Personalized Uncertainty Quantification in Operating Room (PUQOR): Optimizing Surgical Time Estimation with Conformal Prediction

Presentation Time: 03:30 PM - 03:45 PM

Abstract Keywords: Clinical Decision Support, Healthcare Economics/Cost of Care, Machine Learning, Surgery, Informatics Implementation
Primary Track: Applications
Programmatic Theme: Clinical Research Informatics

We developed a machine learning algorithm incorporating personalized uncertainty quantification through Conformal Prediction to estimate surgical duration, outperforming the existing Epic estimates and was nearly equivalent to surgeons’ estimates. We demonstrated the model’s superiority in various surgery subgroups, especially those historically misjudged by Epic, and showcased promising improvements. Our model substantiated its potential in supporting surgeons for decision-making in operating room management, and proposed a new path of optimizing personalized surgery planning in precision medicine.

Speaker(s):
Sylvia Cheng, Ph.D. Candidate
University of California, Berkeley

Author(s):
Vincent Liu, MD, MSc - Kaiser Permanente; Bradley Cohn, MD - Kaiser Permanente; Patricia Kipnis, PhD - Kaiser Permanente; Alejandro Schuler, PhD - University of California, Berkeley; Brian Lawson, PhD - Kaiser Permanente;

Predicting the total cost of care in oncology using modifiable covariates

Presentation Time: 03:45 PM - 04:00 PM

Abstract Keywords: Healthcare Economics/Cost of Care, Administrative Systems, Behavioral Change
Primary Track: Applications
Programmatic Theme: Clinical Informatics

A linear regression model has been constructed to predict the total cost of care (TCOC) for multiple myeloma, utilizing baseline episode data from The Enhancing Oncology Model. The model incorporates modifiable and actionable covariates, with a particular focus on drug usage. This model can be utilized to estimate the TCOC of oncology care.

Speaker(s):
Ping Ye, PhD
The US Oncology Network, McKesson

Author(s):

Making the Case for Patient-Centered Clinical Decision Support: Exploring Approaches for Projecting and Demonstrating Return on Investment

Presentation Time: 04:00 PM - 04:15 PM

Abstract Keywords: Healthcare Economics/Cost of Care, Clinical Decision Support, Evaluation
Primary Track: Applications
Programmatic Theme: Clinical Informatics

Patient-centered clinical decision support (PC CDS) advances the translation of patient-centered outcomes research findings into practice and integrates patient-centered data to support healthcare decision-making. Lack of information about the return on investment (ROI) for PC CDS hinders uptake. The Clinical Decision Support Innovation Collaborative conducted a landscape assessment to document approaches for determining ROI, identifying factors that in combination provide a holistic approach that considers financial viability with the needs of patients and care teams.

Speaker(s):
Priyanka Desai
NORC at the University of Chicago

Author(s):
Priyanka Desai - NORC at the University of Chicago; David Lobach, MD - VP Health Informatics Research, Elimu Informatics; Krysta Heaney-Huls, MPH - NORC @ the University of Chicago; Sofia Ryan, MSPH - NORC at the University of Chicago; Andrew Chiao, MPH - NORC at the University of Chicago; Kensaku Kawamoto, MD, PhD, MHS - University of Utah; James Swiger, MBE - AHRQ; Edwin Lomotan, MD - AHRQ; Prashila Dullabh, MD - NORC at the University of Chicago;

Design and Implementation of a Medication Co-payment Estimation Tool: A Database-driven Approach for Enhanced Medication Co-pay estimation

Presentation Time: 04:15 PM - 04:30 PM

Abstract Keywords: Clinical Decision Support, Data Mining, Information Retrieval, Data Transformation/ETL
Primary Track: Applications

This paper presents the design and implementation of a novel medication co-payment estimation system aimed at addressing the challenge of medication non-adherence due to cost barriers. The system employs a database-driven approach to accurately predict medication co-payments at the point of care. Utilizing data extracted from the University of Utah Health Enterprise Data Warehouse, including insurance coverage, prescription fills, and historical co-payment amounts, the system facilitates transparent price discussions between healthcare providers and patients. Through an iterative data processing methodology, the system manages and updates medication dispense data, incorporating factors such as insurance coverages, prior authorizations, and coupon usage to enhance prediction accuracy. The architecture enables healthcare providers to access estimations of medication co-payments. Testing results of the system indicate high accuracy levels in co-payment estimations, reflecting its potential to improve medication adherence by enabling informed prescription decisions. The findings underscore the importance of leveraging data in addressing cost-related barriers to medication adherence.

Speaker(s):
Nikolay Lukyanchikov, PhD Candidate
University of Utah

Author(s):
Kensaku Kawamoto, MD, PhD, MHS - University of Utah; Claude Nanjo, MPH - Biomedical Informatics Department, University of Utah; Joshua Choi, MD - University of Utah;

Clinical and Financial Impact of a Machine Learning Powered Screening Program for Abdominal Aortic Aneurysms

Presentation Time: 04:30 PM - 04:45 PM

Abstract Keywords: Healthcare Economics/Cost of Care, Machine Learning, Population Health, Healthcare Quality
Primary Track: Applications
Programmatic Theme: Public Health Informatics

Abdominal aortic aneurysms (AAA) are readily treatable if identified early but have a 60-70% mortality rate if they rupture. Geisinger has implemented a machine learning-powered AAA management program, in which high-risk individuals are identified and contacted for screening. This program has delivered a substantial decrease in AAA mortality and an increase in evidence-based screening ultrasound scans performed, all within a financially sustainable operating model.

Speaker(s):
Elliot Mitchell, PhD
Geisinger

Author(s):
Alexander Pretko, MS - Geisinger; James Elmore, MD - Geisinger; David Vawdrey, PhD - Geisinger; Rebecca Maff, MS; Elliot Mitchell, PhD - Geisinger;

Decision and Cost Analysis of a Machine Learning Model to Predict Unplanned Cancer Readmissions

Presentation Time: 04:45 PM - 05:00 PM

Abstract Keywords: Clinical Decision Support, Healthcare Quality, Healthcare Economics/Cost of Care, Machine Learning
Primary Track: Applications
Programmatic Theme: Clinical Informatics

This study examines the expected cost of using a machine learning (ML) model to predict unplanned cancer readmissions and guide the scheduling of 5-day follow-up calls. Results highlight potential cost savings, especially after optimizing ML sensitivity thresholds. Implementing this approach could lead to significant financial benefits, underscoring the value of ML in healthcare decision-making.

Speaker(s):
Scott Vennemeyer
University of Cincinnati, College of Medicine

Author(s):
Scott Vennemeyer - University of Cincinnati, College of Medicine; Tripura Vithala - University of Cincinnati; Daniel Schauer, MD - University of Cincinnati College of Medicine; Danny Wu, PhD - University of Cincinnati College of Medicine; Mark Eckman, MD, MS, FACMI, FACP - University of Cincinnati Medical Center;