Articles & Publications
Catch up on the latest insights and updates from LKS-CHART.
Looking for more clinical data science learning? Check our Webinars, LKS-CHART Tools and Recommended Resources pages.
Featured Insights & Updates
Articles & Publications
August 07, 2020
Intensive care medicine is well positioned to harness modern advances in artificial intelligence (AI) given the abundance of data generated in the intensive care unit (ICU). This is the first in a series of articles aiming to improve clinician understanding of the development and application of AI solutions in the intensive care environment, through exploring three core themes: enabling data, AI development, and AI deployment.
Nursing resource team capacity planning using forecasting and optimization methods: A case study
November 16, 2019
Nurse absences can be filled by a Nursing Resource Team (NRT) at lower hourly cost than staffing agencies or nurses working overtime, but an NRT must be appropriately sized to minimize total cost. Using registered nurse (RN) absences, we developed a generalized additive model (GAM) to forecast the weekly frequency of absences. We found that the frequency of RN absences in a hospital can be forecasted with high accuracy and the use of forecasting and optimization can substantially reduce the cost of filling RN absences.
Incorporating domain knowledge from subject matter experts is critical for building useful models. As statistician George Box said, “All models are wrong, but some are useful.” How useful our models are depend on how well we apply domain knowledge to extract the signal from the noise...
Most patient information is stored only in the form of transcribed, dictated notes. CHARTextract is a rule-based information extraction tool with the ability to quickly and easily refine rules on-the-fly. By combining physician knowledge and keyword-based pattern matching, we can extract patient attributes (e.g., diagnosis, country of birth, procedures, etc.) from free-form text notes. With this tool, hospital leadership, clinicians and clinical researchers everywhere can automatically extract patient information from text notes and perform large-scale analyses of patient data. The tool, as well as pre-existing rulesets, are available for download.
As health care data grows exponentially, Family physicians will increasingly need to leverage advances in data acquisition, storage, and analytics to make sound clinical decisions. However, the progress in the areas of informatics and analytics training has been slow. With the current focus on competency-based medical education, Canadian family medicine (FM) residency programs must establish a minimum level of competence in informatics and analytics for graduates to succeed.
We sought to determine whether addition of a snowfall variable improves emergency department (ED) patient volumes forecasting. Our secondary objective was to characterize the magnitude of effect of snowfall on ED volume.
The implementation of health information technology (HIT) is complex. A method for mitigating complexity is incrementalism. Incrementalism forms the foundation of both incremental software development models, like agile, and the Plan-Do-Study-Act cycles (PDSAs) of quality improvement (QI), yet we often fail to be incremental at the union of the disciplines. We propose a new model for HIT implementation that explicitly links incremental software development cycles with PDSAs, the QI-HIT Figure 8 (QIHIT-F8).
Optimal nurse assignment is essential for ensuring optimal patient care and improving staff satisfaction at the St. Michael's Hospital Emergency Department. Nurse assignment has traditionally been a manual task, averaging 1 to 1.5 hours to generate assignments for one day. In this blog post, we aim to highlight the process we took to creating an assignment and some challenges we faced throughout the development.
This publication looks into how advanced analytics can uncloak insights from big data.
January 16, 2019
Rewriting the same code blocks is a time consuming process and often times involves recreating the wheel for any new employees. One of the main programming languages used at CHART is the R language. With the R language you can write packages that contain these blocks of code and share it across the team. This post will provide easy steps to creating an R package in RStudio.
November 23, 2018
CHART data scientists use a lot of raw text data that are often not captured in EHR. Since some of the data contain misspelled words, correcting the spellings during the preprocessing stage is important. This blog post will demonstrate how to correct misspelled words using the R programming language.on...
Calculating Volumes from Timestamps Using R
October 11, 2018
The following brief tutorial will walk you through how to set up time series Emergency Department data in formats appropriate for answering the questions 1) How many patients will be arriving to the ED at some time point in the future (arrival volumes) and 2) How many patients will be sitting in the ED at some point in the future (patient census) using R.