Summary of A Kalman Filter Based Framework For Monitoring the Performance Of In-hospital Mortality Prediction Models Over Time, by Jiacheng Liu et al.
A Kalman Filter Based Framework for Monitoring the Performance of In-Hospital Mortality Prediction Models Over Time
by Jiacheng Liu, Lisa Kirkland, Jaideep Srivastava
First submitted to arxiv on: 9 Feb 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: None
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary The proposed study aims to address the issue of evaluating binary classifiers in a long-term time period, where the size and distribution of incoming patients can vary. Unlike traditional machine learning studies, there is little control over these factors in real-world scenarios. To overcome this challenge, the authors propose adjusting performance metrics such as AUCROC and AUCPR for sample size and class distribution. This allows for a fair comparison between different time periods. The study also proposes a Kalman filter-based framework to estimate the mean of performance metrics and understand changes over time. The efficacy of this method is demonstrated on both synthetic and real-world datasets, including a 2-day in-hospital mortality prediction model for COVID-19 patients. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary In this study, researchers want to find a way to fairly compare how well a computer program predicts whether someone will get sick or not over time. Right now, it’s hard to do this because the number of people being tested and the types of tests being used can change over time. To solve this problem, the authors suggest adjusting how we measure how good the program is by taking into account these changes. They also propose a special way to use math to better understand how well the program is working over time. The study uses fake data and real data from COVID-19 patients to test their idea. |
Keywords
* Artificial intelligence * Machine learning
