
Our Work
Publications from the GEMINI Team
Explore our featured publications to learn more about how we help physicians, health care teams, and hospitals use data to gain insights into patient care and improve patient outcomes.
Automated identification of unstandardized medication data: a scalable and flexible data standardization pipeline using RxNorm on GEMINI multicenter hospital data
August 8, 2023
Objective: Patient data repositories often assemble medication data from multiple sources, necessitating standardization prior to analysis. We implemented and evaluated a medication standardization procedure for use with a wide range of pharmacy data inputs across all drug categories, which supports research queries at multiple levels of granularity.

An update to the Kaiser Permanente inpatient risk adjustment methodology accurately predicts in-hospital mortality
June 9, 2023
The study aims to update and validate the Kaiser Permanente inpatient risk adjustment methodology to predict in-hospital mortality, using open-source tools to measure comorbidity and diagnosis groups, and removing troponin which is difficult standardize across modern clinical assays. Using GEMINI data of adult general medicine inpatients at 28 hospitals in Ontario, Canada, between April 2010 and December 2022, the KP method accurately predicted in-hospital mortality of general medicine inpatients. This updated method can be implemented in a wider range of settings using common open-source tools.

Boosting Delirium Identification Accuracy With Sentiment-Based Natural Language Processing: Mixed Methods Study
December 20, 2022
The study aimed to improve machine learning models that retrospectively identify the presence of delirium during hospital stays by using natural language processing (NLP) technique of sentiment analysis. Using GEMINI data, a detailed manual review of medical records was conducted from nearly 4000 admissions at 6 Toronto area hospitals. Among the eligible 3862 hospital admissions, 994 (25.74%) admissions were labeled as having delirium. Our machine learning learning model that included NLP produced valid identification of delirium with the sentiment analysis, providing significant additional benefit over the model without NLP.

Diversity among healthcare leaders in Canada: a cross-sectional study of perceived gender and race
March 14, 2022
Previous research indicates that healthcare leadership in Europe and the United States is thought to lack gender and racial diversity. However, the degree to which these imbalances exist across Canadian healthcare institutions is not clear. This study sought to better understand the issue by reporting the perceived race and gender makeup of Canadian healthcare executives across the country’s largest hospitals and provincial and territorial health ministries. While the results suggest that gender parity exists among Canadian healthcare leaders, racialized individuals were significantly under-represented, with racialized women representing fewer than 5% of executives in each province. This work calls on healthcare institutions to increase racial diversity in leadership.

Outcomes in patients with and without disability admitted to hospital with COVID-19: a retrospective cohort study
January 31, 2022
A small number of studies suggest that COVID-19 patients with disabilities are at elevated risk for severe disease and death. To inform equitable pandemic supports for all, GEMINI data was examined to assess severe COVID-19 outcomes among hospitalized patients with a broad range of disabilities. Analyses revealed that patients with disabilities experienced longer hospital stays compared to those without disabilities, an effect that persisted after adjusting for age, sex, long-term care facility residence, medical comorbidity, dementia, and psychiatric disorders. Additionally, patients with disabilities (age ≤64 years) were more likely to experience unplanned 30-day readmissions than those without disabilities. The findings highlight the importance of a pandemic response that prioritizes the needs of people with disabilities in hospital and after hospitalization.

Bedspacing and Clinical Outcomes in General Internal Medicine: A Retrospective, Multicenter Cohort Study
January 18, 2022
Hospital patients are typically cared for on cohorted wards according the specialty of the physician taking care of them. However, because hospitals are often over capacity, patients may be assigned to locations other than their designated ward. Termed “bedspacing”, this phenomenon is a common approach to managing hospital capacity strain but may compromise care. To study the effects of bedspacing on general internal medicine (GIM) patients in Toronto area hospitals between 2015 and 2017, GEMINI data examined 40,440 GIM admissions, 10,745 (26.6%) of which were bed-spaced to non-GIM wards and 29,695 (73.4%) were assigned to GIM wards. After multivariable adjustment, bedspacing was associated with no significant difference in mortality, slightly shorter hospital length-of-stay, and fewer readmissions across all participating hospital sites. Although potential harms in high-risk patients remain uncertain, the findings are generally reassuring, suggesting that bedspacing of GIM patients is not consistently associated with poorer clinical outcomes.

Variations in Processes of Care and Outcomes for Hospitalized General Medicine Patients Treated by Female vs Male Physician
July 16, 2021
GEMINI data from 171,625 hospitalizations were applied in a cross-sectional study that found patients under the care of female physicians in the general medicine wards had lower in-hospital mortality than those cared by their male counterparts. While the differences were significant only when hospital and patient characteristics were adjusted, the results also support existing literature on female-mediated patient care which account for factors that are not otherwise characterized by hospital data variables through electronic medical records.

Using machine learning to predict severe hypoglycaemia in hospital
June 17, 2021
The aim of the study is to predict the risk of hypoglycaemia using machine-learning techniques in hospitalized patients. We conducted a restrospective cohort study of patients hospitalized under general internal medicine (GIM) and cardiovascular surgery (CV) at a tertiary care teaching hospital in Toronto, Ontario. Hypoglycaemia occurred in 16% of GIM admissions and 13% of CV admissions. Among the patients at the highest decile of risk, the positive predictive value was approximately 50% and the sensitivity was 99%.

Managing drug shortages during a pandemic: tocilizumab and COVID-19
May 5, 2021
The COVID-19 pandemic has revealed weaknesses in global manufacturing and distribution of medications, exacerbating many pre-existing limitations and inequities in drug supply and creating new shortages. Tocilizumab is a life-saving therapy for adults admitted to hospital with COVID-19; demand for this drug is likely to outstrip available supply in Canada. We estimated that for every 1000 adults in hospital with COVID-19, 429 would be eligible for tocilizumab; treating them all may prevent 17 deaths and prevent 12 instances of invasive mechanical ventilation.

Patient characteristics, clinical care, resource use, and outcomes associated with hospitalization for COVID-19 in the Toronto area
December 16, 2020
In this study, we described all adult discharges from inpatient medical services and medical-surgical ICU between November 2019 and June 2020 at 7 hospitals in Ontario. Out of 43,462 discharges, 1,207 (3.0%) had COVID-19 and 783 (2.3%) had influenza. Patients with COVID-19 had similar age to patients with influenza and other conditions. Patients with COVID-19 were more likely to be male. Compared to influenza, patients with COVID-19 had significantly greater mortality, ICU use, and hospital length-of-stay. COVID-19 patients did not have a significantly different 30-day readmission rate.

Assessing the quality of clinical and administrative data extracted from hospitals: The General Medicine Inpatient Initiative (GEMINI) experience
November 4, 2020
Large clinical databases are increasingly being used for research and quality improvement, but there remains uncertainty about how computational and manual approaches can be used together to assess and improve the quality of extracted data. The GEMINI database extracts and standardizes a broad range of data. We describe computational data quality assessment and manual data validation techniques that were used for GEMINI. Manual data validation revealed that GEMINI data were ultimately highly reliable compared to the gold standard across nearly all data tables.