enabled CDS tools.

The goals of the Center are to 1) create AI algorithms for CDS tools to address unmet clinical needs, 2) demonstrate internal and external validity of AI algorithms, 3) ensure algorithmic fairness, 4) obtain approval from the U.S. Food and Drug Administration (FDA), and 5) monitor algorithmic robustness after deployment. The Center is advancing both the science and the engineering of AI-enabled CDS to improve health and healthcare.

• 1) Create AI algorithms for CDS tools to address unmet clinical needs. CCAI identifies unmet clinical needs that can be addressed by AI using pattern recognition or prediction and builds AI-enabled CDS for use at the point of care.

• 2) Demonstrate internal and external validity of AI algorithms. CCAI evaluates AI algorithms in a variety of ways, including statistical validity, clinical utility, and economic utility. Statistical validity evaluates performance on discrimination and calibration metrics. Clinical utility assesses performance regarding clinical care and patient outcomes. Economic utility evaluates the impact on healthcare services and cost savings.

• 3) Ensure algorithmic fairness. Ensuring fairness is vital as algorithms increasingly play a critical role in clinical decision-making and the potential for harm grows. CCAI ensures that an algorithm is fair and non-discriminatory when it comes to sensitive attributes like age, gender, race, and socioeconomic status.

• 4) Obtain approval from the FDA. The evolving FDA regulatory framework for Software as a Medical Device (SaMD) is expediting the approval of AI algorithms for clinical use. CCAI is developing experience in applying for and receiving FDA certification, which is essential for real-world deployment.

• 5) Monitor algorithmic robustness after deployment. Clinically deployed algorithms must be monitored for performance degradation over time, across geographical locations, and across populations with varying disease severity or prevalence of the outcome. CCAI is developing methods for automatically assessing algorithmic robustness.

CCAI is working on CDS that use human-in-the-loop AI rather than autonomous AI that does not have a human in the loop. These algorithms are focused on monitoring, risk assessment, prognosis, and workflow efficiency in clinical care. The Center brings together clinical experts, AI experts, and clinical informatics experts from the University of Pittsburgh, UPMC, and Carnegie Mellon University.

Current projects include:

• Learning Electronic Medical Record (LEMR) System utilizing AI to enable EMR systems to deliver the right data of the right patient at the right time in the intensive care unit (ICU) (https://www.thevislab.com/lab/doku.php?id=lemr)
• Outlier-based clinical monitoring and alerting leveraging AI to identify outliers and potential errors in the ICU (https://people.cs.pitt.edu/~milos/outlier_alerting/)
• Realtime Evaluation of ADverse Events with AI during surgery (READE.ai) (https://www.reade.ai/)
• Machine-Learning Prediction and Reducing Overdoses with EHR Nudges (mPROVEN) (https://www.cp3.pitt.edu/mproven)
• Post-Induction Hypotension Modeling in Anesthesia
• Risk of Bleeding with Direct Oral Anticoagulants