A sort of Learn Posted in Peer Review American Journal of Health-System Pharmacy It was found last month that machine learning and advanced analytics techniques could successfully identify instances of drug diversion — a term for drugs stolen from healthcare facilities — 160 days faster than traditional, non-machine methods. AI can do this even with vast amounts of data, unlike its human counterparts who have so far been tracking lost drugs.
The study comes from Atlanta, Georgia explore, A software company with technology to analyze and track inventory in healthcare systems.
This retrospective study examined drug diversion in a total of 10 acute care inpatient hospitals belonging to four health systems. The data analyzed included 27.9 million drug diversion transactions from nurses, pharmacies and anesthesia clinicians over a time span of 8 to 24 months.
Mixed together are 22 known cases of drug diversion. The study was designed to see if machine learning and analysis techniques could not only successfully identify these cases from the 27.9 million cases, but if machine learning and analysis techniques could do so faster than current testing standards that initially found these 22 metastases at this point.
Historical methods of metastasis detection include viewing monthly usage reports or daily difference reports. However, these standard detection methods are problematic in some respects. On the one hand, people can hide their pastimes. Second, any metastases that are often detected are not flagged immediately because the data is reviewed weeks or months after they occur. In addition, hospitals have historically relied on clinicians reporting impaired behavior of colleagues to drive investigation of possible triage.
The study noted that machine learning and advanced analytics techniques not only flagged 22 cases of drug diversion, but also an average of 160 days earlier than standard methods. In addition, the machine learning method technique has a high accuracy rate – 96.3%.
“The findings demonstrate that advances in machine learning and analytics are a real game-changer — and could improve detection of drug diversion in hospitals and other healthcare settings,” said Invistics CEO Tom Knight in a press release. This is important given the enormous financial, clinical and emotional burden that drug theft places on the healthcare system, patients and families.”
In addition to Invistics, other institutions participating in the study include: Piedmont Athens Regional Medical Center, Scripps Health, Piedmont Healthcare and EnvisionChange.
“This study speaks for itself for healthcare systems that have not yet leveraged machine learning and advanced analytics tools for drug prevention and detection programs,” said Don Tyson, director of pharmacy at the Athens Regional Medical Center in Piedmont and study author. “Advanced analytics and machine learning techniques can improve the accuracy, efficiency and effectiveness of any drug diversion prevention program well beyond manual resolution, especially when dealing with large volumes of data.”
“Rapid identification of drug diversions is critical to patient safety. Advances in technology have made it possible to detect and investigate potential diversions months in advance,” Pam Letzkus, senior director of pharmacy at Scripps Health and study author, said in a news release. This study has significant implications for both patients and healthcare providers. “
Photo: Valery Brozhensky, Getty Images
