Peer-Reviewed Publication
JAMIA Open2026;9(1):ooag020.February 1, 2026Journal Article
Interpretability of an FDA-authorized AI/ML sepsis diagnostic tool improved by SHAP values.
Gregory L Watson1, Grace Staples1, Robin Carver1, Akhil Bhargava1, Carlos López-Espina1, Lee Schmalz1, Farhan Ali2, Peter S Antkowiak3, Saleem Azad4, Ramona Berghea5, Lavneet Chawla6, Matthew Crisp4, Alon Dagan3, Francisco Davila5, Hugo Davila5, Carmen DeMarco5, Amanda Doodlesack3, Aimee Espinosa5, Neil S Evans7, Clinton Ezekiel4, Andrew Friederich6, Falgun Gosai6, Alexandra Halalau5, Karthik Iyer4, Max S Kravitz3,8, Niko Kurtzman9, John H Lee3, Nicholas Maddens5, Roneil Malkani10, Stockton Mayer10, Vikram Oke4, Ashok V Palagiri11, Roshni Patel10, Lekshminarayan Raghavakurup2, Samuel Raouf2, Eric Reseland3, Farid Sadaka4, Deesha Sarma3, Scott Smith5, Tatyana Shvilkina3, Matthew D Sims5, Sahib Singh2, Bryan A Stenson3, Anwaruddin Syed6, Muleta Tafa2, Kurian Thomas10, Sihai Dave Zhao12, Ruoqing Zhu12, Rashid Bashir13, Bobby Reddy1, Nathan I Shapiro1,3,8
1Prenosis, Inc., Chicago, IL, 60616, United States.
2Lifebridge Sinai Hospital, Baltimore, MD, 21215, United States.
3Department of Emergency Medicine, Beth Israel Deaconess Medical Center, Boston, MA, 02215, United States.
4Mercy Hospital Jefferson, Festus, MO, 63028, United States.
5Corewell Health William Beaumont University Hospital, Royal Oak, MI, 48073, United States.
6OSF Saint Francis Medical Center, Peoria, IL, 61637, United States.
7Davis School of Medicine, University of California, Sacramento, CA, 95817, United States.
8Harvard Medical School, Boston, MA, 02115, United States.
9Emory School of Medicine, Atlanta, GA, 30322, United States.
10Jesse Brown VA Medical Center, Chicago, IL, 60612, United States.
11Mercy Hospital, St. Louis, MO, 63141, United States.
12Department of Statistics, University of Illinois at Urbana-Champaign, Urbana, IL, 61820, United States.
13Department of Bioengineering, University of Illinois at Urbana-Champaign, Urbana, IL, 61801, United States.
Abstract
OBJECTIVES: To assess the interpretability and acceptance of Shapley values for making artificial intelligence/machine learning (AI/ML) tools more transparent, interpretable, and useful to clinicians.
MATERIALS AND METHODS: Structured assessments were conducted with 30 clinicians (15 providers; 15 nurses; 8 assessments per clinician) to evaluate their ability to understand interventional Shapley…
Create a free account to keep reading
Free members get 10 full research views every month across publications, clinical trials, FDA clearances, adverse events, and NIH grants. No credit card required.
Want unlimited research access? See Pro plans