Using machine learning algorithms, hospitals can now significantly reduce emergency admissions.
Health AI’s power is that of any AI: predictive analysis, which is at the heart of pretty much every machine learning application.
Some hospitals in the U.S. are testing AI’s potential in many frontline and back-office tasks such as robot-assisted surgery, virtual nursing, and administrative workflow.
However, many of these issues arise after staff admit the patient into the hospital. Now, staff can use machine learning algorithms to predict their risk for a stay based on their medical history.
This strategy would be particularly effective in reducing wait times for emergency admissions.
AI Schedules Emergency Room Admissions
Before the logistical and financial burden on hospitals, unplanned emergency admissions are unpleasant experiences for patients.
Using AI systems, medical teams could avoid many emergency admissions. Using a new machine learning algorithm, researchers could significantly reduce emergency admission rates.
Researchers at the George Institute for Global Health at the University of Oxford (UK) say machine learning can analyze individuals’ health records to predict their risk of emergency admissions.
“There were over 5.9 million recorded emergency hospital admissions in the UK in 2017, and a large proportion of them were avoidable. We wanted to provide a tool that would enable healthcare workers to accurately monitor the risks faced by their patients, and as a result make better decisions around patient screening and proactive care that could help reduce the burden of emergency admissions,” said data scientist Fatemeh Rahimian who led the research.
For the study, the team used the UK’s Clinical Practice Research Datalink database. This contained data on 4.6 million patients aged between 18 and 100 years old.
Data included “age, sex, ethnicity, socioeconomic status, family history, lifestyle factors, comorbidities, medication and marital status, as well as the time since first diagnosis, last use of the health system and latest laboratory tests.”
The wide range of variables and the incorporation of temporal information substantially improved the performance of machine learning algorithms. It particularly helped in predicting the risk of emergency admissions compared to existing models.
“Our findings show that with large datasets which contain rich information about individuals, machine learning models outperform one of the best conventional statistical models. We think this is because machine learning models automatically capture and ‘learn’ from interactions between the data that we were not previously aware of.”
Researchers said there’s still room for model improvement through the inclusion of more variables and more information about their timing.
“By deploying such models in practices, physicians would be able to accurately monitor the risk score of their patients and take the necessary actions in time to avoid unplanned admissions.”