Feb 26 2021
Clinical research necessitates the mining of data to gain crucial insights. Machine learning, which involves developing algorithms to identify patterns, finds it hard to do this in the case of data related to health records since information of this kind is neither static nor collected regularly.
Prof. Jeremy Weiss, an expert in machine learning and healthcare informatics. Image Credit: Carnegie Melon University Heinz College.
As part of a new study, researchers have developed a transparent, reproducible machine learning tool to allow the analysis of health information. The tool can be employed in clinical forecasting, which can estimate trends and outcomes in individual patients.
Performed by a researcher from Carnegie Mellon University (CMU), the study has been published in Proceedings of Machine Learning Research.
Temporal Learning Lite, or TL-Lite, is a visualization and forecasting tool to bridge the gap between clinical visualization and machine learning analysis. While the individual elements of this tool are well known, their integration into an interactive clinical research tool is new and useful for health professionals. With familiarization, users can conduct preliminary analyses in minutes.
Jeremy Weiss, Study Author and Assistant Professor of Health Informatics, Heinz College, Carnegie Melon University
Time is a crucial part of clinical data collected during health care delivery.
For instance, discussions of patients on rounds, where doctors visit hospital patients to observe how they are doing, involve the use of visual aids medical staff, depicting measurements of progression and recovery.
The widespread use of electronic health records has enabled considerable advances in visualizing clinical data and in clinical forecasting. However, there still remains a gap between the two.
TL-Lite starts with visualizations of data from databases and concludes with visual risk assessments of a temporal model.
Side by side, users can observe the impacts of their design choices using visual summaries at the levels of individuals and groups. This enables users to gain insights into their data more comprehensively and tune the machine learning settings for their analysis.
Weiss showed the usage mode of the tool by demonstrating the model with three electronic health records related to three health matters—predicting acute thrombocytopenia (with abnormally low platelet levels in the blood) during intensive care unit (ICU) stays for patients with sepsis, predicting survival of patients admitted to the ICU a day following admission, and predicting microvascular complications for type 2 diabetes patients.
The central goal of TL-Lite is to facilitate well-specified and well-crafted predictive forecasting, and this visualization tool is meant to ease the process. At the same time, organizing the clinical data stream into meaningful visualizations can be aided by introducing machine learning elements. These approaches are complementary, so leveraging the benefits of one where another hits roadblocks results in a better overall solution.
Jeremy Weiss, Study Author and Assistant Professor of Health Informatics, Heinz College, Carnegie Melon University
The study was supported by the CMU Center for Machine Learning and Health, Amazon Web Services, and Microsoft Azure.