Projects per year
Abstract
Rationale: Late recognition of patient deterioration in hospital is associated with worse outcomes, including higher mortality. Despite the widespread introduction of early warning score systems and electronic health records, deterioration still goes unrecognised.
Objectives: To develop and externally validate a Hospital-wide Alerting Via Electronic Noticeboard (HAVEN) system to identify hospitalised patients at risk of reversible deterioration.
Methods: A retrospective cohort study of patients 16 years of age or above admitted to four UK hospitals. The primary outcome was cardiac arrest or unplanned admission to the intensive care unit (ICU). We used patient data (vital signs, laboratory tests, comorbidities, frailty) from one hospital to train a machine learning model (gradient boosting trees). We internally and externally validated the model and compared its performance to existing scoring systems (including NEWS, LAPS-2 and eCART).
Measurements and Main Results: We developed the HAVEN model using 230,415 patient admissions to a single hospital. We validated HAVEN on 266,295 admissions to four hospitals. HAVEN showed substantially higher discrimination (c-statistic 0.901 [95% CI 0.898-0.903]) for the primary outcome within 24 h of each measurement than other published scoring systems (which range from 0.700 [0.696-0.704] to 0.863 [0.860-0.865]). With a precision of 10%, HAVEN was able to identify 42% of cardiac arrests or unplanned ICU admissions with a lead time of up to 48 h in advance, compared to 22% by the next best system.
Conclusion: The HAVEN machine learning algorithm for early identification of in-hospital deterioration significantly outperforms other published scores such as NEWS.
Objectives: To develop and externally validate a Hospital-wide Alerting Via Electronic Noticeboard (HAVEN) system to identify hospitalised patients at risk of reversible deterioration.
Methods: A retrospective cohort study of patients 16 years of age or above admitted to four UK hospitals. The primary outcome was cardiac arrest or unplanned admission to the intensive care unit (ICU). We used patient data (vital signs, laboratory tests, comorbidities, frailty) from one hospital to train a machine learning model (gradient boosting trees). We internally and externally validated the model and compared its performance to existing scoring systems (including NEWS, LAPS-2 and eCART).
Measurements and Main Results: We developed the HAVEN model using 230,415 patient admissions to a single hospital. We validated HAVEN on 266,295 admissions to four hospitals. HAVEN showed substantially higher discrimination (c-statistic 0.901 [95% CI 0.898-0.903]) for the primary outcome within 24 h of each measurement than other published scoring systems (which range from 0.700 [0.696-0.704] to 0.863 [0.860-0.865]). With a precision of 10%, HAVEN was able to identify 42% of cardiac arrests or unplanned ICU admissions with a lead time of up to 48 h in advance, compared to 22% by the next best system.
Conclusion: The HAVEN machine learning algorithm for early identification of in-hospital deterioration significantly outperforms other published scores such as NEWS.
Original language | English |
---|---|
Number of pages | 9 |
Journal | American Journal of Respiratory and Critical Care Medicine |
Volume | 204 |
Issue number | 1 |
DOIs | |
Publication status | Published - 1 Jul 2021 |
Keywords
- machine learning
- physiological monitoring
- retrospective observational study
- early warning scores
Fingerprint
Dive into the research topics of 'Detecting deteriorating patients in hospital: development and validation of a novel scoring system'. Together they form a unique fingerprint.Projects
- 1 Finished
-
HAVEN: Hospital Alerting via Electronic Noticeboard
Briggs, J. (PI) & Prytherch, D. (CoI)
University of Oxford, Health Innovation Challenge Fund
3/08/15 → 2/08/18
Project: Research