AI Tool Cuts Unexpected Hospital Deaths by 26%

AI-powered early warning system reduces unexpected hospital deaths: A study conducted at St. Michael’s Hospital in Toronto reveals that an artificial intelligence tool called Chartwatch has led to a significant 26% reduction in unexpected deaths among hospitalized patients.

How Chartwatch works:

• The AI system monitors approximately 100 inputs from a patient’s medical record, including vital signs, heart rate, blood pressure, and lab test results.
• It analyzes changes in the medical record and makes hourly predictions about a patient’s likelihood of deterioration.
• The tool flags potential issues earlier than traditional methods, allowing for quicker interventions and potentially life-saving treatments.

Key findings of the study:

• Researchers examined over 13,000 admissions to St. Michael’s general internal medicine ward, comparing the impact of Chartwatch to thousands of admissions in other subspecialty units.
• The 26% drop in unexpected deaths was observed only in the unit using Chartwatch, while other units without the tool showed no change in mortality rates.
• Chartwatch appears to complement clinicians’ judgment and leads to better outcomes for fragile patients, helping to avoid sudden and potentially preventable deaths.

Development and implementation:

• Unity Health AI team began developing Chartwatch in 2017 based on staff suggestions for predicting deaths and serious illnesses.
• The technology underwent several years of rigorous development and testing before being deployed in October 2020.
• Chartwatch is one of the first AI technologies in Canada to be implemented for daily patient care in a hospital setting.

Practical applications:

• In one case, Chartwatch alerted staff to a patient with a high white blood cell count several hours before routine checks, leading to prompt treatment of cellulitis.
• The tool helps identify patients who may require intensive care or are at risk of death, allowing for timely interventions.
• In some cases, Chartwatch assists in providing early palliative care when patients cannot be rescued.

Broader context of AI in healthcare:

• AI technologies are being explored for various medical applications, including early cancer detection, hypertension identification, and concussion diagnosis.
• These tools have the potential to address staff shortages in Canada’s healthcare system by supplementing traditional bedside care.
• However, concerns about privacy and the need for further research in multiple contexts remain important considerations.

Study limitations and future directions:

• The research was conducted during the COVID-19 pandemic, which presented unique challenges to the healthcare system.
• The study focused on a single unit within one hospital with a distinct patient population, limiting its generalizability.
• Further research is needed to validate Chartwatch’s effectiveness in different hospital settings and patient populations.

Expert opinions:

• Dr. John-Jose Nunez, a psychiatrist and researcher at the University of British Columbia, praised the study for providing a “real-world” example of how machine learning can improve patient care.
• Nunez also emphasized the need to consider patient privacy when implementing AI technologies in healthcare settings.

Potential for wider implementation:

• The Unity Health team is working towards expanding Chartwatch’s use within their Toronto-based hospital network and beyond.
• GEMINI, Canada’s largest hospital data-sharing network for research and analytics, is collaborating with over 30 hospitals across Ontario to test Chartwatch and other AI tools in various clinical settings.

Analyzing deeper: While the results of this study are promising, it’s important to recognize that AI tools like Chartwatch are meant to augment, not replace, human clinical judgment. As these technologies continue to evolve, maintaining a balance between technological innovation and personalized patient care will be crucial for their successful integration into healthcare systems. Additionally, addressing potential biases in AI algorithms and ensuring equitable access to such technologies across diverse patient populations will be essential considerations for future research and implementation efforts.