Back to Resources

Towards Fairness and Interpretability

John Sartori, PhD

John Sartori, PhD

Co-founder, CTO

Dr. Sartori is an Associate Professor in Electrical and Computer Engineering whose research expertise spans multiple fields.

This blog article summarizes a study co-authored by Q-rounds cofounder John Sartori, Ph.D., and a team of researchers.

Machine learning is a game changer in healthcare, enabling providers to make more accurate predictions and informed decisions, which can lead to improved patient outcomes. Its prediction of acute coronary syndrome (ACS) positive patients is becoming increasingly common, but there’s little to no research in terms of assessing fairness and interpretability of these methods. 

Therefore, various methodologies were explored to reduce fairness-related disparities in ACS predictions for different demographic sections of the population. The existing interpretability technique called LIME was also used to understand predictions made using fairness methodologies.

Enhancing transparency through interpretability

Transparency is vital in the pursuit of patient-centric care. Patients need to understand the decisions being made about their health, but machine learning models can seem empty, making it challenging for patients to trust them. Techniques like LIME can make machine learning decisions more interpretable, fostering trust between patients and their healthcare providers.

Rounding it up

The methodologies discussed have the potential to reduce treatment disparities, generalize well to different healthcare settings in the United States and reduce overall false negative rates.

Machine learning can provide personalized insights and recommendations, helping tailor medical decisions and treatments to the individual needs and preferences of each patient. It can also address disparities by ensuring all patients, regardless of their demographics, receive equitable healthcare predictions and interventions. 

Q-rounds is founded on the same desire to ensure patients, families and care teams are informed and engaged, regardless of demographics. It lets families attend rounds virtually, removing distance or transportation access as a barrier to being present, in addition to allowing families or providers to request interpreter services.

Interested in learning more? Read the full abstract or see how Q-rounds can help.