AI-enabled palliative care : from algorithms to clinical deployment
Abstract/Contents
- Abstract
- Healthcare is one of the most promising application areas for Artificial Intelligence (AI) to have a positive impact on society. There has been impressive progress in predictive modeling with health data in recent literature, even matching or exceeding expert-human level performance on a variety of tasks. Yet, translating these machine learning advances into improved patient care has proven to be particularly challenging. While developing accurate and well calibrated models (i.e. the machine learning problem) is necessary to make AI-enabled healthcare applications even possible, a careful understanding and analysis of the healthcare problem is just as essential for bridging the gap between accurate predictions and improved clinical care for the patient. Acknowledging and addressing both these problems is crucial for a successful AI clinical deployment. In this work, we consider the healthcare problem of improving access to palliative care for hospitalized patients. We frame it as a machine learning problem and validate that the framing is indeed appropriate for the healthcare problem at hand by conducting a prospective analysis study involving palliative care specialists. Our technical contributions include a novel survival loss (SurvivalCRPS), evaluation metric (SurvivalAUPRC), a gradient boosting algorithm for probabilistic prediction (NGBoost), among others. We perform a cost-benefit analysis and study the impact of various factors affecting care delivery to inform the design of a clinical workflow to increase access to palliative care services of hospitalized patients. We report on our experiences in operationalizing this workflow, powered by the above algorithmic advances, at the General Medicine service line of Stanford Hospital.
Description
Type of resource | text |
---|---|
Form | electronic resource; remote; computer; online resource |
Extent | 1 online resource. |
Place | California |
Place | [Stanford, California] |
Publisher | [Stanford University] |
Copyright date | 2022; ©2022 |
Publication date | 2022; 2022 |
Issuance | monographic |
Language | English |
Creators/Contributors
Author | Avati, Anand Vishweswaran |
---|---|
Degree supervisor | Ré, Christopher |
Thesis advisor | Ré, Christopher |
Thesis advisor | Ma, Tengyu |
Degree committee member | Ma, Tengyu |
Associated with | Stanford University, Computer Science Department |
Subjects
Genre | Theses |
---|---|
Genre | Text |
Bibliographic information
Statement of responsibility | Anand Avati. |
---|---|
Note | Submitted to the Computer Science Department. |
Thesis | Thesis Ph.D. Stanford University 2022. |
Location | https://purl.stanford.edu/pc436ft1327 |
Access conditions
- Copyright
- © 2022 by Anand Vishweswaran Avati
Also listed in
Loading usage metrics...