NTT Medical Center Tokyo and UBIC Using AI to Lower In-hospital Injury
March 16, 2015
According to the World Health Organization (WHO), the global population is rapidly aging, and by 2050 the number of people 60 years and over will increase to 2 billion. As this trend continues to increase, risk factors that are particular to this demographic will also see a dramatic rise. One of these issues are falls, and the WHO points out in the US alone, 20 to 30 percent of the elderly who fall suffer moderate to severe injuries such as bruises, hip fractures, or head traumas. For countries with very high elderly population rates, such as Japan, public and private entities are trying to find solutions to alleviate the problem. That is why NTT (News - Alert) Medical Center Tokyo and UBIC have come together to conduct a joint research on an artificial intelligence based system to mitigate patients’ risk of falling in hospital settings.
One of the goals the NTT Medical Center Tokyo has set for itself is to reduce the risk of patients harming themselves from falls. To that end, the organization has come together with UBIC, provider of international litigation support and big-data analysis services, to come up with a solution that can use the available data in a healthcare facility using AI technology.
The research program is looking to detect signs of potential falling from the information gathered in electronic medical records and analyzing it to reduce in-hospital injury. They want to give healthcare practitioners information that will allow them to recognize patients who are at risk and provide a safe environment in the facility.
There were many challenges in deploying a solution that could accurately identify patients that were prone to falling. The task was made harder because the average period of stay for patients in a one year period between 2013 and 2014 was only 10.6 days at the Tokyo facility. With almost half of these patients being discharged within six days. Therefore, the solution had to identify patients that were at risk of falling quickly.
The AI technology will sift through the text data of 16,749 patients-per-day, which contain information provided by medical staff to calculate their risk of falling for each patient. Based on this information and other indicators, 1,000 patients were identified and deemed to be high risk.
The result, as published by UBIC, revealed the AI technology has the potential to detect the risk of individual patients so they won’t have to experience an unexpected adverse event, such as a fall.
“The present system can identify such signs in ways similar to human tacit knowledge from a large amount of text information derived from electronic medical records. Therefore, it is very likely to be more efficient and useful for patient care than other existing adverse event prevention measures,” said an official at the Department of Medical Safety of NTT Medical Center Tokyo.
The AI software program used in this research was the Virtual Data Scientist (VDS). The platform incorporates expert judgment, known as tacit knowledge, and uses it for big data analysis. Eventually they want to develop a system that will deliver better patient care by predicting falls and other events in hospitals based on what the platform learns from the staff and electronic medical records.
Edited by Dominick Sorrentino